Latest from Tom's Guide AU in Gpus

I gamed for 250 hours on AMD Radeon RX 9070 GRE — it’s a $549 nightmare for Nvidia’s RTX 5060 Ti

Jason England — Tue, 02 Jun 2026 00:00:00 +0000

I get it. One read of my Best GPUs buying guide is enough to make your bank account weep. RAMageddon has sent graphics card prices into a crisis, but AMD is fighting back with the Radeon RX 9070 GRE.

Eagle-eyed PC gamers may have already seen these out in the wild over the past year, and they're readily available in China over the past year and starting to leak from retailers across the U.S. Now, I can confirm it’s official, and I’ve been testing one for the past few weeks. In short, if they can keep to their $549 promise, this could be a rare value win in GPUs in 2026.

In raw gaming performance, this closely matches Nvidia’s RTX 5070 (which is around $150 more on average) and comfortably breezes past the 16GB RTX 5060 Ti ($20 more on average). For that out-of-the-box, ray-traced gaming performance without any AI trickery, the RX 9070 GRE is a winning 1440p formula.

Not to say it’s the perfect formula, and you’ll have to decide what matters most for you here. It doesn’t have the same productivity chops as an Nvidia card (in both creator workloads or AI). And while FSR Redstone support is starting to expand, and it’s a great AI-infused game performance booster, Nvidia’s DLSS 4.5 continues to have a clear lead in feature set and game support.

But with all that being said, provided this suggested price of $549 holds true (AMD did confirm to Tom’s Guide that a lot of stock has been produced), if your sole focus is gaming bang for your buck, this is a top option that challenges the extortionate GPU prices found elsewhere.

AMD Radeon RX 9070 GRE: Cheat Sheet

What is it? This is a mid-range PC graphics card.
Who is it for? This is for people who are looking for great 1440p gaming performance at maxed-out settings.
What does it cost? The SEP is $549, and I’m praying it stays that way.
What do I like? For $549, this is an impressively performant card for 1440p gameplay, with FSR 4 unlocking some 4K experiences. It's a great bang-for-your-buck pick!
What don’t I like? I’m suspicious of that price tag (hopefully I’m proven wrong), and, much like the RX 9070 XT, while it’s a great raw performer, FSR 4 support doesn’t match Nvidia’s DLSS 4 game compatibility, and AI and productivity aren't as strong here.

AMD Radeon RX 9070 GRE: Specs

Specification	RX 9070 GRE	RTX 5060 Ti
GPU Architecture	Navi 48	Blackwell (GB206)
Cores	3,072	4,608
Base Clock	1,420 MHz	2,407 MHz
Boost Clock	2,790 MHz	2,572 MHz
Memory Type	GDDR6	GDDR7
Memory Size	12 GB	8 GB or 16 GB
Memory Bus Width	192-bit	128-bit
Memory Bandwidth	432 GB/s	448 GB/s
Matix/Tensor Cores	96 4th Gen cores	144 5th Gen cores
RT Cores	48 3rd Gen cores	36 4th Gen cores
Power Consumption (TDP)	220W	180W
PCIe Interface	PCIe 5.0	PCIe 5.0
Display Outputs	1x HDMI 2.1b, 3x DisplayPort 2.1a	1x HDMI 2.1b, 3x DisplayPort 2.1b
Launch Date	June 1, 2026	April 16, 2025
Starting Price	$549 (SEP)	$459 retail average (8 GB), $569 retail average (16 GB)

AMD Radeon RX 9070 GRE: 3DMark tests

AMD Radeon RX 9070 GRE: The ups

So let’s tuck in. For transparency, I tested the RX 9070 GRE in my build with AMD Ryzen 7 7800X3D CPU and 32GB DDR5 RAM. And as I started tweaking settings, playing and testing, I quickly realized the raw capabilities that lie under the hood.

Challenging the gaming throne

(Image credit: Future)

The price of Nvidia’s GPUs has been going crazy in the U.S. Now that the Radeon RX 9070 GRE is here on sale for $549 (for now), I can only anticipate a market correction given how much more this offers for less.

Raw rasterization performance (1440p)
GPU	Cyberpunk 2077 ray tracing ultra 1440p	Forza Horizon 5 max settings 1440p	Call of Duty max settings no frame gen 1440p	Black Myth: Wukong Cinematic 1440p
RX 9070 GRE	36.6 FPS	142 FPS	136 FPS	81 FPS
RX 9070 XT	30.2 FPS	190.8 FPS	173 FPS	62 FPS
RTX 5070	38.2 FPS	158 FPS	139 FPS	84 FPS
RTX 5060 Ti	32.8 FPS	121 FPS	80 FPS	45 FPS

This is coming close to the far more expensive RTX 5070, while pulling comfortably ahead of the 5060 Ti. With those improved ray tracing cores, you’re starting to see AMD get more comfortable with shiny surfaces and reflective windows.

At 1440p, there’s enough headroom in that 12GB of video memory, and it forms a better value for money option than the RX 9070, and is frankly miles ahead of Nvidia’s RTX 50-series lineage in performance-per-dollar.

FSR Redstone is getting better

(Image credit: Playground Games)

When I reviewed the RX 9070 XT, the idea of FSR Redstone was only a glimmer in AMD’s eye, and the broader support for its upscaling/frame generation tech was left reeling. Now, over 200 games support it, and the AI trickery looks great here with extremely minimal ghosting around fast-moving objects and impressive detail even on a character’s hair.

Of course, one key difference here from the likes of Nvidia’s DLSS 4 is how it handles frame generation. With Team Green, you’re getting multi-frame generation where it can stuff up to six AI-generated frames in between two GPU-generated ones. With FSR Redstone, you’re only getting one.

But I have a counterpoint to that. With the frame rates you see in these games, does a couple of hundred frames per second even matter? In my gameplay time, I can honestly say it didn’t.

And the second point, the latency is actually a few milliseconds lower without the additional frames. Food for thought!

AMD Radeon RX 9070 GRE: The downs

That doesn’t mean the RX 9070 GRE is an all-around Nvidia beater. There are two clear places where this mid-ranger does falter — one comes down to Nvidia being in the AI game for longer, and the other is a side effect of when you throw all your eggs in the gaming basket.

Nvidia keeps the compatibility advantage

(Image credit: Tom's Guide)

Chances are, you’ve seen a whole lot of AAA games in 2026 come with an Nvidia logo in their opening credits. Team Green has been busy building relationships across the board, and while AMD is catching up, Nvidia still has an advantage in this space.

Not only in-game compatibility, but also in the number of tools available. Team Red is offering the core tools needed to make all this work well (and it does), but whether you think they’re controversial or not, Nvidia’s AI trickery is evolving quickly as more games come under its umbrella.

Productivity performance is lacking

(Image credit: Tom's Guide)

Gaming’s all good and everything — definitely the key priority for most of you reading this. However, if you need something more multipurpose, it’s clear that AMD is focusing on gaming here. You can see that in the Blender and AI benchmark results.

Productivity performance
GPU	Blender combined score (higher is better)	Procyon AI image gen Stable Diffusion XL
RX 9070 GRE	975	1516
RTX 5070	2390	2438
RTX 5060 Ti	1127	1811

While Nvidia has the drivers and capabilities of doing more across the board, Team Red is focusing in. No such problem for most of you looking at this, of course, as chances are you just wanna play games fast here. But just a heads up to the creators and AI workers out there.

AMD Radeon RX 9070 GRE: Verdict

(Image credit: Tom's Guide)

So, to answer the fundamental question, yes, the RX 9070 GRE is a great 1440p GPU with better rasterized performance than its similarly priced competitors. Productivity performance is underwhelming compared to Nvidia, but if gaming is your main thing, this is a stellar pick.

But as the hours continued to melt by in all my favorite games, one thing became clear. I look at the grim backdrop of GPU pricing and can only pray that the RX 9070 GRE holds true to its $549 retail price. Because if it does, it could move the needle and encourage price drops across the board.

Update your Nvidia GPU drivers now to protect your PC from 9 "high-severity" vulnerabilities — here's what's at risk

scott.younker@futurenet.com (Scott Younker) — Wed, 20 May 2026 20:57:39 +0000

If your Windows or Linux computer uses one of best graphics cards made by Nvidia, you're going to want to make sure you're using the latest drivers. This week, the company issued a security alert and driver updates to combat a variety of vulnerabilities.

The alert highlights 15 issues, nine of which Nvidia has marked as "high-vulnerability." The high-risk flaws run the gamut of what bad actors can do to your PC. That includes letting hackers get access to your PC kernel, inject malicious code, steal crucial data, or gain administrative access. All the stuff you don't want happening.

For both Windows and Linux, you can download the driver update directly from Nvidia. On Windows, you'll want to make sure you upgrade to driver version 569.49.

On Linux, you want to make sure you update to version 590.48.01. The new versions were released about a week ago, so most people who have automatic updates turned on should have the update. But check your driver version just in case.

According to the alert, all Nvidia drivers before version 596.36 — version 482.53 for GTX 10-series and below — are potentially at risk from these vulnerabilities.

Keep your PC safe from malware

(Image credit: Getty Images)

This is a general reminder to ensure all your drivers are up to date. Most driver releases patch security flaws like the ones outlined by Nvidia.

To be fair, some driver updates can cause issues, but later this year, Microsoft will automatically roll back bad Windows drivers to the most recent stable version. Still, keeping your drivers up to date is good practice.

If you're on PC, Microsoft releases new security updates every second Tuesday of each month.

Additionally, you'll want to ensure that Windows Defender is enabled. It largely does a great job of catching threats before they do damage.

For extra protection, you should consider the best antivirus software. Paid antivirus solutions usually update regularly, plus you often get access to VPNs, a password manager and other security goodies.

New vulnerabilities crop up all the time, but if you practice good cyber hygiene you devices and data should stay safe.

The end of unlimited AI: Why Google’s Gemini leak is a warning for every power user

amanda.caswell@futurenet.com (Amanda Caswell) — Wed, 06 May 2026 16:53:57 +0000

For the past year, AI felt like you could freely ask your questions and generate endlessly. And many users have, treating ChatGPT like a faster Google. The company behind Gemini was known to give away their best models "for free" so users could prompt heavily. Now, Google is making it clear those days are over.

In details first spotted by 9to5Google, what looked like a routine update revealed new usage limits across Gemini’s most powerful features. And the company isn’t just tweaking a pricing tier, it’s signaling that AI isn't unlimited, and it never really was. Once you see it, you can’t unsee it.

What actually changed with Gemini

(Image credit: Future)

Google’s latest update reshapes how access to Gemini works, particularly for higher-tier users. Limits are now:

Tied to specific features (like Deep Research and advanced tools)
Adjusted based on usage patterns
Structured more like a quota system than a free-flowing assistant

That might sound like a small backend change, but it fundamentally shifts how AI feels to use. Because now, there’s a ceiling. Whether we like it or not, our access has changed. You've probably already noticed it if you've felt your AI responses suddenly got shorter, less helpful or the response were slower.

To be clear, this isn't just happening with Gemini. There’s a growing pattern across tools like Claude, ChatGPT and Perplexity AI, too. When users hit a "soft limit" the responses change. If you use a premium feature too often such as Gemini Veo 3.1 or ChatGPT-5.5 Thinking, you may find access gets throttled. The model itself isn't getting worse, but the version you're using might be. In other words, you think you're using a more advanced model, but due to usage limits, you've quietly been dropped down to a lesser model.

This shift is subtle, and most people aren't even realizing it's happening. They may get a generic response and simply prompt again, only to push their usage limits even more.

The real reason this is happening

(Image credit: Shutterstock)

Big Tech is paying a lot of money for data centers, energy and training models, and users are now being taxed with helping to offset these soaring costs. This isn't about Google being restrictive, but about physics and the cost of massive GPU clusters, constant energy consumption and very expensive real-time inference.

That means, every “ask” isn’t free, it's a compute event. And, as millions of people start using AI like a daily assistant, those costs scale fast. So instead of offering unlimited access, companies are quietly shifting to tiered usage, feature grating and soft caps. It's kind of like how the best streaming services all started raising their prices once they had gotten enough people to sign up.

For power users, this means AI use is something you have to budget as we enter the "AI credit" era. Even if these companies aren't calling it out yet, that's what it is. Frankly, it's no different than mobile data plans, streaming tiers or cloud storage limits.

The takeaway

This shift changes how you should actually use AI. For example, generating AI images is fun, but it will cost you more than it used to. And, instead of treating AI like an endless chat, it's time to be more intentional with prompts and use different tools for different tasks. If you have heavy requests, save them for when they really matter.

The people who win with AI going forward will be the ones who know how to use is better and more strategically. What do you think? Will these limits make you rethink how you use AI? Let me know in the comments.

I stress-tested Pragmata’s path tracing — it looks like a dream and Nvidia’s DLSS 4 makes it play like one (provided you use it right)

Jason England — Thu, 16 Apr 2026 12:05:43 +0000

As you may have seen, we rather love Pragmata right now, and I can give all you PC gamers out there the lowdown on how well it’ll run on your rigs. Spoiler alert: just like Resident Evil Requiem, path tracing adds a night and day difference.

So I threw it at my pricey RTX 5080 and mid-range RTX 5060 Ti rigs to see how far I can stretch those fidelity settings with DLSS 4, and how well it performs with this incredibly demanding graphical tech running.

What settings should you use?

(Image credit: Capcom)

As I always say, Nvidia’s DLSS is a helluva tool for extracting more gameplay value out of your GPU — provided you use it the right way. Because sure, a game can look smooth, but if you’re adding all the AI trickery on top of a game that is running at 20 FPS, it will not feel smooth to play.

That’s why you’ll need to go into the settings to tinker the base rate a little before overlaying DLSS features. Just as a general reminder:

If your game is a slower-paced single-player title, aim for 40 FPS
Need quick reflexes? Aim for 60 FPS
For competitive multiplayer, either don’t use it or aim for 120 FPS

From mine and Tony’s time playing Pragmata, you can safely put it in that 40 FPS target category. There is combat that requires you to dodge attacks and shoot, but with plenty of time to telegraph enemy moves, the really fun puzzle elements and button-based quick-time events, you’ll still get a game that feels good to play at this level.

By the numbers

(Image credit: Capcom)

So allow me to introduce you to the machines that will be taking this on (quite proud of my rebuilds):

Remember that PC I built for myself at Scan computers? I’ve done a transplant into the gorgeous Micro ATX case you see in the top image — doing some space saving while preserving power. It packs an AMD Ryzen 7 9800X3D, RTX 5080 and 32GB of DDR5 RAM. You can watch the whole big build here.
The second is a modified RDY Element 9 Pro R07 — sporting an Intel Core Ultra 7 270K Plus, RTX 5060 Ti (16GB) and 32GB of DDR5 RAM.

Same rules apply as my Requiem testing, so I’m stress testing at resolutions fair to each card (4K for the 5080 and 1440p for the 5060 Ti) at maxed out path tracing settings. Of course, as I said above, you can tweak some graphical settings to squeeze even more frames out of the base rate.

And as you can see, the multipliers really start to pile on the additional frames nicely here — making the game look silky smooth.

Latency checker

So there’s big frame rates happening up above, but that’s half the story. As I said up above, the game’s got to feel smooth too, and in this stress test (just like what happened with Requiem), this is where it becomes clear that DLSS 4.5 performance is your go-to option if you wanna go all-out in the settings.

DLSS always comes with a small penalty to that base frame rate, as the GPU’s attention is turned to the AI background work. And while these are mostly playable, Quality or Balanced upscaling does lead to some noticeable latency. Performance is your sweet spot.

Path traced sci-fi fun

Path tracingFuture

No path tracingFuture

If you follow my advice from up above, you’ll have a great experience with what has become one of my personal surprise favorite games of 2026. Path tracing really does make a material difference to the visual fidelity here and draws you into the immersion of this fascinating sci-fi world.

I demoed this game a few months ago and immediately fell in love with the hybrid shooter/puzzler gameplay, but I did feel a little nervous about how far this could go before it becomes tiresome. Turns out the answer is it never gets old! New layers of complication are added over time and discoverable elements keep things fresh. Oh, and the boss battles are epic.

I tested DLSS 4.5 Dynamic Multi Frame Generation: It’s not just AI trickery, it’s Nvidia’s cheat code for 4K 240Hz and a true console killer

Jason England — Tue, 31 Mar 2026 13:00:00 +0000

Nvidia has just dropped DLSS 4.5 Dynamic Multi Frame Generation and I’ve been testing it for the past week. This takes the idea of multi-frame generation and adds automation to it, so AI only generates the frames you need rather than going over the top, and in my time using it, Team Green has created a far smoother experience with lower latency because of it.

It’s a better balance of making sure your GPU’s raw rendering skills and AI trickery work hand-in-hand to deliver more consistently stable gameplay. But while you’ll hear a lot about this being big for 4K 240Hz monitors, I found that this is the missing piece to using your PC as a games console replacement — a dream I’ve had for years. Let me explain.

What is Dynamic Multi Frame Generation?

(Image credit: Nvidia)

Up until now, Nvidia has dabbled in Multi-Frame Generation — you can think of this like the manual transmission in a car. Each gear has its benefits for more speed, but if you slow down at these higher gears, it’s hard to get back up to speed. This can be felt in the form of latency if you poorly optimize your game from the jump.

Dynamic Multi Frame Generation switches that out with an automatic transmission, which changes gears (frame multipliers) based on the intensity of the scene — ensuring you’re at the max speed of the display you’re on at all times. This ensures you only get the frames that you need, and produces a far smoother experience with less latency.

In this beta, you’ll see two presets in the Nvidia app:

Preset A: The brute force option. This will take a look at literally everything on screen as one frame and generate. Much wider compatibility, but can come with some issues of smearing in-game UI elements with the background.
Preset B: In this model, DLSS is able to look at both the in-game visuals and the UI separately. This can help keep your on-screen text razor sharp, but compatibility is limited as it requires developers to isolate the layers.

(Image credit: Future)

But as I always say (following on from poorly optimizing your game), it’s critical that you use this the right way. I’ve got a full guide on my resolution scaling and frame generation preferences, but to summarize real quick, it’s important that your base frame rate before turning on any AI trickery is strong.

Because if your game is running at 20 FPS, slapping on this DLSS coat of paint will make it look smoother, but it’ll still feel like 20 FPS.

If you’re in a single player game with slower pacing, 40 FPS is enough.
Need faster reflexes? Start at 60 FPS.
Competitive multiplayer? Don’t use it or build from 120 FPS.

Got it? Good. Let’s get into the results of my testing.

By the numbers

(Image credit: Future)

So let’s test what Nvidia has here — starting with its main focus upon making the most of a 4K 240Hz monitor. I’ve got my RTX 5080 rig that I built with Scan computers here in dear old Blighty, and the Alienware AW2725Q monitor. If you want to see how to build a PC like this for yourself, the guide is just below.

Anyway, let’s get into it. Put simply, monitors of this caliber are no longer just future-proof hardware — RTX 50-series is able to fully utilize them and get 100% of the experience you pay for.

Game (and settings)	Average FPS	1% Lows (FPS)	Latency	Average Dynamic Multi Frame Generation Multiplier	Base FPS
Cyberpunk 2077 (4K Ray Tracing Ultra)	238.99	167.87	43.37ms	4-5x	58.5 FPS
Hogwarts Legacy (4K Ultra)	205.94	79.51	36.34ms	5-6x	74.02 FPS
Dragon Age: The Veilguard (4K Ultra)	237.17	132.58	41.47ms	5-6x	53.15 FPS
God of War Ragnarok (4K Ultra)	211.55	106.93	37.5ms	2-3x	92.47 FPS

In games like “Dragon Age: The Veilguard,” you’re seeing a base frame rate of 53 FPS get effectively upshifted heavily to saturate the monitor’s refresh rate. On the opposite end of this, “God of War Ragnarok” already has a strong base performance, so DFG essentially downshifts to a lower multiplier for the smoothest possible image quality.

(Image credit: Future)

Another way to look at it is those 1% Lows. This is the lowest frame rate you see 1% of the time (as the name suggests), and it’s a good way to track whether the game will visibly stumble at any dramatic rate.

On average (including some loading scenes, so data is a little bit off), you’re looking at an average of 115 FPS across all my tests, which is an excellent floor of the performance — well above the variable refresh rate threshold and eliminating visible judder.

The beauty of this restraint is fully realized in the latency, though. Multiple generated frames are often criticized for adding a floaty input lag. Of course, this comes down to how to tame the technology in many ways. But with an average of 33-45ms, my data proves Nvidia has overcome this hurdle.

To put it in perspective, Cyberpunk 2077 with everything turned up and maintaining only 45ms of latency is a landmark achievement to squeeze up to 240Hz gaming.

To my eyes

Future

As for the visual element of it (numbers are good, but it’s got to look good too), same as I found when going eyes-on back at CES 2026, You’ll struggle to notice when the multiplier shifts — there’s no identifiable switch up or stuttering at all in all the games I tried.

(Image credit: Future)

There are some small ghosting issues I did spot. The most prominent one being in “Hogwarts Legacy” around this teacher’s hair, but most impressively, none of this impacts the UI. One of the most common problems with frame generation is that it can make on-screen elements like a health bar or button prompts smear into the background of the game.

And that’s the magic of Preset B isolating static elements from the game visuals. Menus in “Spider-Man 2,” “Hogwarts Legacy” and “Forza Horizon 5” remain perfectly clear even as the 3D world is being interpolated at a 4-6x rate.

TV time

(Image credit: Future)

Now, you’re going to hear Nvidia talk a whole lot about making the most out of high refresh rate hardware, which Team Green absolutely does here. But I wanted to see whether it could fix one problem I’ve always had — playing PC games on my TV.

It’s been my dream to have a cheaper RTX 5060 Ti rig plugged into my big screen to replace my console setup. A lot of these single-player games that Dynamic Multi Frame Generation is perfect for usually run at 30 FPS on Fidelity mode and 60 FPS on performance.

While a PC can smash that barrier already (and latency issues are felt much less with a controller), there are two key issues:

UI smearing: The brute force frame generation option has viewed the entire game and its UI as one picture. When blown up to a massive TV, that means you can see the UI elements start to smear and ghost — especially when the picture behind it is moving fast (think like a racing game).
The VRR stutter: When framerates can change quite dramatically on a PC game, most HDMI 2.1-armed TVs can suffer from gamma flickering. Without the variable refresh rate of a gaming monitor, the locked refresh rate of the big screen behaves a little weird with this.

Future

With Dynamic Multi Frame Generation (and Preset B), it was my hope that beyond the marketing messages, this could be the key to having my PC become the all-in-one gaming desk machine and TV setup.

And you know what? It absolutely nails it. Of course, if you're looking at this from a keyboard and mouse/your face close to a monitor perspective, some of those latency numbers could be a problem. But for the couch gaming side of things (average response time on a PS5 or Xbox Series X sits between 100ms and 150ms), these numbers are actually a win for the TV rig.

Game (and settings)	Average FPS	1% Lows (FPS)	Latency	Average Dynamic Multi Frame Generation Multiplier	Base FPS
Cyberpunk 2077 (4K Ray Tracing Ultra)	123.46	92.51	83.69ms	4-5x	27.16 FPS
Hogwarts Legacy (4K High)	98.43	30.16	77.12ms	5-6x	31.1 FPS
Dragon Age: The Veilguard (4K Ultra)	132.92	105.26	58.65ms	3-4x	34.42 FPS

Of course, it's worth noting that this is a stress test. I wanted to see how far I could push this on an RTX 5060 Ti rig with settings turned all the way up — lowering the base frame rate target to what a console hits at 30 FPS and seeing how it fares.

(Image credit: Future)

If you want something more responsive, you can tweak those settings down a tad. But for my time playing with a controller, I certainly didn't notice any input latency at all.

The UI remains mostly rock solid thanks to preset B isolating the in-game visuals from the on-screen elements, and generating only the frames I need keeps things feeling responsive while eliminating any flickering.

Verdict: a console killer?

(Image credit: Future)

So the main thing I want you to take away from this is that Dynamic Multi Frame Generation is not just for the enthusiast with the 240Hz monitor. It’s for the rest of us gamers who want their PC to act like a polished, reliable appliance in the living room.

For years, my “PC as a console” dream died at the hands of micro-stutters, UI ghosting and the constant need to babysit settings. While the latter of those three could do with some work (looking at you, Microsoft), Nvidia has finally automated the two headaches away.

It’s the first time that AI hasn’t felt like a “more frames” button, but rather a “better experience” brain. By letting it decide when to push and when to pull back, your PC stops feeling like a hot rod and starts feeling like a refined supercar of a high-performance console.

Put simply, if you’re on an RTX-50 Series GPU, this isn’t just a toggle — it’s the new standard.

Is the RAM price crisis finally ending? I’m starting to see cracks

Jason England — Fri, 27 Mar 2026 12:28:32 +0000

For the past few months, buying a PC has felt like trying to buy a house in a hurricane — prices are up and supply is down. The RAM price crisis has been hitting us hard, and while it’s easy to see how bleak the situation is (I see it too), I want to talk about how in March 2026, I believe the atmospheric pressure is finally shifting.

I looked on PC Part Picker and found that the average price of DDR5 had dropped slightly. Yes, it’s the smallest of margins compared to the lows before this all kicked off, but I’m looking for any good news from anywhere. This got me going down a rabbit hole of research and what I found gave me hope.

(Image credit: PC Part Picker)

Thanks to a perfect storm of a massive legal victory for importers and a sudden pivot in Washington trade policy, this hyper-inflation that defined the hardware dark age of early 2026 is showing its first real cracks. We aren’t out of the woods yet, and there are plenty of reasons to be skeptical (some I’ll go into here). But I look around and I see a hardware thaw beginning.

Uncle Sam owes Big Tech $130 billion — and you might get the change

(Image credit: Shutterstock)

On March 4, 2026, Judge Richard Eaton of The U.S. Court of International Trade issued a sweeping ruling after the Supreme Court found tariffs to be unconstitutional — saying that “every single cent” of the roughly $130 billion to $175 billion collected must be returned to importers.

This would be a massive injection of money from U.S. Customs and Border Protection (CBP) to companies like Asus, Dell, Nvidia and HP. The CBP is currently building an automated mass-refund tool, which the court was told should be ready by mid-April 2026.

On paper, that could mean companies that were pricing in these taxes to protect their margins will have a massive cushion of cash returning to their balance sheets — potentially leading to lower prices.

🤔 Skeptical take: Lower ceiling, but higher floor?

(Image credit: Future)

Now before you go thinking that refund check is in the mail, the Department of Justice is already setting up to appeal the scope of this ruling. The argument being that only companies who filed individual lawsuits (not every importer) should get paid.

And even if this automated mass-refund tool goes live and companies line up to use it, legal experts warn that these requests could be tied up in courts for up to four years.

Plus, there isn’t a legal requirement for a company to pass a tax refund down to the consumer. As Asus’ Director of Technical Marketing, Sascha Krohn said when we asked about the memory inflation: “nobody wants to be the first one to lower prices.” Maybe companies simply use this to pad their bottom lines after a rough year.

Oh, and that’s before even mentioning the administration invoking Section 122 — a temporary 10-15% global surcharge to address the balance-of-payments here.

So while 10% is certainly better than 25%, and I do think that will lower the ceiling of high-end prices, that floor of lowest prices won’t be lowered back to the beforetimes.

The AI Tax is getting a divorce from your gaming rig

(Image credit: Amazon)

This goes to the Presidential Proclamation 11002 — signed on January 14, 2026. On the face of it, the headline you’ll have probably read was about a massive 25% tariff on high-end AI silicon. But scratch beneath the surface, and you’ll find a loophole in new White House policies under Section 232.

This exemption covers “Non-Data Center Consumer Applications,” and it’s a big one. A “get out of jail free” card for gaming PCs, laptops and consoles, which could lead to a significant K-shaped correction.

Do you see what this means? Enterprise and Consumer Computing are getting a silicon divorce, so while the enterprise side of things remains at premium prices, a carve-out of the consumer end could prevent the total disappearance of cheap laptops and mid-range GPUs.

🤔 Skeptical take: A technicality?

(Image credit: Future)

So that consumer shield sounds great on paper, but the technical definitions are rather thin — exempting products based on Total Processing Performance (TPP) and DRAM bandwidth. For full context, here are the triggers for a 25% tariff:

Tier A: A Total Processing Performance (TPP) between 14,000 and 17,500 AND a DRAM bandwidth between 4,500 and 5,000 GB/s.
Tier B: A TPP between 20,800 and 21,100 AND a DRAM bandwidth between 5,800 and 6,200 GB/s.

For anyone keeping track, the RTX 5090 isn’t just “close” to these triggers, it has actually smashed through the floor of the restricted category of TPP. Of course there’s a memory bandwidth buffer and vastly lower interconnect speeds — saving it for not.

But the real risk here is of performance creep breaking that consumer shield. Let’s take Nvidia for example. The RTX 50-series is the first generation where the gaming GPU and the AI GPU are technically the same level of power.

If the next versions (be it RTX 50 SUPER or RTX 60 series) increase memory bandwidth or add even better AI-specialized Tensor cores that push that TPP higher, they could meet both criteria for the hit list. And at that point, the likes of Nvidia and AMD have two choices:

“Nerf” the cards: Intentionally slow them down to stay under the limit (think like the Nvidia RTX 4090D that was released for the Chinese market).
Lobby aggressively: Petition hard for that TPP ceiling to be raised as technology advances.

Put simply, while your gaming rig is safe right now, we could be one breakthrough away from the government deciding your GPU’s path tracing capabilities are a national security asset.

The DeepSeek effect continues

(Image credit: Getty Images)

The reason why DeepSeek R1 shocked the entire market early last year is not because of capability, it was efficiency and low cost. Google recently announced TurboQuant, which is being called the “Pied Piper moment” for AI (“Silicon Valley” is basically real-life now).

Basically, TurboQuant is able to compress AI’s working memory by at least 6x and make it 8x faster — all with no accuracy loss. On paper, this should ease the overwhelming demand on fast memory processing (which companies have been trying to solve by hoovering up all the DRAM it can possibly get) by making the very thing they’re trying to do so much smaller.

You saw the impact as memory company stocks fell immediately following this announcement. There’s of course a lot more technical information to this, but I’m just summarizing the impact here.

🤔 Skeptical take: Jevons paradox

In economics, Jevons Paradox argues the opposite of this. When a resource becomes more efficient to use, demand actually increases rather than decreases. So if TurboQuant compresses AI memory by over 6x, every company on earth could very well want to run over 6x more of it.

Instead of thawing high prices, ruthless efficiency like this could very well trigger a second, even larger wave of hardware buying and keep prices high well into next year.

OK, time for Plan B

(Image credit: Future)

I found this out while speaking to Acer’s EMEA Marketing Director, Manuel Linning, earlier this year, and the switch is incoming.

“Since this is an industry issue, we're all dealing with it in some different ways,” Linning commented. “What we're doing is we're looking at multiple smaller vendors, and also new vendors.”

These are Chinese memory producers CXMT and YMTC, and sources told UNN that Acer, alongside Asus, Dell and HP have been testing their silicon for the past few months.

And in talking to other experts in the area, I’ve found out that these companies are close to certifying these vendors. The transition is happening sooner than you think, and it’s going to change the 'Made in' sticker on your next laptop.

🤔 Skeptical take: The quality lottery

When alternative supply chains are brought in to keep prices low, the consumer can often pay in reliability. There’s a restaurant near me that has managed to keep their prices the same amid inflation, but over time the quality has fallen off. The same could very well happen here with higher memory failure rates and shorter life spans.

So sure, that laptop price could be $499, but we may be entering an era where you enter a lottery of which factory your laptop’s RAM came from.

An omen, not a guarantee

(Image credit: Future)

That being said, this is the first slither of good news that we’ve had in months, and the idea of a hardware thaw is definitely real. However, the path from a courtroom to a price tag in Best Buy is long and riddled with inventory that’s already been priced in at 2025’s peak rates, inevitable appeal battles over that $130 billion refund, and the looming shadow of Jevons Paradox.

My advice is simple: hold until August 2026. If you can wait until the end of back-to-school sales, do it. By then, we’ll know if refunds are happening, the RAM bypass has been successful, whether the consumer shield is holding up, and most importantly, whether the manufacturers are willing to share those wins with you.

We aren’t seeing a fire sale just yet, but the fever has officially broken.

What is Nvidia DLSS? Explaining the AI tech transforming PC gaming — and why it’s getting controversial

Jason England — Fri, 20 Mar 2026 11:47:10 +0000

AI is not just a chatbot you talk to — it’s the primary driver of modern PC gaming performance. Leading this charge is Nvidia’s DLSS, which can unlock high-end fluid visuals without needing top-tier hardware (provided you use it right).

And that’s why I’m here to break it all down. What is DLSS? How does it work? How can you make the most of it? And what is the deal with the DLSS 5 controversy? Let’s get into it.

What is DLSS?

(Image credit: Future)

Deep Learning Super Sampling (or DLSS for short) first debuted in 2018, and it’s a combination of AI tricks to boost a game’s performance — allowing players to prioritize either high-end visuals, faster frame rates or a balance of both. Neural networks have been fed millions of hours of gameplay and trained on Nvidia’s supercomputers to deliver this.

Back in 2017, Nvidia saw an AI revolution coming and introduced Tensor Cores into its GPUs. These are engineered to be the bedrock of deep learning and inference, which can essentially allow your graphics card to work smarter not harder in rendering games

Over the years, many layers have been added onto this AI-fueled cake. But at it’s core, DLSS does business in two ways:

Super Resolution: Rendering a complex AAA game at full resolution is a taxing ordeal — especially when you throw things like ray tracing into the mix. So with Super Resolution, the GPU renders the game at a lower resolution, uses AI to look at the picture and upscale it to look like native quality. This can massively reduce the hardware strain.
Frame generation: To smooth out the frame rate, DLSS is able to analyze every frame generated by the GPU and fill in the gaps with AI frames. With DLSS 4.5, Nvidia is able to inject up to six additional frames.

Beyond that, there are additional things Nvidia has thrown into the mix, such as:

Dynamic Frame Generation: Proactively adjusting how many frames it adds to adhere to the maximum refresh rate of your monitor
Ray Reconstruction: Instead of having the GPU brute force all the light and reflection calculations in path tracing (a crazy intensive task), Nvidia’s AI model replaces this with a neural network trained on how light works in the real world to do all those calculations instead.
Nvidia Reflex: We’ll talk more about latency in the next section, but Reflex is a key technology Nvidia introduced to keep games feeling responsive even when stuffed with AI frames.

And that in a nutshell is DLSS. But just like any technology, you’ll only get the best results if you use it right.

How to use DLSS the right way

If you’re looking at this thinking “I’ll just turn everything up to max, slap DLSS on it and call it a day,” stop it. That’s like trying to win a F1 race in a minivan just because you put high-performance racing tires on it.

And as the guy who has tested all the best GPUs, I’ve learned a thing or two about how to tame this beast.

My Super Resolution preferences

(Image credit: Future)

Yes, DLSS 4.5 super resolution is capable of some bonkers things (like upscaling a 240p picture to a better-than-expected 4K image). But as a general rule of thumb in real-world gaming, the more “input pixels” the AI has to work with, the more aggressive you can be with upscaling.

Here are the four DLSS modes you’ll see in your game settings:

Quality: This renders the game closest to native resolution (normally around 66-70% of resolution).
Balanced: Finding the mix between getting the best possible textures and frame rates, this will render the game at 58% of your targeted resolution.
Performance: One step down to maximize frame rate, this will often go for 50% of the resolution.
Ultra performance: If you want to go all out on frames, ultra performance will target 33.3% of the total resolution being rendered by the GPU.

Now I’m about to start talking about three presets here that you can access from the Nvidia app settings. These presets are three different neural networks Team Green has made to fuel its resolution scaling. Here’s a breakdown of them:

Model K: This is DLSS 4 and much better-built for quality and balanced DLSS modes
Model M: This is DLSS 4.5 with a model tuned for performance mode
Model L: This is an offshoot of DLSS 4.5, which is meant for ultra performance mode

Got that? Good. Now let’s break down which ones are best for you by the resolution you want to play at.

(Image credit: Future)

Gaming at 4K? Use Performance (or Balanced) Mode: In the Nvidia app, this is labeled as Preset M, and in my own testing, I’ve found it to be the sweet spot at 4K. This will typically render a game at 50% of the resolution (1080p), and given how good this model has become over the years, the difference between this and Quality mode is nearly invisible in motion (unless you’re seriously pixel peeping). But at the same time, the FPS gain is massive.
1440p or 1080p? Stick to Quality: Turn on Preset K and stick with Quality (or Balanced at a push). Performance or Ultra Performance can lead to some losses of finer details in complex textures like hair or grass — leading to a soft image. For example, Ultra Performance at 1080p would be trying to upscale from as low as 480p.

Chances are you can get away most of the time with just sticking to Balanced mode. But there are finer tweaks you can make to get even more out of it.

My Frame Generation preferences

(Image credit: Future)

So this is where I get into latency — the difference between how smooth a game looks and how it feels. Let me break it down.

Let’s say you’ve got a game playing at 30 FPS before you turn on any AI trickery. If you take that number and divide a thousand by it (1,000 ÷ 30), you’ll get the gap between frames which is 33.33 milliseconds.

Now if I turn on multi-frame generation at 4X and up that to 120 FPS, it’ll look smoother, but game inputs are locked at that 30 FPS, so it won’t feel as smooth. That’s why it’s incredibly important to start with a solid base rate for the game you’re playing before turning on frame generation.

Over my years of PC gaming, I’ve broken it down into three targets in my mind based on what I’m playing to find that right starting point:

If your game is a slower-paced single-player title, start at 40 FPS: Think like linear story-driven games where twitch reflexes are not needed. RPGs like Baldur’s Gate 3 are great examples of this, and if the game you want to play falls in this region, you don’t need a lightning fast base frame rate and can focus on upping the graphics settings.
If your game requires faster reflexes, aim for at least 60 FPS: Whether it’s racing in Forza Horizon 5 or needing to perfectly time parries in Black Myth: Wukong, if faster reflexes are needed, you’ve gotta close that gap between frames rendered before stuffing more AI frames in between them. This would reduce frametime to 16.67 milliseconds — a nice base to build on.
If you’re in seriously competitive multiplayer, either don’t use it or aim for 120 FPS: A sub 10 millisecond gap between frames is a critical helping hand to being competitive in online lobbies. If you’re particularly clutch and don’t want a single bit of AI trickery getting in your way, don’t use frame generation and focus on turning your settings down instead. But if frame gen is OK, start with a 120 FPS base.

DLSS 5 controversy explained

So now that you’ve got a grip on how to best use DLSS in its current form. Let’s look at where Nvidia’s AI tech is going, because DLSS 5 has caused quite a stir!

Coming this fall, the fifth generation goes beyond the “predictive” modeling of resolution scaling and frame generation, and fuses it with the “probabilistic” elements of generative AI. CEO Jensen Huang calls it the “GPT moment for graphics,” and this comes down to real-time neural rendering.

Basically, DLSS 5 looks at a game on a frame-by-frame basis, and rather than trying to “show you the game better,” it’s now aiming to “show you a better version of the game.” And the results of this pursuit for extreme photorealism is a bridging of the uncanny valley that has the gaming community worried.

Nvidia

Of course, we’re only looking at very small slithers of demos of an early build, but as you can see from certain screenshots, it looks as if the model is adding details that weren’t there originally — like additional hair, fuller lips or a sharper jawline.

There’s changes being made to the cinematic lighting and material depth too, which all contribute to another breakthrough, but one that developers are going to have to figure out the right way to use. Nvidia confirms this is an early build of DLSS 5, and probably with sliders turned all the way to max. So between now and the fall, devs will tune it however they wish.

Outlook

(Image credit: Future)

It’s not been the smoothest of rides for Nvidia’s DLSS. With each new version, there’s been a backlash of some sort. But every single time, it becomes the default and if you use this technology the right way, it can unlock gaming experiences unlike anything you’d see elsewhere.

Because more and more, it’s becoming abundantly clear that we’re reaching the limits of how many transistors we can stuff on a chip to make these things work through hardware. So in a moment like this, AI is being used to continue the march forward.

Is DLSS 5 going to be a moment of overreach? Potentially, but then again this is completely controllable by developers who will hopefully find that right balance. But for the PC gamer of today, DLSS is an essential tool in your arsenal that when tamed can be a breakthrough.

I pixel-peeped DLSS 5 — and now I can’t tell if Nvidia just changed gaming or broke it with AI

Jason England — Tue, 17 Mar 2026 12:12:03 +0000

For the first time in a long while, Nvidia GTC 2026 gave us some huge PC gaming news. CEO Jensen Huang introduced DLSS 5: the “fusion of 3D graphics and artificial intelligence” in his own words, and the response has been…mixed to say the least.

From outlets calling it the most impressive tech they’ve seen in a long time to people throwing out AI slop critiques and comparing it to those face filters you get with certain smartphone cameras, opinions are all over the place. And honestly, I didn’t know how to feel at first — like that damn dress meme in 2015, my mind changed every time I looked at the comparison videos!

@tomsguide
♬ original sound - Tom’s Guide

So I did what I always do. I followed my late Grandad’s advice to “sleep on it, because you’ll know how you feel in the morning.” And in short, DLSS 5 is absolutely a breakthrough, but game developers have until the fall to find the sweet spot for it in their games. Let me explain.

What is DLSS 5? Let’s explain it with pizza

(Image credit: Nvidia)

It feels literally like yesterday that I was talking about DLSS 4.5, but we’re already marching onto the next iteration — and it’s a big one we’ve known was coming for a while now.

“DLSS 5 is the GPT moment for graphics — blending hand-crafted rendering with generative AI to deliver a dramatic leap in visual realism while preserving the control artists need for creative expression,” Huang commented.

It’s a fusion of the “predictive” model that’s fueled DLSS for a while now with the “probabilistic” elements of generative AI to bring photorealism to games with cinematic lighting, enhanced material depth, real-time neural rendering and temporal consistency. And all of its capabilities are controllable by the game developers, as they can tune the intensity, color and masking to find an enhancement balance that is right for them.

To break this down, I’m feeling hungry and you know what that means… Back to the pizzeria I go!

Think about previous versions of DLSS as a magic magnifying glass. I love that 8-inch pizza, but I want more of it, so DLSS takes that pizza and stretches it to look like a 16-inch XL — making it bigger without the crust getting too thin or the cheese burning more easily (this is resolution scaling).

Then every now and again, the chef slides an extra few slices into the box between every real slice you actually order, so it feels like you’re eating more at a faster pace (frame generation).

Now, with the fifth generation, there’s an AI master chef that doesn’t just stretch the pizza, it re-imagines it. With settings turned up to max, the chef looks at that cheap pepperoni slice and says “I know what you were trying to do here, but I can do better,” and swaps it with artisanal, hand-cured salami and fresh buffalo mozzarella — even though you didn’t order those things.

Basically, DLSS 5 has stopped trying to “show you the game better,” and is now resorting to “showing you a better version of the game.”

Bridging the uncanny valley

Nvidia

We can’t say any of us didn’t see this coming. Jensen himself talked about it at CES 2026 in a behind-closed-doors Q&A session and shot his shot at a growing “fusion between rendering and generative AI.”

“In the future, it is very likely that we'll do more and more computation on fewer and fewer pixels. By doing so, the pixels that we compute are insanely beautiful, and then we use AI to infer what must be around it.” Huang said. He talked about the “utterly shocking and incredible” results he saw in the labs that looked like “basically a photograph interacting with you at 500 frames per second.”

And now we have our first glimpse at it. In some of the game videos it’s a significant improvement, but in others you can start to spot some creative challenges that will surely be worked out as we close in on an official launch.

Let's start with the faces (yep, we’ve got to talk about Grace’s face in Resident Evil: Requiem). There was a recent brain scan study that showed our brains process “hyper-realistic AI faces” differently than real faces. A sudden spike in activity around 600ms after seeing an image that triggers an internal uneasy mismatch feeling.

Nvidia

That’s what I believe is happening here, and a lot more of it comes down to the cinematic lighting than I initially thought. Based on my time pixel peeping videos and image comparisons, I’d say about 60% of Grace’s face tweaks here can be explained by lighting and material depth.

The remaining 40% is neural rendering — there are definitely fuller lips and sharper jawlines (not that Leon Kennedy needs it, being the smokeshow he is).

Nvidia

In some places, this photorealism really shines — games like EA Sports FC and Starfield truly benefit from this upgrade. But I can appreciate the view around creative interpretation vs AI creating a jarring effect.

There are some other things I noticed, too. Surfaces and textures have been given a serious upgrade, but with DLSS’ reinterpretation of lighting, some of it feels less stylized or moody.

Nvidia

Take this scene from Nvidia's Zorah Demo, for example. Before, there was a warmer hue and intentional shadowing to add depth, but that is re-interpreted with the cinematic lighting in a way that I feel loses the vibe a little.

All-in-all, DLSS 5 is a diamond in the rough. I can see what the intention is, but this is all completely in the hands of devs to use however they wish. Maybe Capcom rolls back on the face tech, or Warner Bros. alters the cinematic lighting. It will take time to find the right balance.

I’ve got some questions

Nvidia

But of course, this is one PC gamer who’s tested all the best GPUs and the tech that enables them — pixel peeping videos and screenshots. I think it’s something we all have to interact with to get a fuller understanding of DLSS 5.

And if Nvidia’s reading this (hi btw), before I (hopefully) get some hands (and eyes) on time with it, I do have some questions to get a better understanding of what’s going on under the hood:

What is that real-time neural rendering and what has it been trained on?
How big is the DLSS 5 model now, and what is the goal for it by fall? Currently demos showed on 2 RTX 5090s — so optimization/compression is key.
What are the controls for more stylized games? Titles that don't necessarily benefit from photorealism with cel-shaded or more artistic graphics.

DLSS5 outlook

(Image credit: Future)

The timing of these announcements always makes me chuckle. I know it may be unintentional, but it just feels like once another GPU company announces an update to its upscaling/frame generation tech, Nvidia just says “hold my beer” and takes another giant step ahead of the pack.

Intel XeSS 3 with multi-frame generation? How about a little DLSS 4.5. AMD FSR Diamond? Well, here’s DLSS 5. And honestly, it is a breakthrough — but I can totally understand the mixed response.

In some titles, it is a generational leap forward. In others, it can feel like an AI veneer. But ultimately, the tech is here and devs can turn it up or down however they want. Nvidia isn’t adding slop to games, that team is marching forward to photorealism with another feather in the cap of developers to bring their visions to life.

Nvidia says PC gaming will ‘look like a film’ — GTC 2026 could herald 1 million times better path tracing for GPUs

Jason England — Mon, 16 Mar 2026 11:52:58 +0000

In tech, a 2x improvement is great and 10x is generational. So for Nvidia to claim its future gaming GPUs will be 1,000,000x better at path tracing sounds like pure science fiction. But that’s exactly what the company said at GDC 2026, and what’s even crazier is that I think it’s closer than any of us think.

Because today, Nvidia has confirmed you will see the “future of real-time rendering” in CEO Jensen Huang’s GTC 2026 keynote — I think a sneak peek is possible. The obstacle to this at the moment is current GPU hardware is hitting a physical wall where we can’t just throw more electricity at the problem. And the fix? Well, it’s Nvidia’s bread and butter: AI.

Catch the future of real-time rendering in Jensen’s keynote tomorrow👀 https://t.co/KFv1JoTsDuMarch 15, 2026

What is Path Tracing?

This has been the new hotness in PC gaming. I tested it in Resident Evil: Requiem, and it’s rather incredible. But what it is and how to get there is an impossible-level challenge if using the old ways.

Put simply, it’s physical accuracy. It’s not just a reflection in a puddle, it’s how light from a neon sign bounces off a wet pavement, hits a character’s chrome jacket and subtly tints their skin red. It’s not just the ray traced shiny surfaces of old; path tracing calculates how light interacts with literally everything in a scene in a realistic way.

Path tracingFuture

No path tracingFuture

The end result is probably going to be subtle in some games, but in others (like Requiem) where the finer details really matter in building up the fear, it can be a game changer.

The death of brute force

(Image credit: Future)

Like I said, it’s an impossible challenge through raw hardware power. If you had your GPU just brute force full path tracing, your PC would probably melt before it rendered a single frame of Cyberpunk 2077. But as you may have seen reading my Requiem test, Nvidia bridges that gap with DLSS 4 — AI trickery with Ray Reconstruction.

Basically, instead of making the GPU seriously sweat trying to track a single ping-pong ball thrown in a room filled with a billion mirrors, these neural rendering technologies bring in an expert who has seen billions of balls thrown into that maze. They can just look at where you threw it and draw exactly where it’s going to land.

By offloading that demanding task to AI, path tracing becomes much more efficient to do, and then that 1 million time jump starts to sound more possible.

Neural rendering is the whole thing

(Image credit: Future)

In our time speaking with Jensen Huang, it’s clear that DLSS is just the start of AI’s impact on PC gaming. The limits of how much more complex GPU hardware can get are starting to be reached, and neural rendering will be the thing that fills in the gap — turning your graphics card into less of a calculator and more of an imagination machine.

And the results in his words are “utterly shocking and incredible.” He talks about it looking like “basically a photograph interacting with you at 500 frames per second.” And this ties in with what John Spitzer, vice president of developer and performance, said when talking about the future of PC gaming graphics at GDC 2026.

"We're still not where we want to be. We want the real-time images to look indistinguishable from reality. We want them to look like a film,” Spitzer commented.

GTC 2026 is the starting pistol

(Image credit: Nvidia)

And as these stars align around everything Nvidia is saying, that brings me back around to this “future of real-time rendering” promise for Huang’s keynote today. To be clear, this is not going to mean new gaming GPUs.

We already know that much from the supply issues around the current RTX 50-series, as Nvidia’s attention is being turned to building the picks and shovels for the AI gold race (or bubble…whichever way you look at it). But Team Green continues to march on in the gaming space, and this announcement should get PC players hyped.

I don’t think it’ll be a finish line, it’ll be a look at the next steps where we could go. After all, Vera Rubin is an architectural glimpse of what gains we may see in RTX 60-series, and that could fuel these big ideas.

Nvidia is bringing a killer app to the Apple Vision Pro — and it's good news for sim racers

Jason England — Tue, 10 Mar 2026 15:43:06 +0000

Nvidia’s going big at GDC before going even bigger at GTC. But there’s one quiet announcement amidst DLSS 4.5 dynamic multi-frame generation and GeForce Now upgrades that could be a massive one for anyone who owns an Apple Vision Pro — looking for reasons to use it.

Team Green’s CloudXR technology is able to stream PC VR games wirelessly from your gaming PC to the Vision Pro in glorious 4K 120 FPS, and their first target is the ultimate sim rig experience with support for iRacing and X-Plane 12.

Obviously, good news for sim pilots (shout-out), but as you may already know about me, I’m the sim racing guy on the Tom’s Guide team (my bachelor party is literally dedicated to it). One thing is clear: this is a breakthrough for VR drivers like me. Let me explain.

What is Nvidia’s CloudXR?

One thing I have to make clear here is that CloudXR is a very different thing to GeForce Now. Both are game streaming, but while GFN is from Nvidia’s cloud servers, CloudXR streams directly from your own hardware. You’ll need a beauty tower too, because its capabilities are significant.

Think of CloudXR as me vs a strong person taking in a bunch of groceries from the car. I can handle two bags, but if I try to handle more, it becomes a struggle. So my muscly friend comes in and takes the remaining eight bags. We’re technically working together, but let’s be real — they’re doing the heavy lifting (literally).

It’s GPU-accelerated VR delivered wirelessly to a headset with low-latency, which brings a serious graphical turboboost of that Nvidia GPU that you wouldn’t be able to do on the silicon of that standalone set.

And while some people have been talking about beta testing this new feature on the Vision Pro subreddit, this is the first time Nvidia’s officially letting the cat out the bag.

(Image credit: nDreams)

The big win here is foveated PC streaming. Basically, thanks to Apple Vision Pro’s eye-tracking capability, Nvidia’s CloudXR tech can see where you’re looking and render that part of your vision in full glorious detail — while leaving stuff in your peripheral vision and beyond less so. PSVR 2 does something similar with rendering graphics on the PS5, but it doesn’t hold a candle to this.

It’s a huge efficiency gain for your GPU, and provides much more headroom to be able to achieve up to full 4K gaming at 120 FPS in VR.

The Vision Pro’s killer app?

(Image credit: Future)

That question is a tricky one to answer — it’s a $3,500 headset after all. But any new features are warmly welcomed to try and get more value from what has quickly become the beta test device for the version of visionOS that will run on a future pair of smart glasses.

But take it from a VR sim racer: there’s no experience quite like it (…except maybe actually driving a race car, but there’s no risk factor here).

I know we’re all quick to talk about the “immersion” of something, and don’t get me wrong, it absolutely is. But there’s something here about it that actually makes you a better driver.

Take trail-braking for example (instead of hammering on the brakes to slow down for a corner, using a more gentle combination of brake and acceleration to maximize corner speed). It’s a tricky technique to learn and while I’m absolutely getting the feedback I need from my sim rig to ensure I’m not over-driving, being able to see the path of a corner is also critical.

(Image credit: Future)

Now, on the flat screen of a TV or monitor, that can be next to impossible on long-swooping turns or oddly-angled hairpins that disappear off the side. In VR, I can look to the apex and corner exit and aim my car for it.

Yes, I know this sounds small and that I’m getting really geeky here (sim racing is extremely my thing). But it’s oh-so critical to extracting the slight performance improvements that are the difference between a win and a loss.

Not the only XR upgrade

(Image credit: Future)

For the other update, we look over to Nvidia GeForce Now, where on March 19, Apple Vision Pro and Meta Quest headsets are getting a streaming upgrade to 90 FPS. It’s great to see these platforms continue to get supported — nothing quite like throwing on my Quest 3S in bed and playing PC games on a gigantic screen on my ceiling.

And as the subreddit community seems to confirm about beta tests, CloudXR is something that can be used across multiple games, which would be huge for the future of PC VR gaming. Getting those cables outta the way is a step towards true plug and play simplicity that can be the biggest thing holding potential players back.

I went eyes-on with Nvidia’s DLSS 4.5 Dynamic Multi-Frame Generation launching March 31 — and it’s a game-changer

Jason England — Tue, 10 Mar 2026 15:30:00 +0000

Nvidia launched DLSS 4.5 back in the middle of January and (spoiler alert) I’ve been really impressed with just how much detail is AI-infused upscaler is able to get from a rendered picture from just 25% the resolution. It feels like a real value booster for mid-range GPUs!

But there’s one feature that we’ve all been waiting for — Dynamic Multi-Frame Generation — the evolution of the frame gen we have right now. Basically, instead of just packing an arbitrary amount of AI-generated frames in between frames rendered by your GPU, dynamic frame gen is able to see what the max refresh rate of your screen is, and only generate the frames it needs to max that out.

We now have a launch date of March 31st (starting in beta), and I got some eyes-on time with the tech. And I gotta be honest with you, this feels like the frame gen tech that tackles the problems PC gamers have head on. Let me explain.

How does Dynamic Multi-Frame Generation work?

(Image credit: Nvidia)

Think about it like driving up a steep mountain, and you’re in a car with a manual transmission. To maintain speed, you’ve got to manually shift your gears (like shifting between the different multi-frame gen options), and those gear shifts can cause a clunky pause in your momentum.

Dynamic multi-frame gen works like a car with automatic transmission — intelligently monitoring how hard the engine is working (your system load) and automatically shifting gears (the frame multiplier) up or down depending on what the terrain demands.

And when the road flattens out and the workload lightens, the frame multiplier seamlessly downshifts so your system computes exactly what is needed in every scenario. The end result of this in my testing is an uninterrupted gaming experience, where you’ll really struggle to notice when that multiplier changes, and virtually eliminates any tiny hints of latency.

Gamesmaxxing

(Image credit: Future)

So I saw Black Myth: Wukong running on an RTX 5080 machine with multi-frame generation cranked up manually to 6X, and then The Outer Worlds 2 with RTX 5060 Ti with that dynamic frame gen turned on.

The first observation when it comes to Black Myth is just how minimal the latency impact is at full 6X. We are talking only a few milliseconds here, which I didn’t feel at all in the controller inputs — all while showing off exactly what DLSS 4.5 is able to do here.

For example, while toggling to DLSS 4 shows that some of the finer particle details and distant textures are smoothed out/removed, switching to 4.5 really brings these intricacies back to life. You can see it just by looking at the sparks of light around your weapon.

(Image credit: Future)

Then, here comes dynamic frame gen, which targets the max frame rate (240Hz) of the monitor. I got the help of an Nvidia rep to hide the stats in the top corner of the screen with his hand — so I can try to guess when the frame multiplier changes based on feeling alone.

Because visually, I could take an educated guess, right? Like, when the screen is not so busy, that multiplier would go down, but when there’s much more textures and characters to render, it’ll go back up. But the transitions felt oh-so-smooth that I didn’t notice a change to the speed of my inputs whatsoever.

That comes down to three key things:

Eliminating wasteful latency: Let’s say you’ve got a 144Hz monitor, and you’ve got DLSS running a game at 240 FPS. Those extra 96 frames are never seen, but the GPU still spends time computing them. This can create an interpolation gap — you’re over-exhausting your GPU with AI tasks it doesn’t need to do. By aiming for the max frame rate, the backlog is cleared and waste is eliminated.
Smoother frame pacing: Another side-effect of aiming squarely for your monitor’s exact frame rate is a consistent delivery of said frames. If you manually set to a certain multiplier, there can be a jittery feeling as you go between GPU-intensive levels and calmer ones. This stops the over-delivery of frames in those quieter moments and has it all running at the same pace.
Prioritizing the base frame rate: Any AI game frame generation is only as good as the base rate of how fast your GPU can render a game. Sometimes your game can look fast, but feel slow, and sometimes that can be exacerbated by clogging it with AI tasks. With Dynamic MFG shifting from 6X down to 2 or even 0X (native), this ensures that the card focuses more on rendering real frames faster.

The end result is a more tamed DLSS frame generation — only using its AI witchcraft when it’s absolutely necessary and making a game feel smoother and more responsive for it.

Outlook

(Image credit: Future)

So ahead of its March 31st debut (launching as part of an opt-in Nvidia app beta that you can turn on as an override), should you use it? Based on my time testing it, the answer is “absolutely,” but results may vary.

Of course, this is going to start life as a beta, so there’s that disclaimer out the way. Plus, I tested it in a very specific space ship environment in The Outer Worlds 2 — what would happen if I was to be in the throws of combat on a planet? How will results change across different games?

The answer, as we all explore this tech together, will change on a game-by-game basis — as more and more optimizations are made to work around it.

That being said, if this is what we can look forward to across the over 200 titles that support DLSS 4.5 (and the 20 native integrations currently in the works for upcoming games), this is the final piece of the puzzle that brings the vision of smooth, responsive AI fueled gameplay to life.

Nvidia GTC 2026: The biggest reveals we expect to see

tony.polanco@futurenet.com (Tony Polanco) — Sat, 28 Feb 2026 07:30:00 +0000

Nvidia GTC 2026 kicks off on March 16 and promises to deliver some big announcements from Team Green. While this is sure to be an AI-heavy show, we’re also expecting important hardware announcements, as CEO Jensen Huang recently teased that his company will unveil "several new chips the world has never seen before."

You may be wondering why you should tune into GTC, since it’s not as consumer-focused as CES. Even though there will be plenty of talk about data centers and AI, Nvidia is likely to drop hints or announce technology that could end up in the devices you use every day, like the best gaming laptops.

We’re covering all things GTC to keep you updated on Nvidia’s plans. Here are some of the biggest announcements we anticipate.

Nvidia’s mystery chip

(Image credit: Shutterstock)

Given Huang’s comments about a potentially revolutionary AI chip, we can expect a major reveal of a new inference-focused chip, possibly related to the Feynman architecture designed for faster AI tasks. This wouldn’t be a typical GPU, since it’s built for “agentic AI,” where systems act autonomously.

What does this mean for us? Imagine your phone or laptop running AI apps that can edit photos, summarize emails, or plan your day almost in real time. While these features exist, they can feel slow or unreliable. A chip like this could make AI more useful for everyday tasks, and do it all on-device instead of through the cloud. For gamers, it could lead to better upscaling tech, especially for lower- and mid-range PCs and laptops.

The Arm-based N1X CPU

(Image credit: Future)

Speaking about chips, there have been rumors about Nvidia’s N1X, an Arm-based CPU for the best computers and best laptops. But with N1X delays reported, GTC might only offer a sneak peek rather than a full launch. If this processor appears, expect discussions on integrated graphics that could challenge Intel’s Panther Lake.

Windows on Arm is now viable, thanks to Qualcomm’s Snapdragon X chips, so an Nvidia Arm chip would be huge. It could create a four-way race between Nvidia, Intel, Qualcomm, and Apple for who can make the fastest, most energy-efficient laptop chip. We might not get much N1X news, but it’s what I’m most excited about as a laptop reviewer.

Next-gen GPU tease

(Image credit: Shutterstock)

As we recently reported, Nvidia won't release new gaming GPUs in 2026, as they’re focusing on making chips for AI data centers rather than consumer GPUs. However, we might see some architecture teases.

I expect detailed looks at the Rubin GPU, which aims to provide five times the power of Blackwell for AI workloads. Feynman could be a significant leap and might use TSMC’s advanced 1.6nm process.

If you’re a PC gamer, pay close attention to what gets announced, as Nvidia’s data-center tech often influences what runs your PC. For example, Rubin could mean GPUs capable of 8K gaming or ray-traced worlds without overheating your system. Feynman’s efficiency could also improve laptop battery life. Plus, any new tech could lead to an upgraded version of DLSS, making games look and run better.

Physical AI and robotics

(Image credit: Nvidia GR00T )

I expect Nvidia to also focus on “physical AI,” which combines simulation with real-world robotics. We saw some of this at CES, and I expect the company to expand on it at GTC. Look for sessions on AI factories and multimodal systems that combine vision, language, and action.

It sounds like sci-fi, but if you’ve seen videos of robots doing coordinated dancing, you know we’re getting closer to a Jetsons-like world. Maybe one day we’ll see affordable home robots that can clean, cook, or help those in need, all powered by Nvidia’s Omniverse platform. Nvidia will certainly paint an exciting picture, and I’ll be curious to see if any of it becomes real products.

Agentic AI

(Image credit: Future)

You’ll hear about agentic AI, where models go beyond chatbots to multi-agent systems that work together on complex tasks. GTC will likely include workshops on LLM augmentation, prompt engineering, and deploying inference at scale. Essentially, making AI more proactive rather than reactive.

While current LLMs like ChatGPT are useful, they’re not perfect or truly “intelligent.” Agentic AI could deliver genuinely smart AI systems that learn instead of just predict. We’ve heard promises of AI serving as virtual assistants, but these often need a lot of guidance, which defeats the purpose. I remain skeptical about seeing true agentic AI soon, but GTC might start changing my mind.

Outlook

Although Nvidia GTC 2026 will focus heavily on AI, it’s worth tuning in to see what Nvidia is working on and how it could benefit our work and gaming experiences.

Nvidia GTC runs from March 16 to 19, and we’ll be covering it all. The “GTC Live 2026” pregame show starts on March 16 at 11 am ET (8 a.m. PDT), before Jensen Huang delivers his keynote speech at 2 pm ET (11 a.m. PDT). I’ll be watching to admire his leather jacket and hear his insights on Nvidia’s latest plans. Stay tuned for more!

It’s been 25 years since Nvidia GeForce 3 — and I think gamers accidentally built the AI era

Jason England — Fri, 27 Feb 2026 18:06:50 +0000

Nvidia GeForce GPUs have been around since 1999, but it was with GeForce 3 that changed the rules of the game forever. Today is the 25th anniversary of the third generation GPU tech, and you can draw the past quarter of a century of generational leaps in gaming back to this one pivotal moment.

And that change? Programmability. Instead of just feeding data to the chip, people were able to run small programs directly on the GPU — giving birth to the rampant progression towards the incredible fidelity we see on PC games. But in an ironic twist for gamers in 2026, it’s that very programmability that makes it impossible to find the best GPUs at MSRP.

Plus, you can also see how the sheer size of GPUs have grown over time too… Yes this is technically the GeForce 210 that I picked up at Guang Hua Digital Plaza last year. But this is roughly the same dimensions of a GeForce 3 GPU back in the day (around 6.5 inches in length).

‘Your world, programmed’

So let’s look back. In 2001, Nvidia CEO Jensen Huang talked about GeForce 3 as a way to “unleash cinematic realism,” all by moving away from “fixed-function” chips and giving creators a blank canvas to make with.

This came down to the introduction of Programmable Vertex and Pixel Shaders. Think of a 3D video game like a movie set. To get a character on screen, the computer has to figure out where things are and what they look like.

Before GeForce 3, these jobs were hardwired into the chip and developers were stuck with them. If you wanted to draw water in a harbor scene, you were stuck with the water Nvidia’s engineers gave you.

(Image credit: Wikipedia)

Now, with this increased programmability, the Vertex shaders (the “where”) are able to be controlled by the coder — like a construction worker moving the scenery around. And on top of that, the programmable pixel shaders (the “what color”) was also able to be controlled.

You could start telling the lighting crew what to do, figure out whether that floor is supposed to be wet, dusty, bumpy, or whatever you want it to be!

And it was this that gave birth to those moments of cinematic realism. I stared at the water in “Elder Scrolls: Morrowind” for far too long, and “Doom 3” was utterly terrifying. Plus, with GeForce 3 architecture powering the original Xbox, there was a night and day difference between it and the PS2.

Paving the way

(Image credit: Future)

Fast forward to today and you can see how that programmability of GeForce 3 paved the way for CUDA and Tensor cores. This is the age of neural rendering, and Jensen said as much in my Q&A panel with him back at CES 2026.

And the aim seems to be moving beyond the architectural limits of GPU silicon, looking past cinematic realism and producing extreme photo realism: “basically a photograph interacting with you at 500 frames per second” in his own words.

Path tracingFuture

No path tracingFuture

You can see just how much neural work goes into something like “Resident Evil Requiem” — using a combination of DLSS 4.5 and Ray Reconstruction AI trained on millions of hours of gameplay to make a game run smoother and look prettier.

It’s the difference between a developer in 2001 saying “I will write a program to tell this pixel to look blue,” and one in 2026 saying “I’ll let an AI model look at the scene and decide what the pixel should look like, based on billions of previous examples.”

And it’s that very breakthrough that has made Team Green the belle of the AI ball, and that’s come at a bit of a detriment to gamers.

The cost of intelligence

(Image credit: Amazon)

You saw it in Nvidia’s recent earnings and the ominous warning that RTX 50-series GPUs were going to be hard to come by because of the RAM price crisis. While the company still makes billions of dollars on gaming tech, it makes hundreds of billions fueling the AI factories you see popping up all over the planet.

The silicon that once lived in our towers is being diverted to build the future of autonomy, and while I don’t fault the decision (I’d make the same one too in Jensen’s leather jacket), you can’t help but feel like as the GPU figured out how to learn, the world now wants it for everything except games.

The legacy

(Image credit: Future)

“Without GeForce, there would be no AI today. Without AI, there would be no DLSS today. It's harmonious," Huang said back in January.

While it’s a little tricky to see the harmony with GPUs far exceeding their MSRPs, I think it’s even more than that. In 2001, the programmability of GeForce 3 was initially thought to be a way to give us nicer water effects in games.

But Jensen’s long-term bet was that if you make a chip programmable enough, the world will find a use for it beyond games. And boy did they, looking at the most valuable company on the planet.

We are now living in the world that GeForce 3 built. 25 years ago, it delivered on the promise of “infinite effects” so well that the technology outgrew the toybox and the world realized that the most powerful tool for the future of humanity was sitting inside our gaming rigs all along.

And with Vera Rubin being essentially a sneak peek at what the RTX 60-series architecture will be able to do, this train isn’t stopping. But we may have to wait a while for the stock alerts to catch up.

Resident Evil Requiem made me believe in path tracing — and I wish it hadn’t

Jason England — Wed, 25 Feb 2026 15:01:24 +0000

So I’ll be honest. I’ve been a path tracing doubter for a while now. The idea of slightly more realistic-looking textures, shinier lights and better reflections just felt a little bit unnecessary in the games I had tested so far on Nvidia’s RTX 50-series GPUs.

But then I got my hands on Resident Evil Requiem, and I’ve been converted. The difference is night and day with path tracing; a new level of fear and tension is added in the subtleties of light and dark, and the immersion is taken to new heights I’ve never seen before.

What is path tracing?

This is like the gold standard of computer graphics — a generational leap over ray tracing that is able to calculate how all direct and indirect light in a scene bounces off surfaces.

In the past, this effect was done through rasterization, which is to paint these lighting effects onto the scene itself. Then, ray tracing was the next step to calculate how direct light sources interact with one another (like a reflection in a window).

Now, path tracing is able to handle both direct light and indirect illumination, while calculating where that light will bounce multiple times in a scene. This is, of course, incredibly demanding on a GPU, and in Resident Evil Requiem, DLSS is required to turn it on.

And on top of that, Nvidia’s Ray Reconstruction tech helps things along by using an AI model to recognize those traced rays of light in real-time and enhance the image quality.

By the numbers

(Image credit: Future)

With that in mind, let’s get straight into the testing. I ran Resident Evil 9 on three different machines:

The first I built myself at Scan computers, which packs an AMD Ryzen 7 9800X3D, RTX 5080 and 32GB of DDR5 RAM. You can watch the whole build here.
The second is an RDY Element 9 Pro R07 with AMD Ryzen 7 7800X3D, RTX 5060 Ti (16GB) and 32GB of DDR5 RAM. Typically, this prebuilt comes with an AMD Radeon RX 9070 XT, but I switched the GPU for this test.
The third is a Quoted Tech Frontier, a prebuilt sporting an AMD Ryzen 7 9700X, RTX 5070 and 32GB of RAM. It’s a PC that can muster up solid 4K performance when pushed to its limits.

To make this a stress test but fair to each card, I targeted each one’s preferred resolution — RTX 5080 at 4K and RTX 5060 Ti at 1440p. While an RTX 5070 is generally better suited to 1440p, I decided to turn this up to 4K to see how far I could push it. All have their graphics settings turned up to the max with path tracing turned on.

Latency checker

Now, I take frametime from each baseline test without frame generation because this is important to measure any potential latency. It’s a technical explanation, and there is low-latency tech Nvidia uses to tackle this. Put simply, your computer is “blind” to your mouse movements between frames, so the longer that is, the more prominent the latency can feel.

Throw frame generation on top of that, and while you’ll have a game that looks smooth, it may not feel smooth. But, that rule doesn’t apply with the same level of importance across the board, and Resident Evil Requiem is the peak example of this.

I normally follow a golden rule: start with a base rate of 60 FPS and build from there. That way you get much better responsiveness in your games. However, given the slow, methodical gameplay in Resident Evil, I didn’t feel that latency impacted gameplay — even at the higher end of 26.1 milliseconds.

If gameplay were fast and furious (like a multiplayer shooter that required twitch-like responses), this would be a different story. But as I’m about to show you, the pace of Resi 9 means you can turn up that fidelity and not break your immersion. This is a real masterpiece.

Turning down the difficulty

(Image credit: Future)

Yes, that’s not just a cool subheading. That literally happened. I’ve always been a bit of a coward when it comes to horror games, but with path tracing, I noped out at my first monstrous jump scare after a pitch-perfect build of tension and turned down the difficulty. It’s seriously a huge noticeable uplift.

Path tracingFuture

No path tracingFuture

Whether it’s light dancing on floor tiles covered in blood, your terrified face staring back at you in the glass of a fire extinguisher case or the shadows being so accurately cast on a bunch of zombies shuffling towards you, this is the first time I looked at path tracing and thought “I get it now.”

And when it comes down to the differing levels of DLSS 4 to 4.5 — moving from a balanced AI upscaling to one leaning towards performance and ultra performance — there are some telltale signs like a little bit more fuzziness, especially in the rain-soaked intro.

DLSS 4 BalancedFuture

DLSS 4.5 Ultra PerformanceFuture

However, it is quite the achievement to have the game looking this good from such a small picture. For context, DLSS 4.5 ultra performance takes an image rendered by the GPU at a third of the size. That means a 720p picture blown up to 4K, and a measly 480p image up to 1440p.

Yet another example of how Team Green runs this town when it comes to AI trickery to extract better bang for your buck from your GPU.

A new level of fear

(Image credit: Future)

I will never doubt path tracing again. While it is the case that in some titles that sport it, the difference is small, when used intentionally, like what we see here, it makes for a huge upgrade.

Not only is Resident Evil Requiem a visual tour de force through-and-through (thanks to Nvidia’s tech) with frame smoothness and no noticeable impact to latency, it’s a great game, too — even for scaredy cats like me, I had a great time.

If you’ve been looking for titles to both show off what your gaming tower is capable of, while also actually getting an all-immersive single-player experience out of the deal, this is the one to get.

DLSS 4.5 wins big: 48% of gamers pick Nvidia over AMD FSR (and native) in blind gaming test

Jason England — Thu, 19 Feb 2026 12:42:09 +0000

The choice of which GPU to buy for your PC isn’t really about just the hardware anymore — it’s also about the AI trickery to make your games look better and run smoother too. And there are two key players at the moment: Nvidia’s DLSS 4.5 and AMD FSR 4 (a core component to FSR Redstone).

But which one looks better? Well, In a blind PC gaming test conducted by ComputerBase, 48% of players chose Nvidia DLSS 4.5 over AMD FSR and even native rendering — signaling a strong lead in AI upscaling.

By the numbers

(Image credit: Future)

ComputerBase’s study took the total of 6,747 votes across six games — all of which are pretty well-optimized for both AMD and Nvidia GPUs. To test this, three systems were shown to gamers:

One with Nvidia DLSS 4.5 running (set to quality mode)
One with AMD FSR 4 running (set to quality mode)
One with neither — running at native resolution with Temporal Anti-Aliasing (TAA — more on this later)

And here are the results.

(Image credit: ComputerBase)

What is abundantly clear here is that while AMD has made some significant steps forward with its new FSR tech, Nvidia still holds a commanding lead with DLSS 4.5 and all of its AI tech.

FSR 4 still falls for some of the classic things you see with AI upscaling like fizzling jaggedness on finer geometry, ghosting around fast-moving objects, and some shimmering around brightly-lit scenes.

But, interestingly, while there were huge gaps between DLSS 4.5 and FSR 4 in “Satisfactory” and “Horizon Forbidden West,” many users (23%) couldn’t see much of a difference between Nvidia and AMD in “Cyberpunk 2077.”

So while Team Green has a fairly large lead on average, in certain circumstances (like 4K high-fidelity environments in this particular game), that gap is narrowing.

The biggest surprise?

(Image credit: Future)

The one that shocked me the most was to see DLSS 4.5 so handily beat native resolution too! But there is a logical reason for this, because the idea of “native” is almost never raw pixels.

You’ll see most games use something called Temporal Anti-Aliasing (TAA) to smooth out the jagged edges by blending the current and previous frame together — no AI required. But in smoothing them out like this, it can often create a blurry image during movement.

(Image credit: Future)

Because DLSS replaces this native TAA with its own AI-trained transformer model, which uses a massive training dataset to “know” what specific objects should look like, it can produce a much sharper and stable image.

What is a little odd, though, is how AMD’s FSR 4 came in third, because surely this should land above Native + TAA. But ComputerBase did have an answer for this, saying that the results don’t show “what is perceived as the second and third best.”

The question asked was to rate what had the “best image quality,” so to say that “FSR Upscaling looks worse than native” would be wrong given the skewing of the data based on how the study was done.

Nvidia's Jensen Huang says it will ‘surprise the world’ with a new mystery chip — here's what to expect

Darragh Murphy — Thu, 19 Feb 2026 12:26:31 +0000

Nvidia GTC 2026 is gearing up to be quite the show, with CEO Jensen Huang now hyping up the annual AI conference with "several new chips the world has never seen before." But what chips can we expect?

During an interview with media site Korean Economic Daily, Huang expressed that GTC 2026, taking place from March 16, will unveil "a chip that will surprise the world." While there's no indication of what this chipset could be, it's easy to think the upcoming Nvidia N1X for PCs is a likely candidate. However, this may not be the case.

Huang is teasing something big from Team Green next month, and while it's looking to be next-gen chips to power future AI data centers, it could be the beginning of the architecture that we'll see in RTX 60-series GPUs.

A mystery chip to 'surprise the world'

(Image credit: Getty Images)

During the interview, Huang notes that "nothing is easy because all technologies are at their limits," but with the partnership between Nvidia and SK Hynix, one of the three semiconductor companies, "nothing is impossible."

This only makes Huang's statement of revealing "several new chips the world has never seen before" at GTC 2026 more bold, and while the long-anticipated Nvidia N1X and N1 chips for desktops and laptops are a good contender, looking to take on Intel's Panther Lake CPU lineup (especially in terms of integrated graphics performance), we've heard N1X is rumored to be delayed.

As with past GTC events, Nvidia generally focuses on the expansion of AI-driven software and chipsets, and we've heard about Team Green's new Vera CPUs and Rubin GPU that offers 5x more power than Blackwell to drive the AI data centers of the future at CES 2026.

Nothing is easy because all technologies are at their limits
Jensen Huang, CEO of Nvidia

This indicates we could see its new Rubin GPUs for AI take the spotlight at the show, as they've been in full production as of January 2026, as announced by Huang during CES 2026. However, Nvidia could throw a curveball by revealing its Feynman microarchitecture for next-gen GPUs.

That's considering Huang stating, "this team has worked incredibly hard to meet the great challenge of Vera Rubin and HBM4," referring to SK Hynix and its development of HBM4 (High Bandwidth Memory 4), designed for AI data centers and cutting-edge computing.

This remains to be seen, of course, but if Nvidia is set to unveil a world-surprising chip, then we can expect Team Green to bring the next evolution of implementing AI in the entire tech industry.

A glimpse of RTX 60-series?

(Image credit: Future)

With reports of Nvidia not releasing a new gaming GPU this year, this not only pushes back RTX 50 Super GPUs, but also RTX 60-series. And this is due to the RAM crisis. For consumers, this is disappointing, and having a new chip for AI data centers likely being the big surprise at GTC 2026 isn't ideal.

However, consider this. Nvidia Blackwell architecture (what you'll find in the latest RTX 50-series GPUs) was first designed for data centers, and then for consumer GPUs, around a year or two later. With this in mind, with the developments of Rubin now in place, we may see how this architecture comes to Team Green's next-gen graphics cards.

For now, we'll have to wait and see what Huang has up his (leather) sleeve, and it won't be much longer, seeing as Nvidia GTC 2026 is kicking off on March 16. Don't expect all-new RTX GPUs or N1X, but do expect a big reveal that could deliver a taste of Nvidia's consumer graphics cards lineup down the line.

Nvidia won't release new gaming GPU for 'first year in three decades' due to RAM shortage — and it's also slashing RTX 50 production

tony.polanco@futurenet.com (Tony Polanco) — Thu, 05 Feb 2026 16:00:27 +0000

For the first time in three decades, Nvidia is reportedly taking a year off from the gaming GPU market.

According to a new report from The Information, Team Green has scrapped its plans for any new graphics card releases in 2026. The reason? As you might have guessed, the ongoing global memory shortage, or what we at Tom’s Guide refer to as “RAMageddon,” has become so severe that Nvidia is being forced to choose between gamers and feeding the AI beast.

Unfortunately, AI is winning. Here’s everything you need to know about this latest development in the ongoing RAM crisis saga

The end of a 30-year tradition

(Image credit: Future)

NVIDIA has released at least one new gaming GPU every single year since the early 90s. Even during the height of the crypto-mining craze and the pandemic supply chain collapse, the company managed to get something onto store shelves.

But 2026 is different. RAMageddon has hit a point where AI data centers are simply gobbling up the raw materials for memory.

According to The Information’s sources, Nvidia has indefinitely delayed "Kicker"—the code name for the incremental RTX 50-series refresh, likely the "Super" cards that we expected to see this year. Even though the designs were reportedly finished, the company decided that it couldn't justify the memory costs.

RTX 50 production slashed

(Image credit: Future)

It’s not just about the lack of new cards, either. The report also states that Nvidia is slashing production of its current RTX 50-series GPUs. With memory supply this tight, Nvidia seems to be prioritizing its limited stock for AI chips.

This is obviously bad news for gamers. We’re already seeing RTX 50 prices spike at retail due to scarcity, and these production cuts mean that mid-range favorites like the RTX 5070 and 5060 Ti are going to be even harder to find at MSRP.

Don’t expect the RTX 60-series any time soon

(Image credit: Future)

If you were hoping to skip this generation and wait for the RTX 60 series, you're likely going to wait longer than you expected.

The delay in the 2026 refresh has reportedly pushed mass production of the next-generation "Rubin" gaming GPUs to the end of 2027. If that’s the case, then we are looking at a nearly three-year gap between major GPU generations.

On a related note, since we’re discussing delays, Moore's Law is Dead reports that Nvidia's N1X CPUs are experiencing bugs and software issues, delaying when we'll see Arm-based Windows laptops. Apparently, neither Microsoft nor Nvidia is rushing to ensure the SoCs (System-on-Chip) are fully supported on Windows.

Outlook

This latest news, unfortunately, validates everything we’ve been saying: the PC market is effectively in survival mode. When Apple CEO Tim Cook and Nvidia's own leadership are sounding the alarm on memory prices, you know things are serious.

If you’re currently building a PC, my advice remains the same: stop waiting for the next big thing and grab what you can find now. Whether that's an existing RTX 50 card or building around a DDR4 platform to save cash, the goal for 2026 is to create a rig that works, even if it doesn’t have the latest and greatest components.

As always, we’ll monitor this situation and anything else related to the RAM crisis, so stay tuned!

EXCLUSIVE: Intel confirms XeSS 3 Multi-frame generation is coming to Arc B580 next month — and it’s now my top GPU recommendation

tony.polanco@futurenet.com (Tony Polanco) — Tue, 27 Jan 2026 16:55:44 +0000

If you've followed my coverage over the years, you know I'm usually the first person to tell you to be wary of Intel Arc GPUs. While I've always appreciated how "Team Blue" tried to break up the Nvidia and AMD duopoly, the driver issues and inconsistent performance of the early Alchemist era made them a hard sell for anyone who just wanted their games to work.

But the world has changed. We're in the middle of a full-blown RAMageddon crisis that has sent the cost of VRAM-heavy cards into the stratosphere. While the GPU market feels more volatile than ever, Intel is doing something impressive: they’re actually making their budget card better through software.

Intel has confirmed to Tom's Guide that the Intel Arc B580 is officially getting XeSS 3 with Multi-Frame Generation (MFG) next month. My colleague Jason England reviewed this card last summer, and between his testing and this new confirmation, the verdict is clear. At $300, this is a legitimately great choice for budget-conscious gamers.

The Intel Arc B580 is easily the best budget gaming GPU you can buy right now. At $299, you won’t find anything else new that comes close for 1440p gaming. Just mind that ray tracing setting and the multipurpose utility of the card aren't the best.View Deal

Why the B580 makes sense in 2026

(Image credit: Future)

I normally wouldn't recommend a $300 Intel card when you could save for a GeForce equivalent. But the RAM crisis has tripled the cost of memory, leading to some truly obscene price hikes. When an RTX 5070 Ti is pushing toward $1,000, looking for alternatives isn't just "skepticism"—it’s financial survival.

The Arc B580 currently costs $300. For that price, you get a generous 12GB of VRAM and a 192-bit bandwidth. In Jason’s testing, he found this allowed the B580 to punch above its weight class in 1440p gaming:

Cyberpunk 2077 (1440p Ultra, No RT): 90 FPS
Forza Horizon 5 (1440p Ultra, No RT): 106 FPS
Black Myth: Wukong (1440p Medium, No RT): 77 FPS

It’s important to note that performance results are sensitive to your specific rig. If you're running at lower resolutions or using a slower CPU, you might hit a bottleneck where the GPU can't fully stretch its legs. But for pure texture-loading and future-proofing, having 12GB of memory at this price point is a huge advantage over 8GB competitors.

DLSS vs. XeSS

(Image credit: Future)

Now, let’s be fair—Nvidia is the pioneer here. Their DLSS (Deep Learning Super Sampling) ecosystem is the gold standard for a reason. With support across more than 450 titles and the recent DLSS 4.5 demo at CES, Nvidia has a massive lead in maturity. You can even access newer features via overrides in the NVIDIA App, which is the kind of forward-looking support that pro-sumers love.

However, Intel’s XeSS 3 is finally bringing "secret sauce" parity to the budget level. Intel’s AI-driven multi-frame generation is arriving for the Arc B580, and based on what we’ve seen on Core Ultra Series 3 laptops, the results are startling. Turning a 60 FPS experience into 200+ FPS via AI is no longer an Nvidia-exclusive luxury.

By force-enabling this tech on titles that already support XeSS 2, Intel is closing the gap. In my own tests on a Samsung Galaxy Book6 Pro, I saw Cyberpunk 2077 hit 65 FPS at 1800p on Ray Tracing Low—impressive for integrated tech, and a sign that the desktop B580 is about to get a massive second wind.

The trade-offs: RT and productivity

(Image credit: Future)

Of course, you get what you pay for. If you like Ray Tracing, Nvidia still holds the crown. In Alan Wake 2, the B580 drops to 45 FPS at 1440p High when RT is involved—playable, but not as smooth as what you’d get from an RTX card.

Additionally, if you use your PC for professional AI work or video rendering, Nvidia’s CUDA cores still smoke the competition. In the Procyon AI benchmark, the RTX 5060 maintains a noticeable lead. For productivity, Nvidia is still the clear choice.

This tall ABS Cyclone Aqua PC delivers the goods for under $1000! This one packs an Intel Arc B580, Intel Core i5-13400F, a hefty 32GB of DDR4 RAM and a 1TB SSD. With top-down RGB air coolers, too, expect this machine to stay nice and cool will blazing through games at 1440p.View Deal

Bottom line

The Arc B580 isn't an Nvidia killer, but it is a value king. Thanks to the current memory shortage sending prices through the roof, it’s the only card in the $300 range that doesn't feel like a compromise on VRAM.

Nvidia continues to lead the way in software maturity and high-end Ray Tracing, but for the average gamer who just wants to play modern titles at 1440p without spending a month’s rent, the Intel Arc B580 is a serious, budget-friendly contender that's only getting better with age.

‘MSRP is a myth’? I disagree — here’s how to find a GPU and avoid the RAMageddon markup

tony.polanco@futurenet.com (Tony Polanco) — Fri, 23 Jan 2026 11:30:00 +0000

Building or upgrading your PC isn’t easy nowadays, thanks to the ongoing RAM crisis — or RAMageddon. Due to the global shortage of GDDR7 memory and the growing demand for AI-capable hardware, GPU prices continue climbing — seemingly with no end in sight. If you’re a tech nerd and gamer like me, this isn’t a fun time for our favorite hobby.

While Nvidia says all RTX 50-series GPUs will 'continue to ship,' stock and supply tell a different story. Right now, it’s virtually impossible to find popular graphics cards like the RTX 5070 Ti at its original $749 price, or find it at all.

However, you don’t have to pay these inflated prices for a new GPU. If you’re fine with alternatives from AMD and Intel, using automated tracking tools, or even doing what I did during the 2021 chip shortage and buying a pre-built PC, you can get a reasonably priced GPU that won’t crush your bank account.

Below, I detail some of the strategies you can take when buying a GPU this year, and what prices you should (and shouldn’t) settle on. Let’s dive in!

Price check

(Image credit: Future)

Before you start shopping for a GPU, it’s good to know what cards are supposed to cost. Retailers often treat MSRP as a suggestion, as you’ll often see prices at more than 10-15% their original cost. If you see these inflated prices, I advise you to walk away. Here’s a handy reference guide for you to follow.

Nvidia RTX 50-Series (Blackwell)
Model	Launch MSRP	Current status
RTX 5090	$1,999	AI demand is pushing street prices toward $3,000+. Unless you have cash to burn, please do not pay this much for a graphics card.
RTX 5080	$999	Expect long waits or bundle-only deals.
RTX 5070 Ti	$749	Beware $1,100+ third-party listings.
RTX 5060 (8GB)	$299	8GB VRAM is easier for factories to source.

Alternatives (AMD and Intel)
Model	Launch MSRP	Current status
RX 9070 XT	$599	Often found at MSRP at Micro Center, especially in-person.
Arc B580 (12GB)	$249	The most stable stock in 2026.

Red flags to watch out for

(Image credit: Future)

This is something I learned while doing research for this article. Scalpers can hide behind the “Third-Party Seller” tag on legitimate sites like Amazon and Newegg.

That’s a frightening reality, but here are ways you can vet each listing to ensure you’re not getting ripped off.

The "sold by" rule: Only buy if the listing says "Sold and Shipped by [Retailer Name].” If it says "Sold by GPU-King-2026," or another ridiculous name you’ve never heard of, that’s a marked-up price.
The VRAM premium: High-VRAM cards (16GB+) are seeing massive hikes because GDDR7 is scarce. If you mostly play games at 1080p, I recommend sticking to 8GB or 12GB cards to avoid the "AI-tier" memory premium. I know that sounds low, but trust me, that’s more than enough VRAM for 1080p gaming.
The "old stock" rumor: In a Q&A at CES 2026, Jensen Huang hinted that Nvidia might bring back older GPUs to solve the crisis, calling it "within the realm of possibility" (via TechRadar). Please do not pay 2026 prices for 2023 tech! If you see an RTX 3060 for $400, that’s a trap.

Buy a pre-built PC

(Image credit: Future)

If you’re not having any luck finding a standalone GPU, you can always do what I did and just purchase a pre-built PC. I know buying an entire computer for the GPU sounds wild, but during the chip shortage, it was how I snagged an RTX 3080 Ti. It worked during the chip shortage, and now it also works during RAMageddon.

Major brands use long-term contracts to effectively lock in component prices. Because of that, it’s often cheaper to buy a whole new PC with an RTX 5070 than to buy the card alone in the secondary market. Here are some tips when buying a pre-built PC.

Dual-channel RAM: Make sure your PC has two sticks of RAM. A single 16GB stick can bottleneck performance by 15%, so please avoid that.
Standard parts: If possible, try to avoid proprietary motherboards. If they aren't standard ATX, you won’t be able to easily upgrade later. You want industry-standard parts.
Storage size: With many new and upcoming games hitting 150GB+ of storage, don't settle for less than a 1TB Gen 4 NVMe SSD.

Automate your searches

(Image credit: Future)

Manually searching the web for good GPU prices isn’t efficient. Here are some tips on how to automate your hunt:

HotStock.io: This app pings your phone the moment a card hits MSRP at a major retailer, such as Best Buy. Be sure to buy your GPU manually and avoid the “Autobuy” feature, as some users report it can be unreliable.
Micro Center (in-store only): If you live near one, go in person. They often have "secret" stock reserved for local customers to prevent bot-buying.
Best Buy Founders Edition: Nvidia’s "Founders Edition" cards are the only ones locked to their original MSRP. Set your alerts specifically for these.
Price trackers: Use CamelCamelCamel. If the price graph shows a vertical spike in the last 30 days, wait for the dip. I use the CamelCamelCamel extension when writing deals posts, so I highly recommend it.

Which cards to buy?

(Image credit: Intel)

If you’re willing to step outside Nvidia’s green halls, you can save big on a GPU that delivers excellent performance for modern games. Here are a few you should consider.

Intel Arc B580 ($249): Intel’s Arc GPUs are often overlooked (even by me), but given the current climate, it’s good to expand your options. The Intel Arc B580 has an MSRP of $249. This is a good GPU for 1080p gaming.
AMD RX 9070 XT ($599): If you want a higher-end GPU, AMD’s RX 9070 XT is a great option that can offer similar performance to the RTX 5070 Ti, but at a much lower price.
Nvidia RTX 5070: If you’re strictly Team Green, then you can buy a pre-built RTX 5070 machine from the likes of CyberPower or Newegg’s ABS brand.

Bottom line

I know that finding a GPU for a reasonable price is tough these days, but if you’re patient and strategic, you can score a graphics card without paying through the nose.

The window for finding a “true” MSRP card is shrinking fast, so you have to decide if you’re willing to pay exorbitant amounts of money for an Nvidia GPU, or if you’d rather look to the likes of Team Red or Team Blue to get what you need. Then again, integrated graphics have gotten surprisingly good recently, so there's always that option to hold you over until you can find a GPU you like at the right price.

In short, try to avoid buying a GPU from third-party vendors you’ve never heard of, automate your searches, or buy a pre-built with the components you want.

Nvidia RTX 50-series GPU production reportedly 'paused,' but company insists it will 'continue to ship' all cards

Darragh Murphy — Thu, 22 Jan 2026 12:55:41 +0000

The state of Nvidia's RTX 50-series GPUs is in flux. With the RAM crisis causing prices to skyrocket and supply to dwindle, Nvidia was put in the position to confirm that it will “continue to ship” its graphics cards. But a recent leak claims otherwise.

Speaking with industry sources, reputable tipster Moore's Law Is Dead (MLID) claims Nvidia is now putting its entire RTX 50-series GPU production on hold, and it's in order to meet AI demand. What's more, the RTX 5060 with 8GB of GDDR7 VRAM will be cut for the foreseeable future.

According to the leak, which sees several sources commenting on the cut in RTX 50-series production, RTX 5060 supply will "get worse until at least Q4 [2026]," while other models in the lineup, including the RTX 5060 Ti with 16GB, RTX 5070 Ti, RTX 5090 will be "unobtanium" — in other words, unobtainable.

Yep - now the 5060 is done...and it get's worse...In tonight's video, I leak that @nvidia's effectively pausing production on RTX 50 Series for the foreseeable future in order to meet capacity for AI customers.🤦‍♂️Yes. The entire series. Effectively.https://t.co/t5ZTe9ADZbJanuary 22, 2026

As MLID puts it on a post on X, the "5060 is done," and "the entire series" is put on hold. But going into the comments, there's more to it, as GPUs such as the RTX 5050, RTX 5060 Ti with 8GB, RTX 5070 and RTX 5080 will still be available, albeit still in low supply.

Previously, Asus had allegedly put the RTX 5070 Ti into End of Life (EoL) status, but this was retracted after Nvidia claimed RTX 50-series would continue to ship as usual. However, with the memory shortage hitting the tech market hard, and Gigabyte stating Nvidia's GPU strategy is about "revenue per gigabyte," it's looking more likely that Team Green's graphics cards won't be easy to buy this year.

Is RTX 50-series production shifting?

(Image credit: Future)

We've heard that Nvidia's plan is to prioritize GPUs with lower video memory, which means that since the RTX 5060 Ti comes in bother 8GB and 16GB models, Team Green will push production of the 8GB version. One source speaking to MLID appears to back this up.

"I [Distributor] just got done with multiple phone calls with AIBs — it sounds like the 5060 is done for the next 6 months. Everyone says that Nvidia vastly overbooked Al sales, and therefore needs to pause almost all RTX 50 Series Production until at least Q3 2026. The 5090, 5070 Ti, 5060 Ti 16GB, and 5060 will soon be unobtanium... and the 5080, 5070, and 5060 Ti 8GB will trickle in with incredibly low volume."

So, while some GPUs, which include many of the best graphics cards you can get, will soon be near-impossible to grab, the rest will still be in production, but still at a "low volume." Or, at least, until Q3 of 2026 (around July to September). However, according to an unknown major retailer, there's a different idea.

"On Monday we [Major Retailer] were told that we should soon expect 5060 supply to get worse until at least Q4. In fact, we were specifically told that the 5050 and 5060 Ti 8GB are going to be most of the volume through summer. Yes, there will be a trickle of 5080s and 5070 supply... but not enough. Nvidia will 're-examine' in Q4."

Yes, there will be a trickle of 5080s and 5070 supply... but not enough. Nvidia will 're-examine' in Q4.
Anonymous retailer via Moore's Law Is Dead

With this in mind, it's looking like the RTX 5050 and RTX 5060 Ti with 8GB will be most seen on shelves, while Nvidia will rethink its production plans towards the end of 2026. In any case, it's another reason to believe that we'll soon see the RTX 5060 on its way out. The last source also supports this claim.

"For the foreseeable future, Nvidia's effectively only selling us [AIB] 5080, 5070, 5060 Ti 8GB, and 5050 kits. As for "why" Nvidia's effectively pausing 5060 production - it all comes down to cost & RAM. The 5050 uses GDDR6, and the 5060 Ti uses the same amount of GDDR7 as a 5060 — so they are preferable. Oh, and get ready for a ~30% baseline price increase to GPUs imminently."

Not only can we expect a cut in production on Nvidia's most popular GPUs, but now, it appears we may also see a 30% price increase. Of course, this isn't entirely surprising, considering Intel also expects costs to rise after 'about 9 to 12 months' of stock on laptops runs out, but it's still something to anticipate — if these comments ring true.

Nvidia's official statement brings this into question

(Image credit: Future)

While these leaks and rumors continue to come out, Nvidia has stuck to its comment confirming all RTX 50-series GPUs will continue to ship, barring the challenges with memory supply.

Responding to Tom's Guide, Nvidia states: "Demand for GeForce RTX GPUs is strong, and memory supply is constrained. We continue to ship all GeForce SKUs and are working closely with our suppliers to maximize memory availability."

Despite many rumors swirling around RTX 50-series GPUs being in troubled waters this year, Team Green clearly states there's no trouble when it comes to shipping its graphics cards. If anything, that's good news to all looking to grab a GPU this year, and hopefully, this plays out.

Is a GPU drought coming?

(Image credit: Future / Tom's Guide)

It's still uncertain just how Nvidia's GPU strategy will play out, but with even more sources pointing to a limited supply of RTX 50-series graphics cards, it's looking like GPUs will become a precious commodity in the near future.

Interestingly, Tom's Guide got to chat with CEO of U.K. PC supplier Scan, Elan Raja, who said the "RTX 5070 GPU is the sweet spot" when it comes to what PC gamers are actually buying. Fortunately, this is one of the graphics cards that apparently won't be in such short supply (for now), and that's due to its offering 12GB of VRAM, as opposed to its sibling, the RTX 5070 Ti, delivering 16GB of video memory.

In any case, if MLID's sources are predicting the near future, expect many of Nvidia's GPUs to be in incredibly short supply, with some being near impossible to find. So, would you grab an RTX 50-series GPU now or wait for supply to return after the RAM crisis passes? Let us know your thoughts!

‘RTX 5070 is the sweet spot’: Scan CEO reveals what Nvidia GPU PC gamers actually want

Jason England — Tue, 20 Jan 2026 10:00:00 +0000

As you may have seen, the question of whether you can actually buy an Nvidia RTX 50-series GPU has been quite the topic of conversation.

From uncertainty spread by Asus to Nvidia correcting the record and leakers disputing said corrections, it’s been an interesting few days that’s been capped off with Gigabyte’s CEO sharing more about Team Green’s possible plans to focus on increasing “revenue per gigabyte” on each GPU sold.

In short, Nvidia may focus its video memory capacity on the higher-end GPUs. And after talking to Scan UK CEO Elan Raja, I think this decision may be validated after hearing the “sweet spot” is not the RTX 5060 Ti with 16GB of VRAM, as I assumed, it’s actually the RTX 5070.

What is Scan UK?

To my fellow Brits, chances are you won’t need me explaining what Scan UK is. But for context (and to all my U.S. readers), think of Scan as the British equivalent of a Micro Center, or as Elan says, “Argos on Steroids.”

And being honest? It’s one of the coolest places I’ve been to in the U.K. While their bread and butter is gaming PCs — building custom computers to the specs you want (even letting you tune into a live stream of them building it if you want to watch and learn) — the business itself touches much more than that.

@tomsguide
♬ Spooky, quiet, scary atmosphere piano songs - Skittlegirl Sound

These include prosumer-level systems and hardware, setting up cloud compute, AI systems (including Nvidia’s mini supercomputer, the DGX Spark), and even robotics. During my time there, I got to see everything, and I felt like I was in Willy Wonka’s computing factory — building a computer, testing robots and getting to talk to Elan.

(Image credit: Future)

Expect much more coverage from my time with Scan. A lot of fascinating discoveries, including how sim games could train the robotics of the future, how the RAM crisis may be more cyclical than you think, and how (in Raja’s eyes) it will never kill PC building.

“Doesn’t break the bank”

(Image credit: Future)

So in my sit-down interview with Elan, we covered many topics — all of which I will be covering in a few more articles coming up. But let’s start with this question: of course retailers like Scan focus on the higher-end, but what is the sweet spot most folks are buying?

“Our customers specifically are focused at the very high end,” Raja commented. “But there's always technology for every budget.” Of course, not everybody needs the bleeding edge (and the astronomical cost that comes with it).

(Image credit: Future)

“Entry level now is considered an RTX 5060, all the way up to the extreme 32GB RTX 5090,” Raja added. “Anywhere in between, depending on what you use, is a sweet spot.” That’s a big range from $300 all the way up to $2,000 (and that’s only MSRP), so I pushed him a little to narrow it down.

“If we were talking about the general market, I'd probably say a 5070 is a decent sweet spot,” Raja confirmed. “You know that it offers great performance per buck, but at the same time, it doesn't break the bank. It's your definition of a sweet spot.”

(Image credit: Future)

In leaks surrounding the RTX 50-series supply issues (and the decisions Nvidia is reportedly making), a call allegedly hasn't been made on the RTX 5070 and its 12GB of VRAM.

This comment was made before all of this came out last week (I interviewed Elan last month), but with new context, it's interesting to see how this declaration of a sweet spot could tie to what the GPU market could look like this year.

I wasn’t a fan of RTX 5070, but does DLSS 4.5 change that?

(Image credit: Future)

It’s an interesting take, because when I reviewed the RTX 5070, I called it the “awkward middle child” of the bunch.

Granted, things have gotten a little better now, as you can grab the GPU for just a touch above MSRP ($50 more than the $549 asking price), rather than when it had ballooned to around $750 at the time of reviewing it. But it’s in between a rock of the RTX 5060 Ti 16GB and a hard place of the RTX 5070 Ti.

That being said, though, DLSS 4.5 could be the difference-maker here. As Nvidia CEO Jensen Huang said himself in a behind-closed-doors Q&A to Tom’s Guide, the vision of DLSS tech is to overcome the “physical limits” of chip technology. “Moore’s Law ended, and we make these chips as large as we possibly can,” Huang added.

And what I’ve noticed while testing DLSS 4.5 is that while you’ll take a frame rate penalty if you use it wrong and turn on Quality or Balanced mode, using the models for Performance and Ultra Performance mode does show some significant improvements. Namely, much improved clarity around moving objects, a much more stable image, and less jagged edges.

I’ll have to find out for myself whether this combination can overcome my initial impressions. Standby for a re-review…

Gigabyte CEO teases Nvidia’s 'revenue per gigabyte' push — mid-range RTX 50-series GPUs could be at risk

Darragh Murphy — Mon, 19 Jan 2026 15:00:00 +0000

Nvidia has been in the spotlight amid the ongoing RAM crisis, with the supply of its RTX 50-series GPUs in question despite Team Green stating it will "continue to ship" its graphics cards. Now, we may have further insight into the company's plans.

The memory shortage is already raising the price of consumer RAM, and it's set to affect tech across the board, including laptops, phones, consoles and more. That also includes GPUs and their video memory (VRAM), putting Nvidia in focus. In fact, to the extent that the RTX 5070 Ti with 16GB VRAM and others were tipped to be put into an EoL (End of Life) status.

However, the tech giant responded to Tom's Guide with the following:

"Demand for GeForce RTX GPUs is strong, and memory supply is constrained. We continue to ship all GeForce SKUs and are working closely with our suppliers to maximize memory availability."

This confirms that all RTX 50-series GPUs will be sticking around, but Nvidia will still struggle with the short supply of RAM. According to Gigabyte CEO Eddie Lin, in an interview with our sister site Tom's Hardware, the company may prioritize certain graphics cards over other popular choices — and it's all to do with "how much revenue contributes per gigabyte of memory."

A plan of GPU action

(Image credit: Future)

With RAM in short supply, GPUs with higher VRAM are set to be a premium, and that includes popular mid-tier graphics cards with 16GB video memory. While Nvidia claims these will still be in development, it's looking to shift focus to other GPUs that will gain more revenue.

"They cannot produce only high-end or low-end [products]... but they can, for example, they have 1, 2, 3, 4, 5, five segments," Lin states to Tom's Hardware. "They focus on 1, 3, and 5, and reduce the percentage on 2 and 4, because on 2 and 4, the revenue contribution per gigabyte of memory is lower. They will calculate how much revenue [each segment] contributes per gigabyte of memory."

They will calculate how much revenue [each segment] contributes per gigabyte of memory
Gigabyte CEO Eddie Lin via Tom's Hardware

For context, this means that instead of pushing the supply of all its graphics cards, Nvidia's strategy is to focus on GPUs that offer the most revenue per gigabyte.

Lin continues to explain, starting with a $300 GPU like the RTX 5060 with 8GB GDDR7 VRAM: "The memory contributes $35 per GB of revenue, whereas for a $400 8GB GPU, that product would contribute $50 per GB of memory. For a $500 [card] with 16GB of memory, that puts you at only $32 of revenue per GB, then the [contribution] is lower."

As an example, since the RTX 5060 with 8GB VRAM delivers $35 per gigabyte in revenue and the RTX 5060 Ti with 8GB offers $50, then Nvidia will allocate more production of the latter, seeing as there will be a bigger return.

With this in mind, we expect certain RTX 50-series GPUs to be in greater supply compared to others, and that's bad news for those looking for an RTX 5060 Ti with 16GB and RTX 5070 Ti with 16GB.

Shorter supply

(Image credit: Future)

From Gigabyte's comments, Nvidia's GPU strategy will see many popular GPUs be harder to find on shelves this year, even though they will still be in the works. If anything, it's likely that we'll see 8GB VRAM become the norm, with 16GB VRAM on its last legs.

According to a chart of revenue per gigabyte from Tom's Hardware, here are the numbers on what we can expect. This will paint a better picture of the GPUs that will be in focus:

	Revenue per gigabyte
RTX 5060 (8GB VRAM)	$37.38
RTX 5060 Ti (8GB VRAM)	$47.38
RTX 5060 Ti (16GB VRAM)	$26.81
RTX 5070 (12GB VRAM)	$45.75
RTX 5070 Ti (16GB VRAM)	$46.81
RTX 5080 (16GB VRAM)	$62.44
RTX 5090 (32GB VRAM)	$62.47

In the entry-level GPU category, you'll find the RTX 5060 Ti with 8GB GDDR7 VRAM offers the most revenue compared to its siblings, the RTX 5060 with 8GB and RTX 5060 Ti with 16GB. So, expect to find plenty more 8GB RTX 5060 Ti models around.

As for the mid-range, the RTX 5070 Ti with 16GB obviously uses more video memory, which will increase its price in production. The RTX 5070 with 12GB, however, delivers similar revenue, so this is likely to be prioritized over the more popular Ti model (sadly).

In the higher range, there's the RTX 5080 with 16GB and RTX 5090 with a whopping 32GB. Following the mid-tier, the RTX 5080 is looking to see more of a focus, considering it's largely reduced use of RAM.

It's unfortunate, but the RTX 5060 Ti with 16GB and RTX 5070 Ti with 16GB may be a precious purchase down the line, following this strategy. The RTX 5090 is already tough to get, with prices going well over MSRP. So, expect to see more RTX 5060 Ti with 8GB, RTX 5070 with 12GB and RTX 5080 with 16GB at retailers more often than not.

As the RAM crisis continues, we'll start to see how the market for all upcoming tech plays out, and Nvidia GPUs will certainly be in the spotlight. Who knows, we may even be renting a PC from the cloud in the future.

Nvidia says all RTX 50-series GPUs will 'continue to ship,' but stock and supply tell a different story

Jason England — Fri, 16 Jan 2026 11:38:39 +0000

Well, it’s been an interesting 24 hours for the best Nvidia graphics cards. To catch you up, the RAM price crisis has hit the computing industry hard — AI data center demand is eating up a lot of the supply, which is impacting how much you pay for your tech.

And as was reported yesterday, Asus had allegedly placed the RTX 5070 Ti into End of Life (EoL) status. Then, Nvidia commented saying that the company will “continue to ship” all RTX 50-series GPUs, and Asus walked back its statement.

This was then followed up on again by Moore’s Law Is Dead with a report that claims that Nvidia has killed the “RTX 5080, 5070, 5060 Ti to supply AI bubble! (Laptops Included).” Here’s a timeline of everything that happened, how we got here and what my advice to you is right now.

How it started

As we originally reported yesterday, Hardware Unboxed reported that Asus had effectively killed its RTX 5070 Ti, as Nvidia was realigning memory prioritization to its AI data center business.

This is reportedly part of a bigger move to prioritize more performant consumer GPUs with the same amount of memory. An example from HEKPC makes this clear: “When two different models, such as the RTX 5060 8GB and RTX 5060 Ti 8GB, share the same memory capacity, the higher-tier model (the RTX 5060 Ti) will be prioritized for supply.”

(Image credit: Future)

Not only is the RTX 5070 Ti brought into question, but also the RTX 5060 Ti with 16GB VRAM — with the connection being made to the RTX 5080 sporting the same amount of memory.

You’ve seen PC gaming come under real pressure, as AI companies snap up a significant chunk of the memory allocation. Looking further back, this started with the RTX 50-series being potentially scrapped outright. Talking to companies on the CES 2026 show floor, it seemed clear to me that the expectation was for new GPUs, but they didn’t materialize.

The follow-ups

(Image credit: Future)

This is where it gets a little messy. We reached out to Nvidia for comment, and Team Green confirmed that the company will “continue to ship all GeForce SKUs and are working closely with our suppliers to maximize memory availability."

Hardware Unboxed also got a clarifying statement from Asus walking back their original comments.

Sorry everyone but we’ve just been provided with ANOTHER clarifying statement from Asus. This one completely walks back their original statement to us“We would like to clarify recent reports regarding the ASUS GeForce RTX™ 5070 Ti and RTX™ 5060 Ti 16 GB. Certain media may…January 16, 2026

“The GeForce RTX 5070 Ti and GeForce RTX 5060 Ti 16 GB have not been discontinued or designated as end-of-life (EOL),” the statement reads. “Current fluctuations in supply for both products are primarily due to memory supply constraints, which have temporarily affected production output and restocking cycles.”

This, in Asus’ words, will result in supply appearing “limited in certain markets.” So to quote the timeline of this story directly from Hardware Unboxed’s X post:

We request RTX 5070 Ti samples from Asus (and other partners).
An Asus PR rep looks into it, comes back and says they cannot provide 5070 Tis due to supply constraints, saying their models are “end of life.”
We ask Asus to clarify whether the RTX 5070 Ti is end of life. They confirm it’s end of life.
We reach out to retailers to see whether they can purchase RTX 5070 Ti stock for their stores, to fact-check Asus’ claims. They say there is no supply.
Given we have received an on the record statement from Asus and confirmed the supply constraints with retailers, we publish a video with this information.
Nvidia says all GeForce SKUs are being shipped.
Asus reaches out to clarify that Nvidia told Asus that the RTX 5070 Ti is not end of life, but that Asus are “streamlining some models.”
We publish that statement.
Asus reaches out again to provide another statement (the third statement we’ve received from Asus), now saying the 5070 Ti is not discontinued or end of life. This directly contradicts the original statement.
We immediately request RTX 5070 Ti samples now that the cards are not discontinued or end of life. We haven’t heard back yet.

And those memory supply constraints can be felt in some of the price variations. Our friends at Tom’s Hardware have been tracking the street prices vs MSRPs, and have spotted RTX 50-series going up by as much as 32%.

It could be worse

(Image credit: Future)

Plot twist. More reports came out during this period of time, too. First, according to MEGAsizeGPU (and reported by Tom’s Hardware), Nvidia has allegedly slashed GPU supply by 15-20%, and there won’t be any new GeForce cards until 2027.

While it’s easy to connect this potential drought to the above price jump I mentioned above, this could also be to do with panic buying surrounding this evolving story. But in a world of supply and demand where the demand is outstripping the supply (and could be made worse based on this rumor), prices could continue to go up.

Then, Moore’s Law Is Dead posted a video saying that “the reports are true.” According to the video, the presenter, Tom, had spoken to his sources close to the matter and found out that it is “not just the 5060 Ti 16GB nor the 5070 Ti 16GB.”

Instead, he claims that Nvidia is “basically exiting high-end PC gaming in 2026.” Anything with more than 8GB of VRAM is “having its supply throttled,” and this isn’t just limited to desktop GPUs. A couple of sources have told Tom that “even laptops with dedicated Nvidia graphics cards” are “receiving far fewer shipments.”

One final note: he also confirms the reports of the RTX 3060 12GB coming back are true — “some retailers” have told him that “they just got a fresh shipment” of these GPUs that launched back in 2021.

Our take (and advice to you)

(Image credit: Future)

In our Q&A session with Nvidia CEO Jensen Huang, he talked about this challenge in two ways. First, he mentioned how the company is a “significant consumer” of GDDR video memory for graphics cards and has been “planning that out with all our suppliers for quite some time.”

And Paul Alcorn from Tom’s Hardware asked a question about bringing back old GPUs, to which Jensen said this is a “good idea” — highlighting that “we could bring the latest generation AI technology to the previous generation GPUs.”

But let’s get down to brass tacks. What does this actually mean for you? There’s a lot of people saying a lot of different things, but one thing is clear: RAMageddon has come for GPUs.

Dramatic wording aside, you’ve got two choices if you’re in the market:

Buy now while the prices are a little inflated.
Wait it out and hope that price comes down, but you’ll be waiting a while.

I’m not going to give a take one way or the other on the evolving situation — this is just a timeline of events. One thing is clear: AI looks set to impact what you pay for PCs in a big way.

'Demand for GeForce RTX GPUs is strong': Nvidia reportedly won't discontinue RTX 5070 Ti GPU

scott.younker@futurenet.com (Scott Younker) — Thu, 15 Jan 2026 23:38:59 +0000

Update 3:45 pm PT, Jan. 15, 2026: Nvidia responded to our queries and gave us this statement:

"Demand for GeForce RTX GPUs is strong, and memory supply is constrained. We continue to ship all GeForce SKUs and are working closely with our suppliers to maximize memory availability."

The translation was correct, and this statement is the same as what Hardwareluxxe received. So, the 5070 Ti with 12GB does not appear to be discontinued after all.

We've read reports today that Nvidia would be discontinuing the RTX 5070 Ti with 16GB of RAM due to the ongoing RAM crisis. Several other 16GB VRAM GPUs were allegedly also on the shopping block.

However, that may not be true. The initial discontinuation claim was made by the YouTube channel Hardware Unboxed. However, later in the day Nvidia reportedly gave German website Hardwareluxx a statement contradicting the Hardware Unboxed report.

"Demand for GeForce RTX GPUs is strong, and memory supply is constrained. We continue to ship all GeForce SKUs and are working closely with our suppliers to maximize memory availability," reads the (machine translated) statement.

What's going on?

(Image credit: Future)

Some have claimed Nvidia is focusing on smaller 8GB and 12GB memory variants of its high-end RTX 50 series models, while only the GeForce RTX 5080 and the GeForce RTX 5090 would feature 16GB and 32GB (respectively).

RAMageddon is squeezing every part of the tech world, from smartphones to laptops and consumer memory. It wouldn't be that surprising if Nvidia had to discontinue a few GPUs to meet demand and guarantee access to memory.

Indeed, Nvidia seems to be trying to find new ways of satisfying customers who can't afford to pay through the nose for high-end GPUs at inflated prices.The company may even bring the RTX 3060 back from the dead, an idea that CEO Jensen Huang didn't dismiss out of hand during a recent interview.

Tom's Guide reached out to the North American arm of Nvidia to confirm the veracity of the Hardwareluxx statement. We will update this article if and when the company responds.

'Things are looking pretty bleak across the graphics card market': Nvidia reportedly set to discontinue the RTX 5070 Ti GPU

tony.polanco@futurenet.com (Tony Polanco) — Thu, 15 Jan 2026 16:17:44 +0000

In a new video from Hardware Unboxed (via PC Gamer), the YouTube channel claims that the Nvidia GeForce RTX 5070 Ti will be discontinued.

Asus confirmed this report, telling Hardware Unboxed that it has placed the graphics card into its EoL (End of Life) product list. The YouTube channel also claims that, according to its investigation, some retailers in Australia are unable to secure the RTX 5070 Ti from manufacturers. As you might have guessed, this is due to the ongoing RAM crisis.

While this might seem arbitrary, there is a method to the madness. According to a report from HEKPC (via Wccftech), Nvidia is realigning production of its entire gaming stack and will discontinue several GPUs with 16GB of VRAM.

Editor's note: Nvidia reportedly gave German website Hardwareluxx a statement, saying: "Demand for GeForce RTX GPUs is strong, and memory supply is constrained. We continue to ship all GeForce SKUs and are working closely with our suppliers to maximize memory availability."This statement, if accurate, would contradict what Asus reportedly said about the RTX 5070 Ti.

So long, 16GB

(Image credit: Future)

“If a model like the RTX 5060 Ti has both 8GB and 16GB versions, Nvidia will focus on supplying the 8GB version,” says the HEKPC report. “When two different models, such as the RTX 5060 8GB and RTX 5060 Ti 8GB, share the same memory capacity, the higher-tier model (the RTX 5060 Ti) will be prioritized for supply.”

So if the RTX 5060 Ti, RTX 5070 Ti and RTX 5080 are all 16GB models, then Nvidia will prioritize the highest-tier model. In this case, the RTX 5080 will get priority.

This report claims that Nvidia hasn’t decided on the fate of the RTX 5070 12GB GPU, but it’s possible that its memory allocation could be allocated to the RTX 5080. As for the RTX 5090 and RTX 5090 D v2, these graphics cards will reportedly not be affected. That said, they might be harder to find since their supply won’t be as plentiful.

In short, Nvidia will reportedly stop manufacturing the RTX 5060 8GB, RTX 5060 Ti 16GB and RTX 5070 Ti 16GB. The company will focus on the 8GB versions of those cards, along with higher-tier GPUs than the RTX 5070 Ti.

It’s a shame that the RTX 5070 Ti is now a victim of RAMageddon. As we said in our Nvidia GeForce RTX 5070 Ti review, it was the Goldilocks GPU in Nvidia’s entire 50-series, with just the right amount of blistering performance, versatility across games at 1440p and 1080p, and a true showcase of DLSS 4 AI trickery with buttery 4K smoothness.

Now, thanks to the memory crisis, folks will either have to fork over an obscene amount of cash for higher-end cards or make do with their current GPU. Since I don’t have $5,000 for an RTX 5090, I’m firmly in the latter camp.

We’ve reached out to Asus and Nvidia for comment, so we’ll update this article if and when we hear back from these companies. Stay tuned!

DLSS 4.5 is now live — I tested Nvidia’s upscaler to see which model you should actually use

Jason England — Wed, 14 Jan 2026 14:07:53 +0000

DLSS 4.5 is out of beta and available to use by everyone. Make sure you update your Nvidia app and GPU drivers and it’s all yours across all the games that already support DLSS 4! Just open the app to update or download it manually here.

But after testing it over the past few days since going eyes-on for the first time at CES 2026, I’ll say it can get confusing to figure it all out. So with that in mind, I want to share the results of all my testing and answer two key questions:

When should I use DLSS 4.5?
What is the difference between Model L and Model M?

But first, an explainer.

What is DLSS?

(Image credit: Nvidia)

Deep Learning Super Sampling is one of Nvidia’s showcase pieces of AI trickery that extracts better performance from your games.

To do so, it will render a game on the GPU at a lower resolution and use a neural network trained on the game to upscale to a higher resolution.

While from the get-go, you can use an automatic mode in most games, which chooses what’s best based on the specs of your PC, there are four modes to pick from:

Quality: This renders the game closest to native resolution (normally around 66-70% of resolution).
Balanced: Finding the mix between getting the best possible textures and frame rates, this will render the game at 58% of your targeted resolution.
Performance: One step down to maximize frame rate, this will often go for 50% of the resolution.
Ultra performance: And if you want to go all out on frames, ultra performance will target 33.3% of the total resolution being rendered by the GPU.

(Image credit: Future)

These can come with a penalty, and these are usually ghosting (faint outlines around moving objects as DLSS fights to keep up with what’s happening on-screen), shimmering (brightly lit objects being a little bit blown out), overdone AI sharpness at times and rendered detail in the background not being clear until you get closer to it.

In terms of model-naming:

Model K = DLSS 4
Model L = DLSS 4.5 targeted at ultra performance mode
Model M = DLSS 4.5 targeted at performance mode and above

When should I use DLSS 4.5?

(Image credit: Future)

Put simply: use it if you’re opting for performance modes. Nvidia has been clear in saying that’s what this model is targeting, and when put through testing at higher quality modes like balanced or quality, you’re not getting any noticeable uplift in image quality but a lower frame rate. By the way, all these tests are done with frame generation turned off.

DLSS 4 on RTX 5070 laptopFuture

DLSS 4.5 on RTX 5070 laptopFuture

DLSS 4.5 test across 3 systems
System	Black Myth Wukong (DLSS 4 Balanced)	Black Myth Wukong (DLSS 4.5 Model M)	Cyberpunk 2077 (DLSS 4 Balanced)	Cyberpunk 2077 (DLSS 4.5 Model M)
RTX 5080 PC	53 FPS (4K Cinematic)	53 FPS (4K Cinematic)	84.53 FPS (4K Ray-tracing overdrive)	68.96 FPS (4K Ray-tracing overdrive)
RTX 5070 Laptop	52 FPS (1200p High)	44 FPS (1200p High)	77.84 FPS (1080p Ray-tracing low)	70.17 FPS (1080p Ray-tracing low)
RTX 5060 Ti (16GB) PC	47 FPS (1440p Cinematic)	45 FPS (Cinematic)	n/a	n/a

Meanwhile, if you do move to performance, you only get taxed a little, but the image is much more stable with reduced ghosting around objects, better distant detail and less shimmering.

Future

DLSS 4.5 performance mode comparison (RTX 5080 PC)
Cyberpunk 2077 settings	DLSS 4	DLSS 4.5
4K Ray-tracing overdrive DLSS performance mode	86.61 FPS	84.53 FPS
4K Ray-tracing overdrive DLSS ultra-performance mode	126.22 FPS	125.75 FPS

Model L vs Model M

Future

In my testing across several games (and after speaking to Nvidia), here’s how to choose between the models:

Model M: This is the model that’s been trained more for clarity, so this one is recommended for performance mode — driving for more detailed graphics.
Model L: This is better for ultra performance. You still get taxed in image quality, but nowhere near as much as you see in DLSS 4.

You can see as much in Cyberpunk 2077. The differences are subtle, but in the sharpness of the lamps and the heatwaves coming up from the food cart and distorting the bottles, you can see a little more ghosting and jagged edges in Model L over Model M.

Future

My recommendation to you is that if you are focused on ultimate fidelity, stick with Model K (DLSS 4). Model L is mightily impressive, but DLSS 4.5 Model M is the better all-rounder that Team Green has confirmed will improve over time into quality and balanced modes. But that transition is still in progress to my eyes.

My next steps

(Image credit: Future)

So where will my testing take me next? Well, first off, I’ve been hearing that the more advanced acceleration parts of this new transformer model have a bigger penalty on GPUs older than the RTX 50 series. I’m in the process of calling in some older cards and gaming laptops to put this to the test.

But let’s get weird, too. I’ve tested the features you’ll get out of the box, but if you know anything about me, I’m a bit rogue when it comes to testing — pushing the theory behind new tech to its extreme.

In the behind-closed-doors Q&A, Nvidia CEO Jensen Huang talked about how “more and more computation” could be done on “fewer and fewer pixels.” Put simply, DLSS on steroids is where Huang sees all of this heading. I’m setting up what’s needed (DLSS Tweaks) to put this theory to the test — seeing how it looks to render a game at 360p and upscale to 4K.

I’ve seen some of my friends do this, and the results are…well, they’re a mess if you want to play using the settings. But when you think about what’s happening (a neural network analyzing the picture and upscaling everything), it’s a mindblowing preview of the progress being made towards this vision.

Watch this space. I’ll be back with more.

After missing CES, don't expect the Nvidia RTX 50 Super GPU series anytime soon

scott.younker@futurenet.com (Scott Younker) — Tue, 13 Jan 2026 21:39:33 +0000

Nvidia had a massive presence at CES 2026, but if you were hoping for good, or any, news about the future of gaming GPUs, it was sorely lacking. As has become the case for most of the tech world, Team Green's focus was on AI.

Which makes a new leak from Tech PowerUp even more poignant, as it reported Nvidia has allegedly told vendor partners that the GeForce RTX 50-series SUPER GPUs were being delayed indefinitely.

Missing from CES

(Image credit: Future)

Prior to CES, it was rumored Nvidia would release the mid-cycle refresh at least by CES, if not in December. A CES 2026 release would fit since Nvidia typically launches the SUPER models 12 months after the release of the base models. The original RTX 50-series debuted at CES 2025.

However, a late September leak claimed that the cards were delayed but would at least be announced at CES 2026. It was speculated that they would still come out in the first three months of the new year.

The closest thing we got at the tech show was an extreme RTX 5090 GPU that won't even fit in most computers. That was a heavily modded version built by MSI that only 1,300 people will even get the opportunity to purchase.

From the jacket's zipper

(Image credit: Future)

Even in a closed-door interview with CEO Jensen Huang, he remained mum on the future of Nvidia's gaming GPUs. At one point, Huang commented that Nvidia could bring the current generation of AI technology to older CUDA cores.

“That’ll require a fair amount of engineering, but it’s also within the realm of possibility. I’ll go back and take a look at this," Huang said in response to a question about Nvidia rereleasing the RTX 3060 to combat the memory shortage.

It's possible that Nvidia is trying to transition its GPUs from the current Blackwell architecture to the new Rubin GPU architecture revealed at CES. However, we may not see that until the RTX 60 series.

AI to blame

(Image credit: Amazon)

According to the TechPowerUp leak, there are three reasons why Nvidia is delaying the SUPERs. The most obvious and easiest to blame is AI and the insatiable need for GPUs and RAM to power the data centers running various models. RAM prices are already soaring and may not improve until 2028.

Many companies like Nvidia, Micron and Samsung are leaning into the AI side of things while profits are skyrocketing for them. Which also means that VRAM prices for GPUs are also going up, a real problem for modern graphics cards that use GDDR7 VRAM.

Finally, the rumor suggests that because AMD didn't announce any new consumer GPUs for 2026, the RTX 50 series doesn't need an update since it will remain competitive with whatever is already on the market. Right now, the top AMD RDNA 4 GPU is the Radeon RX 9070 XT, which isn't getting an update.

Tom's Guide reached out to Nvidia for comment on the alleged delay and will update this article if the company responds.

'Basically a photograph interacting with you at 500 frames per second': Nvidia CEO Jensen Huang shares his vision of the future of PC gaming, and how the company plans to tackle GPU pricing crisis

Jason England — Sat, 10 Jan 2026 06:10:00 +0000

If you were to look at what Nvidia announced at CES 2026, I wouldn’t blame you for thinking Team Green may not be a PC gaming company anymore. A huge portion of its $5 trillion valuation is driven by the AI gold rush, and on-stage, CEO Jensen Huang talked entirely about new data center tech and robotics.

Tom's Guide at CES

(Image credit: Future)

We've picked our best of CES 2026 awards after going hands-on with the biggest gadget announcements. For more, be sure to follow Tom's Guide on TikTok for the coolest videos from the show.

But behind closed doors at a Q&A with Huang, we were really able to get into gaming with him and talk about the future of GPU tech, how AI will continue to push the gaming envelope than what we see now with the likes of DLSS 4.5 and how it’s the key to exceeding physical limitations of traditional chip technology. AI could also be used as a workaround to the RAM pricing crisis that is starting to hit the best GPUs.

Oh, and of course, we got answers to the most important meme question in the computing community — what is it with all the leather jackets?

AI is the key to generational leaps

@tomsguide
♬ original sound - Tom’s Guide

So let’s start at the beginning. After a whole lot of questions about the new Vera Rubin data center tech, we were finally able to get a question about gaming, which made Jensen’s eyes light up.

"GeForce brought CUDA to the world, which brought AI to the world, and then after that, we used AI to bring RTX to gamers and DLSS to gamers,” Huang commented. “And so you know, without GeForce, there would be no AI today. Without AI, there would be no DLSS today. It's harmonious."

(Image credit: Future)

Now, I know some of you reading this are looking at the use of AI here and rolling your eyes. Fake frames is something I’m commonly asked about in reader emails, and in my time testing DLSS 4.5, all I can say is that steps are being taken to generate frames only when you need them. It’s a positive shift from hitting it with a 4x blunt instrument!

But is AI being used to overcome physical chip limitations? Huang says yes, and we may have seen the biggest possible GPU for raw graphics rendering come with the RTX 5090.

“Moore's Law ended, and we make these chips as large as we possibly can. If you look at a 5090, holy cow, it's giant. You're really at the physical limits, and that's the reason why we invented DLSS in the first place.”

(Image credit: Future)

And much like the Blackwell GPU architecture made the move from data center cards to the RTX 40 and 50 series, it seems obvious that you will see the Rubin architecture jump over for the RTX 60 series. So pay attention to what these cards can do.

Neural rendering is the future

(Image credit: Nvidia)

But by steering this PC gaming car into the skid of AI, there’s much more that can be done — you’re seeing the signs of this happening already with some of the RTX neural systems, Nvidia ACE AI NPCs and more. But as Jensen himself says, the future is “neural rendering.”

DLSS is already kind of an example of this in some sort of way: taking some pixels rendered by the GPU and predicting the rest based on a trained model, and DLSS 4.5 in my eyes-on is quite the breakthrough in extracting more performance you wouldn’t normally see. The next step will be leaning more on the AI side of things.

(Image credit: Future)

And in Team Green’s labs, Huang says the company is working on things that are just “utterly shocking and incredible.” To put a more specific point on it, he talked about extreme photo realism: “basically a photograph interacting with you at 500 frames per second.”

These will lead to a growing “fusion between rendering and generative AI” in Jensen's eyes — taking the tradition generative models we know now and having them run at the pace of a video game to create infinite worlds.

AI will touch every part of a game’s DNA

(Image credit: Nvidia)

Now, Nvidia has been showing its AI gaming tech isn’t just limited to making titles look prettier and run smoother. We’ve tested Nvidia’s AI NPCs (named ACE) in the past, and while it’s definitely a fun demo feature at the moment, Huang predicts that AI will move beyond just graphics to control the characters and physics within the game world itself.

“You should also expect that future video games are essentially AI characters within them. Every character will have their own AI, and will be animated robotically using AI,” Jensen predicted.

(Image credit: Future)

And you can see this coming hand-in-hand with some of the AI work Nvidia’s doing in the enterprise world — such as Cosmos driving robotics and full self-driving to breed interoperated human behavior.

Relying on AI processing isn’t just the shortcut to better performance over raw rasterization of graphics, it could be the key to unlocking the next generation of entire titles.

"All the characters as I mentioned earlier will be AI-based, all the animation will be AI-based and, and all of the fluid dynamics won't be computational based, but AI-emulated physics,” Huang said. “Everything is just going to look so incredibly real, and so immersive."

Looking back to survive RAMageddon

(Image credit: Future)

If there’s one topic of conversation that dominated the halls of CES this year, it’s the RAM price explosion. So who better to ask about the great AI gobble-up of DRAM chips than the guy at the center of using them. When asked about rising costs and potential shortages for gamers, Huang went into Nvidia’s role as a major memory purchaser.

“We also purchase GDDR for our graphics cards, and we’ve been a very significant consumer for a very long time, and so we’ve been planning that out with all of our suppliers for quite some time.”

To push further, my friend Paul Alcorn from Tom’s Hardware managed to get a question in, and connected it to the rumor that Nvidia may be bringing the RTX 3060 back from the dead amidst the memory shortage.

(Image credit: Future)

Jensen had an interesting answer to this, saying it’s a “good idea,” while also expanding on what could be done. “We could bring the latest generation AI technology to the previous generation GPUs,” Huang commented. “That’ll require a fair amount of engineering, but it’s also within the realm of possibility. I’ll go back and take a look at this.”

This would be an interesting concept and a huge bet on the capabilities of Nvidia’s AI gaming tech. Rendering on older CUDA cores, while doing the guesswork on the latest Tensor cores could produce some interesting results — both in affordability and performance.

Deciphering the leather jacket lore

(Image credit: Future)

And of course, the most critical question — the main reason why you’re here. What gives with all the leather jackets? All of us who report on Nvidia’s work have our own theories on the jackets picked and worn for each specific event.

Mine is a simple one: the shinier the jacket, the bigger the announcement. So I was so happy that someone brought up Jensen Huang’s signature attire, and asked if he has a road map for his style in the future.

“Somebody else’s road map, I just wear it,” Huang joked. Of course there’s going to be a stylist behind the scenes, and I have so many questions for them — hopefully to validate my theory on the leather jacket lore.

I just saw the most extreme RTX 5090 GPU at CES — and it definitely won’t fit inside my PC

tony.polanco@futurenet.com (Tony Polanco) — Thu, 08 Jan 2026 21:36:40 +0000

The MSI GeForce RTX 5090 32GB Lightning Z might be one of the most extreme GPUs ever released. Unveiled during CES 2026, this graphics card features several “world firsts,” including an 8-inch GPU display, power output beyond 1000W for extreme overclocking, and a full copper, full-cover cold plate to maximize cooling. This card is so power hungry that MSI recommends a 1600W+ power supply. That, my friends, is absolutely insane!

I wasn’t able to independently benchmark the Lighting Z when I met with MSI during CES, but MSI’s internal benchmarks claim some bold numbers. That includes a score of 513,272 on the Geekbench 6 CPU benchmark and 139,907 on the Wildlife Extreme 3D Mark graphics benchmark. Truly staggering stuff!

GeForce RTX 5090 32GB Lightning Z: Specs

	GeForce RTX 5090 32GB Lighting Z
Interface	PCI Express Gen 5x 16
Core clocks	Extreme Performance: 2775 MHz (MSI Center) Boost: 2730 MHz
Cores	21760 Units
Memory	32GB GDDR7
Memory bus	512bit
Output	USB Type-C x 1 (Dedicated for Lightning display panel only), DisplayPort x 3 (2.1b), HDMI x 1 (As specified in HDMI 2.1b: up to 4K 480Hz or 8K 120Hz with DSC, Gaming VRR, HDR)
HDCP Support	Y
Power consumption	OC:800 W EXtreme:1000W
Power connectors	16-pin x 2 (1-to-3 dongle not supported.)
Recommended PSU	1600W
Dimensions	Card: 10.2 x 5.9 x 2.4 inches Radiator: 15.1 x 4.7 x 2.2 inches
Weight	TBD
Maximum displays	4
G-Sync	Y
Digital Maximum resolution	7680 x 4320

On the spec sheet, the boost clock is listed at 2330Hz. Then there’s the “Extreme Performance” profile that can jump to 2775Hz, but you’ll need to enable that through the MSI Center app. There are two settings to choose from, with one being an overclock mode at 800W and Extreme Performance rated at over 1000W. My electricity bill is already screaming at those numbers.

The GPU is huge since it’s an RTX 5090, measuring 10.2 x 5.9 x 2.4 inches. The weight wasn’t disclosed, but it felt heavy when I held it. Connection-wise, the Lighting Z features two 16-pin power connectors. For output, it has three DisplayPorts, an HDMI 2.1 port, and a single USB-C port.

(Image credit: Tom's Guide)

I saw the Lighting Z inside the MEG Maestro 900R, but it’s not exclusively for that PC case. But as you might have guessed, you’re going to need a huge PC case of your own to house this beastly graphics card.

Speaking about what's in the MEG Maestro 900R case I saw at CES, it had the MEG X870E Unify-X Max motherboard, which features a slew of overclocking features, a clean, all-black design, and lets you pack 128GB (2x64GB) of RAM.

Lastly, the rig also featured the MPG Ai1600TS PCIE5, which can deliver the 1600W the Lightning Z requires. In addition to being powerful, this PSU also has GPU Safeguard+, which exists to protect your GPU from unexpected threats. It also has stable power delivery, smart thermal regulation, and is built to be durable.

Outlook

MSI didn’t provide pricing, but it said the GeForce RTX 5090 32GB Lighting Z will release sometime in February. Considering how it’s limited to 1300 units, it’s probably going to cost a pretty penny—if you can snag a unit. But even if you can’t, I’m sure you can appreciate that such a powerful GPU even exists.

Nvidia RTX 60-series GPUs: Rumors, leaks and everything we know so far

Lloyd Coombes — Sat, 03 Jan 2026 09:30:00 +0000

Nvidia is the most valuable company in the world thanks to the AI boom (or bubble), but there’s still the small matter of graphics cards to attend to.

The company’s 50-series GPUs debuted earlier this year, but arrived with mixed responses. There was the price to consider, the fact that they were scalped to high heaven, and the company betting big on its AI technologies to add generated interstitial frames instead of making the cards more powerful.

Still, with rumors floating around that concern the 60-series, and the expectation of RTX 50 SUPER series launching soon, it may not be long until we see new cards from Nvidia. Here’s all we know so far.

When can we expect the RTX 60-series?

Our first leak, in December 2025, suggests that NVIDIA’s 60-series GPUs are slated for a “Q1 2027 launch announcement,” likely to be at CES next year.

There’s also a good chance we end up with a full fleet of models, from 6050 up to 6090, with suggestions of big power gains at the top end.

Rumored specs

(Image credit: Future)

There are some suggestions about what the new NVIDIA 60-series GPUs will include, and also what they might not.

The company’s CEO, Jensen Huang, has been touting the next-generation Rubin GPU architecture at every Nvidia GTC, so there’s a strong likelihood it will debut with the 60-series, as it’s also slated for data centers.

The Blackwell architecture began in data centers, too, and then made the move to consumer cards with the 50-series, so expect a similar play here.

NVIDIA reckons Rubin is twice as fast as Blackwell in AI workloads, but when it comes to gaming, rumors have suggested a 40% increase from the RTX 5090 to the RTX 6090 - a sizeable climb, although expect more power usage.

As for what might not be there, expect NVIDIA to avoid launching cards with 8GB of VRAM. Customers have been voting with their wallets and pushing back on the idea, suggesting we could get more on the GPU front this time around. Sadly, though, there’s a spanner in the works…

The RAMifications

(Image credit: Future / Tom's Guide)

NVIDIA, as with just about every tech giant in the world, is likely to be brought to its knees by the RAM crisis.

If you missed it, RAM is in such high demand for data centers that it’s being priced out of reach for consumers. The AI age has essentially hamstrung the computing industry until prices come down or RAM is more available, and neither looks likely anytime soon.

That could mean one of two things: NVIDIA will be forced to keep relatively low VRAM counts in its GPUs and compromise on features and performance accordingly, or the new cards will skyrocket in price - particularly if they’re scalped.

Neither sounds ideal, but that’s the world we’re living in right now it seems.

Will we get DLSS 5?

NVIDIA has been all-in on AI technologies for years, and as I type this, I can see that the NVIDIA app is now packed with an AI-powered Home Studio, the option to build a chatbot, an experimental AI assistant, and the option to stream games from the cloud, among other things.

Still, DLSS has perhaps been the feature getting the biggest push. It’s no ‘silver bullet’ for unoptimized games, but can have a transformative effect on supported titles, ramping up frame rates at the cost of relatively minor latency and some shimmering edges.

It’s something AMD is still working to catch up on, and NVIDIA will no doubt want to press its advantage here, so expect DLSS 5.0 to be a very likely inclusion in anything 60-series.

For now, though, we don’t have a lot to go on for the tech. We’ll update this page if and when we hear more.

What about RTX 50 SUPER?

(Image credit: Future)

There’s a missing piece to this puzzle, though - NVIDIA’s Super lineup. The company tends to release more powerful versions of its hardware, but it’s been quiet on the Super front.

The RTX 50-series hasn’t had any mid-cycle refresh announcements just yet, prompting suggestions it may have been cancelled, given the current RAM issues we’ve discussed earlier on this page.

That seems to suggest the NVIDIA 60-series will be our next wave, skipping the Super refresh entirely - but we’ll update this page if we hear more. If you’re holding out, it’s not ideal though.

2026 will be the year of cloud gaming — and we may not even have a choice

dave.meikleham@futurenet.com (Dave Meikleham) — Sat, 03 Jan 2026 07:30:00 +0000

Just like “Terminator 2” predicted way back in ye olden times of 1991, Judgement Day is upon us. Artificial Intelligence isn’t the future anymore; it’s the here and now. We’re not quite seeing cities being engulfed in Skynet nuclear flames quite yet, but AI is increasingly hitting fans of the best PC games bank balances hard. No wonder cloud gaming looks like the most sensible option for budget-conscious gamers going forward.

Make no mistake: RAMageddon is absolutely a thing (the individual who coined that phrase deserves all the flowers). The main takeaway for PC gamers looking to buy new sticks of memory (especially DDR5)? Thanks to the combination of a DRAM oligopoly controlling 95% of memory production and colossal demand from AI and data center companies, the cost of RAM has skyrocketed a bewildering 240% on Amazon based on our recent research.

Today’s big budget games normally require at least 16GB of RAM if you hope to hit 60 FPS at high resolutions. Now obviously GPUs and CPUs remain far more important components when it comes to eking out the best performance from your gaming PC. Yet RAM also plays a key part of the equation; especially in memory-intensive titles like Cyberpunk 2077 or Microsoft Flight Simulator 2024.

The cost of RAM has skyrocketed a bewildering 240% on Amazon based on our recent research"

That’s why the best cloud gaming services are becoming more and more enticing for folks who don’t want to fork out absorbent amounts for a couple of sticks of RAM or an RTX 50-series Nvidia graphics card.

For years the notion of gaming through the cloud on a smart device was looked down upon by the tiny minority of the market lucky enough to own one of the best gaming PCs. Just look back to the disastrous, short lived tenure of Google Stadia. The subpar image quality and hideous latency of Stadia’s cursed legacy still informs snobbery against modern game streaming.

On cloud nine (gaming)

(Image credit: Future)

Up until a couple of years ago, I’d have been right there with fellow PC elitists turning my nose up at playing my favorite Steam games through the cloud, rather than cutting edge hardware. Just like Bob Dylan once warbled, though, “The Times They Are A-Changin”. And how.

When I mentioned the best game streaming services a few paragraphs back, I really meant one in particular. That would be Nvidia GeForce Now. Though we might currently rank Xbox Game Pass Ultimate as the cream of the crop for cloud gaming for most people, as a hardcore PC gamer, I couldn’t disagree more with that choice… short of grinding my teeth to enamel oblivion.

Now that the Ultimate tier of Microsoft’s cloud service has jumped from $19.99 to $29.99 per month, its value proposition is nowhere near as appealing as it once was. Especially when the far superior GeForce Now Ultimate sub costs $19.99/£19.99 every 30 days.

It’s commendable that the Xbox cloud service runs on such a wide variety of devices; from Amazon Fire TV Sticks, tablets, MacBooks to the best phones. Compare it to the highest quality version of GeForce now though, and the subsequent fight is about as competitive as pitting a Pomeranian against a polar bear.

When our Managing Editor of Computing Jason England first tested the RTX 5080 tier that recently became available to GeForce Now Ultimate subscribers, he wrote that the high-end cloud gaming service “made my $750 PS5 Pro feel old”. Short of myself and Jason’s consciousnesses being melded by a “Pluribus” style hivemind, I couldn’t agree more with that statement.

Steam vs stream

Whether you're playing the GeForce Now app on a TV or laptop, it's pleasingly snappy to navigate. (Image credit: Tom's Guide)

Out of pure, massively geeky interest, I’ve done a lot of testing comparing my mobile RTX 5080-powered Asus ROG Zephyrus G14 (2025) against the 5080 tier of GeForce Now Ultimate running on the same laptop. Often I’ve found the in-game differences are comparatively razor-thin.

This has brought up a cloud vs native debate that’s made me seriously requestion buying what we consider to be 2025’s best gaming laptop.

Doom: The Dark Ages is the shining poster child for the merits of cloud gaming I’ve seen so far. Setting my native Steam version to ultra settings, DLSS quality mode, with path tracing enabled at 2,560 x 1,600 resolution, bolstered by Nvidia’s X3 frame generation, I could easily hit well over 120 FPS.

On GeForce Now 5080 levels on the same settings via the cloud on my 1GB fiber optic connection? The results were virtually identical: the streaming and native versions of the Doomslayer’s mediaeval killing quest comfortably weighing in at less than 10ms of lag.

Once you get it working, GeForce Now on Steam Deck is incredible. (Image credit: Tom's Guide)

Holy smokes is the RTX 5080-level of GeForce Now incredible on Steam Deck OLED, too. Fully taking advantage of Valve’s palm-friendly gadget, it’s about the most eye-alluring advert for what we consider to be the best handheld gaming console I can think of.

Considering the OG Steam Deck is limited to 60Hz, being able to play modern triple-A games at console-shaming frame rates at 90 FPS on Valve’s pin-sharp screen is an utter delight. The previously mentioned Doom: The Dark Ages easily runs at the Deck OLED’s 90Hz refresh rate with top-tier graphical settings and ray tracing enabled.

The Doomslayer provides gobsmacking, demon-dicing delights in your hands should you have strong broadband/fiber optic. FYI, Valve recommends an internet connection of between 50-100+ Mbps to get the most out of GeForce Now’s RTX 5080 tier.

Great Circle of life

Playing Indy and the Great Circle at 90 FPS with path tracing enabled on a Steam Deck OLED is mind-blowing. (Image credit: Tom's Guide)

Indiana Jones and the Great Circle is even more impressive on Valve’s handheld PC. Once you figure out how to install GeForce Now on Steam Deck (a fairly simple process, especially if you pair the little machine with a mouse and keyboard), Indy’s action-adventure will blow you away.

I distinctly remember first putting MachineGames’ initially super-taxing love letter to the whip-cracking icon through its paces when it first launched on Xbox Game Pass in December 2024. Despite rocking an RTX 4080 in my home-built desktop at the time, I could barely lock to 60 FPS at 4K during the incredibly demanding sun-dappled early Marshall College level.

Fast forward with GeForce Now Ultimate taking advantage of frame-boosting Nvidia DLSS 4 features (which were patched into the Great Circle a few months after its launch), and my Steam Deck OLED can run Indy’s adventure maxed out with full path tracing at 90 FPS. Just as impressively, GeForce Now’s overlay normally tells me the game is clocking in at a responsively low 16ms of lag/ping. It’s a beautiful, buttery smooth experience on Valve’s surprisingly sharp 800p (1280 x 800) screen.

The price isn't right

Sky-high RAM prices could push more and more gamers towards cloud gaming. (Image credit: Shutterstock)

More than any other factor, it’s the cost involved with cloud gaming that makes it increasingly attractive in an era where the perils of AI have made traditional PC components prohibitely expensive. A quick search on both Amazon US/UK shows that your average third-party RTX-5080 costs roughly $1,200/£1,200. While I may generally suck at math, even my Sesame Street Big Bird brain can work out that a single desktop 5080 would cost you around 60 months of a GeForce Now Ultimate subscription.

Does Cyberpunk 2077 natively run a little smoother on my Asus ROG Zephyrus G14 (2025) compared to GeForce Now Ultimate’s RTX 5080 tier? Yes, by a semi-noticeable amount if you’re obsessed with frame pacing. When it comes to image quality though, even my Terminator levels of committed pupils can barely tell the difference between native and Team Green’s highest level of cloud gaming.

Couple RAMageddon with the prohibitive costs of GPUs, and the prospect of cloud gaming has never been more appealing. Don’t get me wrong, the potential of the cloud hasn't been fully realized yet. I recently retested the PlayStation Portal and I think its picture quality absolutely sucks.

But when it comes to GeForce Now Ultimate, you’re looking at a cloud service that genuinely shames both the PS5 Pro and Xbox Series X in several key metrics if you’re fortunate to have strong, stable Wi-Fi.

Couple RAMageddon with the high costs of GPUs, and cloud gaming has never looked quite so appealing"

Ultimately, I can’t forget the RAM-shaped Dumbo in the room. During researching this article, I revisited my Amazon UK orders from 2025. Skip back to May, and I paid precisely £99.95 (around $133 by current exchange rates) for two 16GB sticks of Corsair Vengeance RGB DDR5 RAM.

In the build up to Christmas as I write this, the only current option on the Big A (which is admittedly through a third-party seller) for said amount of the same Corsair memory costs £504. To my American friends, that’s around $674.

Perhaps it’s time to let the age of game streaming reign supreme… as long as Nvidia GeForce Now Ultimate keeps up its immaculate standards.

Nvidia RTX 60-series in early 2027? The leak sounds huge, but the numbers don’t add up

Jason England — Thu, 11 Dec 2025 11:06:14 +0000

We’ve got a whole lot of news about Nvidia’s RTX 50 SUPER Series. First, it was coming and will boost the amount of previous GPU video memory. Next, it was potentially cancelled, but then it was only postponed. But Team Green may have its eyes on a bigger price — the next generation.

According to Red Gaming Tech, Nvidia is already making preparations to launch the next-generation RTX 60-series GPUs — touting a Q1 2027 launch announcement for the RTX 6090 (translation: announced at CES 2027).

So what can you expect from Nvidia’s RTX 60-series? And given the RTX 50 SUPER Series is rumored to launch in Q3 2026, am I right in having some doubts about this early 2027 leak for the next generation? Let's get into it.

What could we see in Nvidia’s RTX 60 Series?

Built on Rubin architecture — possibly using TSMC’s 2nm or 3nm process.
RTX 6090 is rumored to get at least a 40% uplift over RTX 5090 in rendering and ray tracing.
No more 8GB VRAM GPUs? Gamers have made their position very clear on that through sales, and Nvidia may be listening.

The rumors have been small and speculative, but let’s pick some details out of here and look at what’s happening in Nvidia’s AI chips world (because what data centers get, gamers almost always get further down the line).

First off, I think it’s clear that RTX 60-series cards will get the next-generation Rubin GPU architecture. Nvidia CEO Jensen Huang's been talking about this coming in 2026 to AI GPUs for data centers at practically every Nvidia GTC. And just like Blackwell made the jump to consumer cards, Rubin will, too.

What will this give you? On paper, performance-wise, it should be a huge uplift — to AI customers, Nvidia’s talking about Rubin being over 2x faster than Blackwell. Of course, in a gaming GPU, that will be tamed slightly, but it’s a good starting point for how it can handle intensive gaming scenarios like complex level rasterization and path tracing.

(Image credit: Nvidia)

To put a number on it, the rumor mill is pointing towards the RTX 6090 having at least a 40% uplift in performance over the RTX 5090, along with the watts being pumped through it going up again in excess of 600W. Prepare your electric bills.

And given Nvidia has gone all in on AI, you can expect a new version of DLSS to launch alongside it. There are zero substantiated rumors or reports about DLSS 5, so I’m flying a little blind here. But if I were to ask for one thing, please make frame generation smarter.

(Image credit: Future)

By that, I mean instead of adding a fixed amount of extra frames that can lead to slight increases in gameplay latency, move to variable frame generation — dynamically adding frames as needed to hit a particular target.

Beyond that, it’s fair to expect enhanced path tracing and an even deeper AI integration into the likes of image reconstruction and denoising. Maybe even a more fully fleshed out version of Nvidia ACE, too, for more intelligent NPCs.

(Image credit: Future)

As for the memory side, Rubin has been touted to support much faster video memory speeds, but how much video memory will we actually get? Well, if we look at the postponed RTX 50 SUPER Series and how that was set to be a direct response to gamers just not buying 8GB VRAM cards (not enough for a lot of AAA titles nowadays), I think it’s clear that we won’t see them launch one with this little memory.

And finally, price. The big worry here is just how that VRAM scarcity will impact things here, and I’m sorry to say it probably will. Rumors suggest starting prices of $2,000-$2,500 for the RTX 6090, with some even saying over $3,000.

Analysis: here are my doubts

DRAM pricing crisis has probably thrown a lot of launch plans into flux.
Nvidia’s RTX 50 SUPER Series has been reportedly postponed, further throwing my own personal doubt on an early 2027 launch of RTX 60 Series.
My bet is that RTX 50 SUPER Series gets cancelled, and Nvidia goes full steam ahead on its next-generation.

(Image credit: Future)

Before I start, I want to say one thing. Given what we are seeing happening right now, if you can afford to buy one of the best GPUs at retail price (like a lot of them are right now), then you should do that. RTX 60 Series will be impressive for sure, but the ever-rising cost of TSMC’s foundry process, paired with VRAM spiking, makes me very nervous about what the final total will be.

That being said, my doubts about RTX 60 Series launching in 2027 are not planted in any logic if you were to look at the release cadence of Nvidia’s gaming GPUs. Consistently, generational jumps have happened every two years (ish — give or take a pandemic).

(Image credit: Future)

But we are in a very different situation with the RAM pricing crisis, which has postponed the launch of RTX 50 SUPER Series GPUs to Q3 of 2026 — due to the problems getting the video memory chips needed for it to happen. From a consumer tech perspective, does it make sense to launch something new so close to the next generation? I don’t think so.

Becoming the picks and shovels company in this AI goldrush (or AI bubble, whichever way you look at it) has shifted Nvidia’s priorities to where it makes the lion's share of its money — meaning the timeline of gaming GPUs is more in flux, and items are more vulnerable to being canned.

Reading the tea leaves of what is coming down the line and pairing it with this leaker’s allegation, I think one of two things will happen in 2027:

RTX 50 SUPER Series launches next year, and RTX 60 Series is postponed to late 2027/early 2028.
My bet: RTX 50 SUPER Series gets cancelled, and RTX 60 Series launches in 2027.

Of course, all of this is my own opinion paired with a little guesswork, but I’d love to know what you think in the comments!

AMD’s FSR Redstone is finally here to take on Nvidia’s DLSS 4 — here’s everything you need to know

Jason England — Wed, 10 Dec 2025 16:38:25 +0000

It was first announced way back at Computex 2025, but finally, AMD has launched FSR Redstone — its own platform of AI trickery to make supported PC games perform even better. This is Team Red’s effort to close in on the reigning champ of resolution scaling and frame generation: Nvidia’s DLSS 4.

So what does FSR Redstone do exactly? And how does it stack up to what Nvidia has cooked up? Let’s get into it.

What is AMD FSR Redstone?

FSR Redstone is now available via an AMD Adrenalin driver — available only on RX 9000 GPUs (like the RX 9070 XT I reviewed recently).
AI/machine learning-aided resolution scaling, ray tracing and frame generation.

In a way, Redstone is AMD realizing its initial approach wasn’t working — a plan B approach. You see, previous versions of FSR didn’t use machine learning/AI to do all of its performance optimizations. Instead, it would rely on preset algorithms for resolution scaling and frame generation.

This always meant it was a step behind what Nvidia was doing with DLSS, fully embracing AI. FSR Redstone is the all-important transition point, as AMD is now all-in on similar machine learning capabilities to do the following things:

(Image credit: AMD)

FSR Upscaling: This is the same machine learned resolution scaling that you saw in FSR 4 on any AMD Radeon RX 9000 series GPU so far. Basically, it renders the game at a lower resolution, then uses an upscaling model trained on the game itself to boost the resolution back up — easing the pressure of intensive games.
FSR Frame Generation: AMD’s adding some machine learning sauce to its frame generation, too. A neural network running under the hood will analyze every rendered frame and the underlying game data, and merge it with generated frames to provide a smoother frame rate.
FSR Ray Regeneration: Ever looked at ray tracing in a game and thought something was ever so slightly off? Maybe it’s way too shiny or too noisy. That’s what Ray Regeneration aims to resolve, by using machine learning to denoise ray-traced reflections and a neural network to do it efficiently (rather than testing your GPU’s limits through raw rendering).
FSR Radiance Caching: Another neural network at play here, but for the complexities of ray tracing and path-traced lighting. All of that indirect lighting is such a challenge — even for the latest and greatest hardware, so machine learning is here again to calculate the radiance of illumination within the level you’re playing.

(Image credit: AMD)

The end result of all of this is a less clunky process flow for optimizing game performance. Using neural networks, particularly for frame generation, virtually eliminates ghosting compared to AMD’s past efforts.

Plus, it will make the calculations for ray tracing and upscaling far less GPU-intensive, delivering sharper textures and smoother frame rates at higher levels of detail (at least that’s what I’ve seen from my friends testing it out).

How many games support AMD FSR Redstone?

Over 200 games will support FSR Redstone by the end of 2025, including Arc Raiders, Call of Duty: Black Ops 7 and God of War Ragnarök.
But 32 games are currently listed to support FSR Frame Generation (machine learning)

(Image credit: AMD)

One look at the list on AMD’s website shows a whole lot of titles that support FSR Redstone technology. Specifically, Team Red claims over 200 games will support it by the end of 2025.

But scroll down this list a little bit, and you see it's not that simple. There’s a secondary list at the bottom titled “AMD FSR Frame Generation (ML) Support.” This seems like the neural frame gen part of Redstone is only coming to 32 games.

How does it compare to Nvidia DLSS 4?

DLSS 4 has a lot of the same features as FSR Redstone, but has been around a lot longer to grow those features.
The Transformer AI used in DLSS 4 is more mature, providing a stable output.
DLSS Frame Gen is now multi-frame generation — two additional frames for every one rendered frame than FSR Redstone.
DLSS 4 supports older GPUs, whereas FSR Redstone is locked to AMD’s newest cards.
They are seemingly tied to Ray regeneration/reconstruction and neural radiance caching.
But AMD can catch up.

(Image credit: Nvidia)

DLSS 4 has been around for a while now and has had much more time to mature into its new feature set. It offers pretty much all the same features as AMD’s FSR Redstone does, but Nvidia seems to be doing these a little bit better, given they’ve been around the block a while.

First up, the Transformer AI used in DLSS 4 has been around a lot longer than FSR upscaling (even when it was called FSR 4 for a couple of months before). That means you should get a more stable game output from Team Green.

Second, frame generation. While the frame interpolation of FSR Redstone is 1 new frame for every 1 rendered, DLSS 4 has multi-frame generation — giving you a 3:1 ratio. And third is compatibility: you can use DLSS 4 features on cards all the way back to RTX 30-series, but AMD is locking Redstone just to Radeon RX 9000 cards and newer (when they’re launched).

The big disadvantage here to FSR Redstone is game support. Nvidia maintains a massive game compatibility lead with over 800 games benefiting from DLSS 4 automatically. AMD’s made an admirable effort with over 200 titles by the end of the year (and they can catch up from that), but it’s a huge gulf.

My take: This could be part of a killer Steam Machine build

(Image credit: Future)

So here’s my take on it. In the Windows PC gaming space, Nvidia should be a little worried. Its top competitor has finally got its act together with FSR Redstone for its GPUs, which are a better value for money for gaming. And while there is still a clear lead in terms of game support, that’s a gap AMD can close.

But I see a far bigger threat coming in somewhere else: SteamOS. While the Steam Machine looks set to disrupt both the console market and PC gaming in a big way next year, that hasn’t stopped people from building AMD-laden machines and installing SteamOS.

(Image credit: Valve)

The outgoing FSR 4 is supported on the underlying Linux after all — it should only be a matter of time before the Proton layer for translating DirectX 12 code used in Windows games gets updated for Redstone.

Imagine a super-powered Steam machine with all of this new AI-aided gaming goodness, but without the bulk of Windows 11… To me, that sounds like a mighty exciting proposition, and if I were in any post-release planning meeting in AMD’s HQ, I’d be shouting for a closer tie with Valve to make something like this happen.

A major RAM price rise will impact laptops, PCs and GPUs in 2026 — Black Friday is your only chance to survive the surge

Jason England — Wed, 26 Nov 2025 09:30:00 +0000

The price of laptops, tablets, GPUs and PC components has already been under pressure this year. But now, thanks to the skyrocketing cost of RAM and SSD storage chips, it's looking like retail prices for tech could moon in 2026.

Look, I know you're hearing a whole lot about Black Friday deals right now, so I'll try to not make this sound too dramatic. But this year's sales season could be the last chance you get computing tech at these prices before huge increase next year.

...well the whole not sounding "too dramatic" didn't last long. But the point stands! Let me explain.

What’s happening?

(Image credit: Future)

The problem has been quietly growing throughout 2026, but now it's here with a bang thanks to the AI boom. You see, to run the likes of ChatGPT and Gemini, AI companies need to buy a whole lot of silicon for its data centers to handle all of your prompts. The critical chips in all of this are DRAM (Dynamic Random-Access Memory) and NAND flash, but you may know them by their other names: RAM and SSD storage respectively.

As the demand for these chips started to skyrocket because of this boom, two things have happened:

The companies that manufacture these types of chips are seeing demand for cheaper, older memory chips vanish, given their biggest customers (companies behind the AI boom) are throwing handfuls of cash at them for the latest stuff. So these older options are being phased out.
Smaller NAND flash chips are being deprioritized in favor of the highest bidders looking to pack their AI data centers with tons of memory.

Both of these together are leading to drastic silicon shortages, which are in turn leading to higher manufacturing costs for PC and smartphone brands, and inevitably, price jumps for us.

How bad could it be?

(Image credit: Future)

In one word: bad. In two words: real bad.

One place we can already see it happening in real-time is with DDR5 RAM kits you can buy for your desktop PC. People are finding prices are up to 171% higher compared to 2024. Looking at what I’ve been tracking to buy — the Crucial Pro 32GB DDR5 kit was $174.99, and it’s now $300 in just over one month. That is astonishing.

As for laptops, TrendForce’s latest investigation (reported by TechPowerUp) has this pinned at a predicted 5-15% price rise in 2026. Meanwhile, AMD and Nvidia are already signaling price hikes for their GPUs, and this could be by at least 10%. Combine GPU and RAM jumps, and that could be a massive increase for prebuilt PCs.

But there is a small window of hope

(Image credit: Future)

Luckily, a lot of companies do pay attention to this stuff and plan ahead — stockpiling the chips they need to prepare for high demand moments (like Black Friday).

That is why I can confidently say that this is your last chance to grab a laptop, GPU or PC before the big old price surge takes effect. The only one to publicly acknowledge this is happening is Lenovo, but I know other companies are too.

So what you have between now and the end of Cyber Monday is a final chance to grab a laptop at MSRP, a GPU at slightly below MSRP, and a desktop at a reasonable price before the memory surge sends those prices up in smoke.

Surge-proof Black Friday tech deals

Laptop deals

This is the lowest price I've ever seen for the 13-inch MacBook Air M4, making this remarkably fast and long-lasting ultraportable far more affordable. This is the entry-level model with 16GB of RAM and a 256GB SSD, but the more powerful models are also on sale if you'd rather have more memory!
MacBook Air 13" M4 (16GB/512GB): was $1,199 now $949
MacBook Air 13" M4 (24GB/512GB): was $1,399 now $1,149View Deal

Looking for plenty of heat under the hood? Then you'll want to save $400 on this Acer Predator Helios Neo 16S AI. Why? Because it boasts an Intel Core Ultra 9 275HX CPU, RTX 5070 Ti GPU, 32GB of RAM and a 1TB SSD. And that display? Expect a 16-inch QHD+ (2560 x 1600) OLED panel with a whopping 240Hz refresh rate. Worthwhile? You know it. View Deal

The discount is minor on paper, but this is actually a drop from $899 and the lowest price on this MacBook Air competitor — packing a gorgeous OLED display, Snapdragon X, 16GB DDR5 RAM and a 512GB SSD in a slim body that weighs just under a kilogram.View Deal

Desktop PC deals

Over $900 off a power-packed RTX 5060 gaming PC is always a desktop to keep an eye on, esepcially when it looks this flashy! This AOACE delivering an AMD Ryzen 7 5700X CPU, an Nvidia GeForce RTX 5060 GPU (8GB), 32GB of RAM and a 1TB SSD. Plus, Wi-Fi 6E connectivity and six ARGB fans for good measrue. For under $900, this will breeze through 1080p gaming (and even a big beyond). View Deal

GTA 6 probably won’t launch on Steam until 2027 (or beyond), yet in the here and now, you can enjoy amazing games like Cyberpunk 2077 at max settings at well over 60 FPS thanks to that 5070 GPU, Intel Core Ultra 7 263F CPU and 32GB of DDR5 RAM. 1TB of NVMe storage means you’ll have to be selective with the games you install, yet this is still an enticing deal when you consider what the ROG Xbox Ally X costs. View Deal

A fairly low price for an RTX 5070 Ti gaming rig! Sporting an AMD Ryzen 7 8700F processor combined with 32GB of DDR5 RAM and a 2TB SSD, this MSI Aegis Z2 (which comes in a slick all-black case and all the RGB-lit fans) will push gaming even further than to its top-line 5070 Ti graphics cards. Hard to go wrong here. View Deal

Tablet deals

This is a nice discount on the base iPad with 128GB of storage, and it's one of the best iPads yet. In our iPad 11 review we lauded this tablet for its improved speed, longer battery life and the fact Apple finally bumped it up to 128GB of storage by default. View Deal

Here's a sweet discount on the latest 11" iPad Air packing Apple's M3 processor, an 11-inch LED (2360 x 1640) display and 128GB of storage. I think this is generally the best iPad for most people because it's better than the basic iPad but doesn't cost as much as the iPad Pro, yet offers nearly all the same features. Check out our iPad Air M3 review to see what I mean!View Deal

Save a nice chunk of change on Samsung's Galaxy Tab S10 FE+, a mid-range tablet that just launched this year and offers a big, beautiful 13.1-inch LCD panel and a 15+ hours of tested battery life. Read all about it in our Samsung Galaxy Tab S10 FE+ review!View Deal

GPU deals

This upper-mid range option is my current top-rated best graphics card for its stellar 1440p and 4K gaming performance, and zippy speeds in creative pro workloads and AI. It's a small but important dip beneath that $749 MSRP. Make sure you use the rebate to get that discount!View Deal

Now this is the RTX 5060 Ti you'll want to bag, as it comes with 16GB of GDDR7 video memory, which is more than enough to handle games like Indiana Jones and the Great Circle, Cyberpunk 2077 and more. Expect smooth 1080/1440p gameplay, and with Nvidia's multi-frame gen, frame rates going well beyond 120 FPS. View Deal

The budget GPU killer is back down to MSRP after this $20 saving — Intel's Arc B580 is plentiful in video memory and packs solid 1080p and 1440p gaming performance.View Deal

RAM deals

As they always say in PC building: when in doubt, bundle it. This isn't the all-time lowest price (you won't find anything like that for a while given the shortage). But in the context of the wider problem we're facing, a bundle with a 4TB PCIe Gen 5 SSD and zippy DDR5 RAM at this price is a steal.View Deal

A modest but warmly welcome price cut on 16GB of DDR5 RAM — more than enough for builds sporting the likes of the RTX 5060 Ti and a mid-range CPU.View Deal

Make sure you use the code "BFAE525" on this one. It's super rare that you'll find a 32GB DDR5 Kit on its own for less than $200 at the moment, so hurry!View Deal

Don’t wait for something new

(Image credit: Shutterstock)

To sum up what my chats have been like with people in the industry over the past couple of months in one word, it would be nervous. There’s a lot of new stuff coming down the line like Intel Panther Lake and Snapdragon X2 Elite, and I’m definitely excited to try them out.

But what if it's way too expensive to even consider an upgrade? And at that point, the prices of those older systems have probably gone up with reduced supply anyway. You’re damned if you do and damned if you don’t.

One thing I’m noticing from seeing what are the most clicked Black Friday deals on our site is that value for money is so critical. And if this trend continues, that feeling of value is going to evaporate quickly from the tech of 2026.

For more discounts, check out our Black Friday deals live blog.

Shop today's best Black Friday deals

Amazon: 50% off Ring, Ninja, AirPods and more
Walmart Plus: 50% off @ Walmart
REI: half-off Arc'teryx, Patagonia, more
Dell: monitors from $69, laptops from $299
Nectar: 50% off mattresses + bundles
Best Buy: laptops from $149
Lenovo: doorbuster deals from $319
Wayfair: up to 80% off home decor
Lululemon: top picks from $9
Newegg: Doorbuster deals from $6
Walmart: holiday gifts from $15
Birkenstock: deals from $20 @ Nordstrom
Saatva: luxury mattresses from $1,099

I’ve got bad news, PC gamers: Nvidia RTX 50 SUPER series may have been scrapped, and current GPUs look set to become more expensive

Jason England — Fri, 07 Nov 2025 11:25:26 +0000

A lot of PC gamers have been holding out for the Nvidia GeForce RTX 50 SUPER series GPUs, which were reportedly set to launch early next year… Notice how I said “were” there? Well, I’ve got some (rumored) bad news.

According to Uniko’s Hardware, Nvidia has cancelled its SUPER cards because of “the crazy shortage” stopping the ability to bring these to market. On top of that, the shortage could raise prices of the current best graphics cards, and make them “more expensive very soon.”

[rumor]due to the crazy shortage recently, 3gb gddr7 cant make it to the consumer market for desktop, thus the super series is cancelled.and the current models are expected to be more expensive very soon, because of the increasing cost of 2gb gddr7. pic.twitter.com/yMSMj1MjqJNovember 7, 2025

All about GDDR7

(Image credit: Future)

The secret sauce to RTX 50 series GPUs (minus the 5050) has been the use of much faster GDDR7 video memory. This has increased the bandwidth and drastically sped up the ability to deliver higher-resolution graphics, as well as delivering much quicker AI performance too.

That’s all fair and good, but one of the common complaints you see flying around is that some of the range packs quite a small amount of it. Because sometimes it doesn’t matter how fast the video memory is — if there isn’t enough to hold the ever-increasing demands of GPU-intensive AAA games like Indiana Jones and The Great Circle, then you’ll run into limitations.

I found out as much when testing the RTX 5070 with 12GB and the RTX 5060 with 8GB. Well, the RTX 50 SUPER series, Nvidia’s fix was reportedly a simple one: switch out the current 2GB GDDR7 modules on the GPU and switch them for 3GB ones. That would be an effective 50% uplift across the board.

GPU	RTX 5080 SUPER	RTX 5080	RTX 5070 Ti SUPER	RTX 5070 Ti	RTX 5070 SUPER	RTX 5070
Price	$999-1,199	$999	$749-$799	$749	$549-$599	$549
CUDA Cores	10752	10752	8960	8960	6400	6144
Video Memory	24GB GDDR7	16GB GDDR7	24GB GDDR7	16GB GDDR7	18GB GDDR7	12GB GDDR7
TGP	415 Watts	360 Watts	350 Watts	300 Watts	275 Watts	250 Watts

The rumors and leaks pointed heavily towards this happening, but these 3GB GDDR7 models are reportedly in extremely short supply, with little hope of them reaching the desktop consumer market.

Instead, these chip modules are possibly being reserved for different uses, and Uniko’s guess is you’ll see them in RTX 5090 laptops and the RTX PRO 6000 Blackwell GPU.

Driving up prices

(Image credit: Future)

More concerning than this is the question of what will happen to the current crop of RTX 50 series GPUs? Given the recent shortages, it looks like the cost of the 2GB GDDR7 modules is going up too, which means “current models are expected to be more expensive very soon.”

Given this is a rumor, we can’t say with 100% certainty whether there’s any truth to this. But what this does pose is a small window of opportunity to those who were looking to upgrade. Right now, most RTX 50 series cards are either at MSRP or slightly below, and given the macroeconomics of the current situation, they may not stay that way for long.

So I have one piece of advice: think about buying now. Of course, there are certain criteria to hit with this — namely if you already have an RTX 40 series card, you don’t need the latest GPU. But if you’re on anything older and you’ve been holding out for SUPER, if this rumor holds water, this could be your final chance to buy before costs go up.

Black Friday RTX 50-series GPU deals in the U.S.

My main gripe with the RTX 5070 was the price to performance at its $549 MSRP. Now it's $70 below, that's a different story!View Deal

It's a small but important dip beneath that $749 MSRP. Make sure you use the rebate to get that discount!View Deal

Always avoid the 8GB RTX 5060 Ti, and that becomes so much easier as the price of the 16GB model has dipped below $400View Deal

Black Friday RTX 50-series GPU deals in the U.K.

The price of the RTX 5070 Ti just keeps tumbling in the U.K. For the first time, it's available for less than £700.View Deal

Another all-time low has been hit on the RTX 5070 too! Once again, at this price, consider my RTX 5070 review a 4-star rather than the lower score I gave it because of the higher cost.View Deal

That's the trifecta for Overclockers right there. Nvidia's 16GB RTX 5060 Ti has taken another tumble too — giving strong 1440p gameplay and shockingly good 4K with DLSS 4 activated at an affordable price.View Deal

Best graphics cards in 2026: I've tested every GPU to find the best bang for your buck

Jason England — Mon, 27 Oct 2025 10:00:00 +0000

You're here to find out the best GPU you can buy in 2026. Well, after testing the latest graphics cards from Nvidia, AMD and Intel, from gaming to video editing and AI performance, my team of experts and I are here to cut through the jargon and guide you to the absolute best graphics cards. But right now, more than ever, that's difficult to pin down.

Today's market is more complex than ever, and we have the RAM crisis to thank for that. Prices are jumping to absurd levels, meaning what was once a more budget-friendly graphics card has now been pushed up to mid-range prices — and so on.

It's been rough. So, at a time when GPUs are seeing truly obscene price hikes, there are huge benefits to finding the sweet spot between price and performance. That's why we've named the Intel Arc B580 the GPU for most, but there are other graphics cards that should be on your radar. So, let’s dive in to find the best GPU to power your setup for years to come.

GPU Price index

GPU prices have always fluctuated, but RAMageddon has tipped the scales so far that it isn't about raw horsepower anymore. It isn't as simple as saying "just get an RTX 5090," unless you want to spend close to or over $4,000. In fact, you'll find that an RTX 5070 Ti, which we rate highly, can be $1,000 and beyond, despite its $749 MSRP (Manufacturer’s Suggested Retail Price).

It's not all "doomer narrative," though, as we now have Nvidia's DLSS 4.5, AMD's FSR 4.1 and Intel's XeSS 3 for multi-frame generation and resolution scaling, boosting performance while working with less. And now, we're gearing up for Nvidia DLSS 5 in fall 2026. Even better, the AMD Radeon RX 9070 GRE GPU can actually be found at MSRP (for now)!

Originally, our golden rule was to never buy a card that is over $50/£50 above MSRP, but costs have gone well above this (but not for all!). Regardless, we're here to find the best possible price, and I recommend keeping an eye out during big sale events. Here's a GPU pricing matrix to ensure you get the best possible price.

GPU	MSRP (US)	Lowest price (US)	RRP (UK)	Lowest price (UK)
Nvidia GeForce RTX 5070 Ti	$749	$979	£729	£809
AMD Radeon RX 9070 XT	$599	$752	£569	£679
AMD Radeon RX 9070 GRE	$549	$549	£489	£489
Nvidia GeForce RTX 5060 Ti (16GB)	$429	$569	£399	£469
Intel Arc B580	$249	$309	£239	£299

Best graphics cards: 3DMark test results

3DMark benchmarks are a good way of seeing how much overall gaming performance your GPU has.

I take the average of all of the tests I do, and while this isn't an exact science (some games are better optimized), here's a guide:

18,000 and above is capable of 4K gaming
11-18,000 is squarely aimed at 1440p
And 8,000 upwards is more than ideal for 1080p

This guide to 3DMark numbers is not a hard and fast rule, as there are some cheat codes to unlock more performance like Nvidia's DLSS, AMD's FSR and Intel's XeSS. This guide is based on what the raw performance can achieve without any AI trickery. You can find out how they compare below:

3DMark average benchmark test results

Higher is better

3DMark average

Nvidia GeForce RTX 5070 Ti (16GB GDDR7)

20271

AMD Radeon RX 9070 XT (16GB GDDR6)

19679

AMD Radeon RX 9070 GRE (12GB GDDR6)

14115

Nvidia GeForce RTX 5060 Ti (16GB GDDR7)

11195

Intel Arc B580 (12GB GDDR6)

9692

07,50015,00022,50030,000

3DMark average Data
Product	Value
Nvidia GeForce RTX 5070 Ti (16GB GDDR7)	20271
AMD Radeon RX 9070 XT (16GB GDDR6)	19679
AMD Radeon RX 9070 GRE (12GB GDDR6)	14115
Nvidia GeForce RTX 5060 Ti (16GB GDDR7)	11195
Intel Arc B580 (12GB GDDR6)	9692

The best graphics cards you can buy right now

Best GPU overall

(Image credit: Future)

1. Intel Arc B580

The best GPU you can actually afford

Strong 1440p and 1080p gaming

12GB of VRAM

Intel XeSS is improving nicely

$249 is a steal

Turn ray tracing off

Not the best for work

This may come as a shock, but ever since bringing Intel XeSS 3 on board, the Intel Arc B580 is now the GPU for most to get. How so? With multi-frame generation and the budget price it continues to (kind of) stick to, you can get a lot out of playing the latest games with Intel's graphics card.

This is a real budget buster of a card that proves you can get more than 8GB of video memory for $250 and a fantastic 1080p/1440p gaming graphics card for a great price. Ray tracing may not be its best friend, but Intel XeSS is improving nicely, and it’s simply the best GPU for many.

Buy it if

👍 You want cheap and cheerful gaming: Tweak the settings to get the optimal levels, and you’re in for a damn good time. At 1440p and maxed out settings, you can comfortably see over 60 FPS. Turn on XeSS, and that goes up to nearly 100 FPS with minimal ghosting around fast-moving in-game objects.

👍 You don’t want to spend a fortune: I mean, it’s $249, and prices haven't skyrocketed — sometimes it's discounted to $239. I’m floored by what it was able to achieve at the price.

👍 XeSS 3 now included: With Intel adding XeSS 3 with multi-frame generation, this GPU scores even more highs with its 12GB of VRAM. Now, it's the value king.

Don’t buy it if

👎 Ray tracing matters: Put simply, just turn off ray tracing here; otherwise, you will have a bad time. No XeSS frame gen can fix the huge fall off in performance you’ll see.

👎 You need it for work too: Whether it’s animation, video editing or an AI workload, the Arc B580 isn’t really built for these, so you won’t see great speeds working with it.

Intel Arc B580 test results

Gaming benchmarks/Productivity and AI benchmarks

Frames Per Second (FPS)

Cyberpunk 2077 (1440p Ultra no ray tracing XeSS frame gen)

Black Myth: Wukong (1440p Medium no ray tracing XeSS)

Forza Horizon 5 (1440p Ultra no ray tracing XeSS)

106

Call of Duty Modern Warfare (1440p High XeSS resolution scaling/no frame gen)

126

Alan Wake 2 (1440p High/Medium no ray tracing XeSS)

037.575112.5150

Frames Per Second (FPS) Data
Product	Value
Cyberpunk 2077 (1440p Ultra no ray tracing XeSS frame gen)	90
Black Myth: Wukong (1440p Medium no ray tracing XeSS)	77
Forza Horizon 5 (1440p Ultra no ray tracing XeSS)	106
Call of Duty Modern Warfare (1440p High XeSS resolution scaling/no frame gen)	126
Alan Wake 2 (1440p High/Medium no ray tracing XeSS)	44
Blender Median score (Animation)	N/A
Procyon Stable Diffusion XL (AI image gen)	N/A
Procyon Llama-3.1-8b (local LLM AI)	N/A

1 of 2

Best value GPU

(Image credit: Tom's Guide)

2. AMD Radeon RX 9070 GRE

The best GPU for 1440p gameplay that's actually at MSRP

Fantastic 1440p gaming

Solid ray tracing performance

FSR Redstone is getting much better

Real value for money…

…provided it stays at $549

FSR game compatibility is still limited

AI/Creator workloads behind Nvidia

In a very welcome turn of events, AMD has released its all-new Radeon RX 9070 GRE at a price that won't blow a hole in your wallet. That's right, $549/£489, and we're hoping this cost stays that way for the foreseeable future. Even at this price, this GPU still delivers exceptional 1440p gaming performance and even greater ray tracing capabilities than Nvidia's similar RTX 5060 Ti with 16GB.

Throw in improvements to FSR Redstone upscaling, and the AMD Radeon RX 9070 GRE offers up rare value that's harder to find these days. No, it's not as strong as its 9070 XT sibling, and FSR's compatibility and performance in games has its limits compared to Nvidia's DLSS, but for lean, clean gaming prowess, this is pure value.

Buy it if

👍 You want the best bang for your buck: If the RX 9070 GRE stays at $549, it’s the rare mid-range GPU that delivers genuine value in today’s market.

👍 You’re aiming for top-tier 1440p gaming: Performance closely matches Nvidia’s pricier RTX 5070 and beats the 16GB RTX 5060 Ti, making it a 1440p sweet spot.

👍 You need enough VRAM for modern titles: With 12GB of GDDR6 video memory, the RX 9070 GRE has the headroom needed for big new releases and plenty of demanding games to come (with FSR to help out).

Don’t buy it if

👎 You prioritize AI/creator workloads: The RX 9070 GRE lags Nvidia cards for productivity and AI performance—look elsewhere if these are your focus.

👎 You want the most powerful upscaling and features: DLSS still leads FSR Redstone in features and game support, so stick with Nvidia's offerings if you're after bleeding-edge upscaling.

AMD Radeon RX 9070 GRE test results

Gaming benchmarks/Productivity and AI benchmarks

Frames Per Second (FPS)

Cyberpunk 2077 (1440p Ultra ray tracing no FSR)

Black Myth: Wukong (1440p Cinematic no FSR)

Forza Horizon 5 (1440p Ultra no FSR)

142

Call of Duty Modern Warfare (1440p Max no FSR)

136

037.575112.5150

Frames Per Second (FPS) Data
Product	Value
Cyberpunk 2077 (1440p Ultra ray tracing no FSR)	36
Black Myth: Wukong (1440p Cinematic no FSR)	81
Forza Horizon 5 (1440p Ultra no FSR)	142
Call of Duty Modern Warfare (1440p Max no FSR)	136
Blender combined score	N/A
Procyon Stable Diffusion XL (AI image gen)	N/A

1 of 2

Best mid-range GPU

(Image credit: Future)

3. Nvidia GeForce RTX 5060 Ti (16GB)

Best mid-range GPU

Supereme 1440p performance

Shockingly good at 4K too

Content creation and AI workloads are a breeze

16GB video memory

High power consumption when idle

Let me emphasize: GET THE 16GB VERSION! The 8GB version of the RTX 5060 Ti is far too limited on the video memory. But the 16GB version is a banger of a mid-range GPU that sees its value unlocked with DLSS 4.

In raw rasterization, you can get impressive frame rates at 1080p and 1440p for sure, but with Nvidia’s AI magic turned on, you can really enjoy 4K single-player gaming too! Plus, you’ll find decent content creation performance too.

Buy it if

👍 You want solid 1440p gaming performance: Raw game rendering speeds are impressive here, and the 16GB of GDDR7 video memory is more than enough to store all those textures and ensure zero bottlenecking of complex games.

👍 You don’t mind a bit of DLSS 4: When it comes to the debate about DLSS 4 AI trickery, I’m more on the side of supporting it, unlocking more value from cheaper GPUs like this. And it absolutely does with zippy 4K gaming performance with low latency.

Don’t buy it if

👎 You need the absolute best in multiplayer: If you’re a competitive player, latency less than 10ms is essential, and in my testing, I couldn’t achieve this in online shooters.

👎 You have big AI plans: There are only minor gains in AI performance over the RTX 4060 Ti, which doesn’t make this a good value card if you’re looking to run on-device AI tasks.

Nvidia GeForce RTX 5060 Ti test results

Gaming benchmarks/Productivity and AI benchmarks

Frames Per Second (FPS)

Cyberpunk 2077 (1440p Ultra/Psycho Super Res)

Cyberpunk 2077 (1440p Ultra/Psycho Super Res DLSS 4 MFG 4x)

132

Black Myth: Wukong (4K Cinematic)

Black Myth: Wukong (4K Cinematic DLSS 4)

Call of Duty Modern Warfare (1440p Max no frame gen)

Alan Wake 2 (1440p Max)

Alan Wake 2 (1440p Max DLSS MFG 4x)

102

037.575112.5150

Frames Per Second (FPS) Data
Product	Value
Cyberpunk 2077 (1440p Ultra/Psycho Super Res)	17
Cyberpunk 2077 (1440p Ultra/Psycho Super Res DLSS 4 MFG 4x)	132
Black Myth: Wukong (4K Cinematic)	25
Black Myth: Wukong (4K Cinematic DLSS 4)	41
Call of Duty Modern Warfare (1440p Max no frame gen)	80
Alan Wake 2 (1440p Max)	15
Alan Wake 2 (1440p Max DLSS MFG 4x)	102
Blender Median score (Animation)	N/A
Procyon Stable Diffusion XL (AI image gen)	N/A
Procyon Llama-3.1-8b (local LLM AI)	N/A

1 of 2

Best GPU for gaming

(Image credit: Future)

4. AMD Radeon RX 9070 XT

Best GPU for gaming

Very impressive 4K gaming performance

16GB video memory

Better gaming than RTX 5070 Ti at a lower price

FSR 4 shines…

…but FSR 4 can be awkward to use

Some ray tracing hiccups

Not quite available at MSRP

Not great for AI and productivity

If your main goal is getting awesome gaming performance, AMD’s Radeon RX 9070 XT nails it while being cheaper than the RTX 5070 Ti at the same time. Overall, you’re getting roughly 95% of the performance of Nvidia’s card (sometimes even faster frame rates) at 80% of the cost.

To me, that’s the sweet spot of gaming GPUs. You could, of course, blow thousands of dollars and get the absolute maximum, but is that really the best use of your money when you see what you can actually achieve with this card?

Buy it if

👍 You want maximum gaming bang for your buck: When it comes to solid gaming performance that either comes close to or beats the $150 more expensive RTX 5070 Ti, the Radeon RX 9070 XT has it in the bag.

👍 You don’t want to fiddle around with complicated connectors: After testing Nvidia cards and growing a little frustrated with the 2x6 12VHPWR connector, it’s a breath of fresh air to go back to a card that just requires standard 8-pin PCIe power connectors.

Don't buy it if

👎 You care about ray tracing and wide game support: Yes, Nvidia’s ray tracing cores still have the edge over AMD. And by that last point, I don’t mean this just won’t work with all games — of course it will. What I mean is Nvidia’s DLSS 4 has far more support (over 175 games) vs AMD’s FSR 4 (over 85 games).

👎 You want to get work done with it too: While this is amazing for gaming, it’s not so hot on creator performance or AI. Nvidia clearly has the advantage here.

AMD Radeon RX 9070 XT test results

Gaming benchmarks/Productivity and AI benchmarks

Frames Per Second (FPS)

Cyberpunk 2077 (1440p Ultra/Psycho Super Res)

Cyberpunk 2077 (1440p Ultra/Psycho Super Res DLSS 4 MFG 4x)

102

Black Myth: Wukong (4K Cinematic)

Call of Duty Modern Warfare (4K Max no frame gen)

Alan Wake 2 (1440p Max)

037.575112.5150

Frames Per Second (FPS) Data
Product	Value
Cyberpunk 2077 (1440p Ultra/Psycho Super Res)	50
Cyberpunk 2077 (1440p Ultra/Psycho Super Res DLSS 4 MFG 4x)	102
Black Myth: Wukong (4K Cinematic)	57
Call of Duty Modern Warfare (4K Max no frame gen)	92
Alan Wake 2 (1440p Max)	92
Blender Median score (Animation)	N/A
Procyon Stable Diffusion XL (AI image gen)	N/A
Procyon Llama-3.1-8b (local LLM AI)	N/A

1 of 2

Best GPU for performance

(Image credit: Future)

5. Nvidia GeForce RTX 5070 Ti

The best GPU we wish we could afford

A 4K monster

Content creation and AI workloads are a breeze

16GB video memory

High power consumption when idle

Don't get us wrong. The RTX 5070 Ti is still one of the best offerings you can get in a GPU, but prices are outrageous right now. If you can find it closer to its $749 MSRP (fingers crossed), then you'll definitely want to go for it. But for over $1,000? It's a stretch.

Still, the RTX 5070 Ti is the sweet spot of Nvidia’s RTX 50-series — bringing just the right amount of blistering performance and versatility across 1440p gameplay, and a true showcase of DLSS 4 AI trickery at 4K to boot. But it doesn’t forget about productivity and AI either, as those cores are finely tuned to deliver huge GPU performance to your prosumer workloads too.

Buy it if

👍 You’re looking for great gameplay across the board: In all of my tests, the RTX 5070 Ti was a stellar, versatile GPU for gaming — bringing esports-level frame rates at 1080p, silky smoothness at 1440p, and even 60+ FPS in 4K in some circumstances. Throw in DLSS 4 and multi-frame gen, and nothing gets in its way of delivering a buttery gaming experience.

👍 You also need a creative powerhouse: Whether you’re editing multiple layers of 4K video in Davinci Resolve or doing AI neural Photoshop edits locally on-device, the RTX 5070 Ti keeps up with it all. Throw in support for the 4:2:2 color format, and you’ve got flexibility and fidelity in all your prosumer tasks.

Don’t buy it if

👎 You’re worried about power consumption: Yes, I know that if you’re buying a GPU, your electric bills may not be high on your priority list. But it’s worth noting that power usage when idle is a bit higher than I’d expect on a desktop graphics card.

Nvidia GeForce RTX 5070 Ti Test results

Nvidia GeForce RTX 5070 Ti test results

Gaming benchmarks/Productivity and AI benchmarks

Frames Per Second (FPS)

Cyberpunk 2077 (1440p Ultra/Psycho Super Res)

Cyberpunk 2077 (1440p Ultra/Psycho Super Res DLSS 4 MFG 4x)

240

Black Myth: Wukong (4K Cinematic)

Black Myth: Wukong (4K Cinematic DLSS 4)

107

Call of Duty Modern Warfare (4K Max no frame gen)

Alan Wake 2 (1440p Max)

Alan Wake 2 (1440p Max DLSS 4 MFG 4x)

125

075150225300

Frames Per Second (FPS) Data
Product	Value
Cyberpunk 2077 (1440p Ultra/Psycho Super Res)	51
Cyberpunk 2077 (1440p Ultra/Psycho Super Res DLSS 4 MFG 4x)	240
Black Myth: Wukong (4K Cinematic)	73
Black Myth: Wukong (4K Cinematic DLSS 4)	107
Call of Duty Modern Warfare (4K Max no frame gen)	86
Alan Wake 2 (1440p Max)	31
Alan Wake 2 (1440p Max DLSS 4 MFG 4x)	125
Blender Median score (Animation)	N/A
Procyon Stable Diffusion XL (AI image gen)	N/A
Procyon Llama-3.1-8b (local LLM AI)	N/A

1 of 2

What's coming up

(Image credit: Future)

Truthfully, I wasn't expecting many big GPU releases this year, but AMD proved me wrong with the launch of its Radeon RX 9070 GRE. With this in mind, stay tuned for more major drops as they happen. Otherwise, there are still big updates and features set to arrive in 2026. Here's a look at what's ahead:

During Nvidia GTC 2026, Team Green announced DLSS 5. This next-gen graphics tech uses real-time neural rendering that brings photo-realistic lighting and materials to RTX 50-series GPUs, and we got a taste of how it looks. It will be available in fall 2026, but we won't know how good it will be until we properly see it in action. For now, it's leaning towards "uncanny valley," but it's another feature Nvidia's GPUs will have down the line if you choose to buy one.
There's good news for AMD Radeon RX graphics card owners. AMD is bringing FSR 4.1 upscaling to older GPUs, meaning you won't need to upgrade to the latest Radeon 9000 series cards to get a bump in frame rates and visuals via AI upscaling (like Nvidia's DLSS). FSR 4.1 will be available for Radeon 7000 GPUs with RDNA 3 in July 2026, while Radeon 6000 series with RDNA 2 will see FSR upscaling in early 2027. So, expect FSR 4.1 support for over 300 games on older GPUs for titles that support it!
Wondering how the latest games look and perform, and the settings you should use on these graphics cards? Check out our thoughts on path tracing in Resident Evil Requiem (it's horrifyingly impressive) and path tracing in Pragmata (it looks glorious).

How do we test graphics cards?

(Image credit: Future)

To ensure fair and accurate results across the board, we here at Tom's Guide test all GPUs in the same system that sports an AMD Ryzen 7 9800X3D CPU, 32GB DDR5 RAM and a 2TB PCIe Gen 5 SSD.

We then test for three key purposes through a range of benchmarks and performance analysis: gaming (average frame rates and 1% lows tested across multiple titles), productivity (GPU-intensive tasks like animation rendering), and AI performance (image generation and local LLM processing).

This is done with a range of 3DMark gaming benchmark tests, benchmarks across the top-selling AAA titles on Steam, Procyon generative AI tests and the classic Blender benchmark.

I completed 10 games on AMD Radeon RX 9070 XT — here’s why this is a gaming champ that makes Nvidia blush

Jason England — Fri, 24 Oct 2025 11:00:00 +0000

The AMD Radeon RX 9070 XT surprised me in the best way possible — reaching higher frame rates in some games than the Nvidia GeForce RTX 5070 Ti, while also being up to $150 cheaper at the same time.

In raw rasterization, this is (mostly) a slick GPU that can run all your favorite titles at buttery smooth frame rates with all the fidelity options turned up to max. Ray tracing has definitely improved, but it hasn’t matched the same levels as Team Green — not that this is a complete dealbreaker as all your favorite AAA titles still look amazing with baked in lights, shadows and reflections.

And AMD’s AI trickery, known as FSR 4, is a fantastic resolution scaler that extracts more performance while eliminating practically all of that telltale ghosting. Put simply, Nvidia has enjoyed its time on top for a while, but there’s some serious competition now. If you’re sitting on $600 and you want the best bang for your buck, this is a darn good way to go.

That’s not to say it’s all sunshine and rainbows. Outside of the ray tracing issues, this GPU doesn’t hold a candle to the RTX 5070 Ti for being a multipurpose card to facilitate AI and prosumer workloads. On top of that, FSR 4 is cool and all, but it’s definitely more finicky to use in AMD’s software compared to DLSS 4 in the Nvidia app.

But if you’re solely focused on gaming and you want to maximize your horsepower in this area specifically while not breaking the bank, the RX 9070 XT is a great buy.

AMD Radeon RX 9070 XT: Cheat Sheet

What is it? This is a higher mid-range PC graphics card.
Who is it for? This is for PC gaming enthusiasts who are looking for stellar 1440p smoothness, and great 4K gaming too.
What does it cost? MSRP starts at $599/£569. In reality, you’re looking at paying $50 more in the U.S., and £20 more in the U.K.
What do I like? This is a beasty GPU — more powerful in some of the PC gaming bestsellers than the RTX 5070 Ti at a lower price to boot. FSR 4 has also seen some impressive gains over time too.
What don’t I like? While the RX 9070 XT may have the edge in gaming, Nvidia clearly has an advantage in multipurpose usage, as AI and productivity are not as strong on AMD’s GPU. On top of that, ray tracing isn’t quite as good, and Nvidia’s software is easier to use with far better DLSS 4 reliability over FSR 4.

AMD Radeon RX 9070 XT: Specs

GPU	AMD Radeon RX 9070 XT	Nvidia GeForce RTX 5070 Ti
Price (MSRP)	$599	$749
Video memory	16GB GDDR6	16GB GDDR7
Ray Tracing Cores	64 Ray Accelerators	70 4th Gen cores
AI/Tensor Cores	128 AI Accelerators	280 5th Gen cores
Shader Units / Cores	4,096	8,960
Power consumption (TDP)	304W	300W
Ports	1x HDMI 2.1b, 3x DisplayPort 2.1a	1x HDMI 2.1b, 3x DisplayPort 2.1b

AMD Radeon RX 9070 XT: 3DMark tests

(Image credit: Future)

AMD Radeon RX 9070 XT: The ups

Just a short while ago, I called the RTX 5070 Ti the beacon of mid-range gaming. Now, well, I take that back.

A gaming monster

(Image credit: Future)

It’s a price-to-performance thing when you look at it, and AMD’s Radeon RX 9070 XT either comes close or beats the $150 more expensive RTX 5070 Ti in key games. Of course, there’s DLSS 4 and multi-frame gen which give Nvidia a leg up in some situations and there are other areas where Team Green stays ahead.

But looking at the math, you’re getting around 95% of the performance of the 5070 Ti at 80% the cost. The dollars and cents make sense here.

Raw rasterization performance (1440p)
GPU	Cyberpunk 2077 ray tracing ultra 1440p	Forza Horizon 5 max settings 1440p	Call of Duty max settings no frame gen 1440p	Black Myth: Wukong Cinematic 1440p
RX 9070 XT	30.2 FPS	190.8 FPS	173 FPS	62 FPS
RTX 5070 Ti	55.5 FPS	196 FPS	169 FPS	86 FPS
RTX 5070	38.2 FPS	158 FPS	139 FPS	84 FPS
RTX 5060 Ti	32.8 FPS	121 FPS	80 FPS	45 FPS

As you can see, you’ve got three different scenarios here: either AMD’s GPU is a bit far off because of the ray tracing requirements (more on that later, but turn RT down to low in these games), it comes damn close or it either surpasses it.

(Image credit: Future)

It’s real sweet spot is at 1440p, but go even further and you’d be surprised at just what this is able to do at 4K — even more so when you turn on FSR 4. For those uninitiated, the 4th generation of FidelityFX Super Resolution is AMD’s version of Nvidia’s DLSS. It’s an upscaler that uses machine learning to look at what’s going on at a lower resolution, then make everything sharper through neural trickery to deliver faster frame rates.

Raw rasterization performance (4K)
GPU	Cyberpunk 2077 ray tracing ultra 4K	Forza Horizon 5 max settings 4K	Call of Duty max settings no frame gen 4K	Black Myth: Wukong Cinematic 4K
RX 9070 XT	25.7 FPS	105.4 FPS	92 FPS	57 FPS
RTX 5070 Ti	33.7 FPS	113 FPS	86 FPS	73 FPS
RTX 5070	17 FPS	95 FPS	67 FPS	32 FPS
RTX 5060 Ti	15.6 FPS	69 FPS	60 FPS	41 FPS

There is no multi-frame gen technically available yet (though there are some hacky ways to get it like Lossless Scaling from Steam). But ultimately, the rasterized baseline trades blows with Nvidia’s vastly more expensive GPU.

Plus there’s the promise of more to come. FSR Redstone looks prime to really make up for the ray tracing short-fall with things like neural radiance caching and ML ray regeneration, as well as frame generation too. But this is all tech coming down the line, and I don’t review products based on the promise of future updates.

Following simple standards

(Image credit: Future)

When it comes to connectivity, this may not matter if you’re getting a prebuilt PC with this GPU already installed. But if you’re upgrading your pre-existing card, it’s comforting that the power plugs are your standard 8-pin PCIe power connectors — rather than the 2x6 ports found on the RTX 5070 Ti.

Especially given the documented cases of the 2x6 12VHPWR connector melting. Of course, this has been linked to poor manufacturing or bending the cables at uncomfortable angles, but with the 8-pin, it’s certainly much less common.

AMD Radeon RX 9070 XT: The downs

This GPU does indeed go hard on gaming, and that is at the expense of a couple other use cases. Plus, there’s one thing that while improved, doesn’t match what Nvidia is capable of.

Nvidia still has some gaming advantages

(Image credit: Future)

Take a look back at those tables again. You may see a gulf in performance in Cyberpunk 2077, which comes down to my testing conditions set to RT Ultra settings. And unfortunately, ray tracing is still not quite there on the Radeon RX 9070 XT.

Turn on FSR 4 and you can see that bump right back up to over 100 FPS, which is great! But on raw rasterization, the RT cores of Team Green continue to be one step ahead.

And speaking of one step ahead, Nvidia’s game support is far more vast for DLSS 4 (over 175 games) vs FSR 4 (over 85 games). And in terms of accessing settings to force DLSS 4, Nvidia’s app seems to be more reliable, whereas AMD Adrenalin’s FSR 4 toggle is a bit more hit-and-miss.

Not for creative pros or AI workloads

(Image credit: Future)

Throwing all your eggs into the gaming basket is fair and all, but if you aimed to do more with your desktop than just play, you’ll be left a little shorthanded with the RX 9070 XT. In terms of GPU-intensive creator tasks, you’ll see AMD’s GPU fall behind the RTX 5070 Ti by 11.2%.

(Image credit: Future)

For the lower price, that could be an OK compromise, though. But it just depends where you sit on how you balance productivity and play.

Future

Then there’s the AI side of things, which…well there’s no better way to say it: Nvidia absolutely trounces AMD here.

AMD Radeon RX 9070 XT: Verdict

(Image credit: Future)

It may not be the most versatile GPU given its lacking creator workload performance, and Nvidia may have an advantage in dedicated game support and ray tracing. However, when it comes down to gaming and extracting the most performance for every dollar you spend, the RX 9070 XT is a fantastic option.

The current pricing is within the region of where I’d feel comfortable buying one without feeling overly scaled (way cheaper than the RTX 5070 Ti), and this card often dukes it out performance-wise with Nvidia’s pricier option.

Nvidia RTX 50 SUPER GPU release window narrows with new launch rumor — what we know

Darragh Murphy — Tue, 30 Sep 2025 10:14:57 +0000

Been waiting for video memory upgrades in Nvidia's rumored RTX 50 SUPER GPUs? You may have to wait a while longer than expected. As in, after CES 2026.

Nvidia's highly anticipated RTX 5080 SUPER, RTX 5070 Ti SUPER and RTX 5070 SUPER GPUs are now "most likely" to arrive between March and May in 2026, according to tech site Benchlife. This is due to industry insiders stating that AIC partners (manufacturers) have yet to receive project notices for the Nvidia RTX 50 SUPER graphics cards.

This means a specific launch time can't be given, as these partners need this project notification to land on a more accurate release time. If this is the case, then it makes sense that we'll see Team Green's upgraded GPUs launch a little later in 2026, seeing as the cards don't appear to be in development currently.

(Image credit: Future)

Rumors of Nvidia's upgraded GPUs have been swirling, despite these cards not being officially announced. They were first thought to arrive before the end of the year, with leakers recently noting a "delay" in when the graphics cards will launch. However, several reports also suggested Nvidia didn't plan on releasing its next RTX 50 SUPER series before the end of 2025.

Previously, industry insiders suggested they expect the cards to at least be revealed during CES 2026, which may still be the case (seeing as this is when Nvidia announced its RTX 50-series GPUs at CES 2025).

That said, we've heard that sources speaking to Wccftech claim that Nvidia's RTX 50-series GPUs are still in full-cycle, and that they may not even show up during the big tech event in January.

In any case, the launch of Nvidia RTX 50 SUPER GPUs is still up in the air. At the very least, we may see these GPUs in March at the earliest, if the expectation proves accurate. Plus, we have an idea of what to expect from them.

GPU	RTX 5080 SUPER	RTX 5080	RTX 5070 Ti SUPER	RTX 5070 Ti	RTX 5070 SUPER	RTX 5070
Price	$999-1,199	$999	$749-$799	$749	$549-$599	$549
CUDA Cores	10752	10752	8960	8960	6400	6144
Video Memory	24GB GDDR7	16GB GDDR7	24GB GDDR7	16GB GDDR7	18GB GDDR7	12GB GDDR7
TGP	415 Watts	360 Watts	350 Watts	300 Watts	275 Watts	250 Watts

According to rumors, we can expect a boost in VRAM from the RTX SUPER 5080, 5070 Ti and 5070 GPUs, up to a whopping 24GB of GDDR7 video memory for the 5080 and 5070 Ti and a welcome 18GB of VRAM for the 5070. That also means an increase in TGP for an uplift in power, which all translates to being able to handle demanding PC games, 3D graphics and more.

Hopefully, we won't see a huge increase in prices for these rumored GPUs either, with only a possible $50 difference in the RTX 5070 Ti SUPER at $749 and RTX 5070 SUPER at $549, if rumors ring true.

While it may be a while until we finally get our hands on Nvidia's RTX 50 SUPER GPUs, we may see other chips from Team Green arrive down the line, including the Nvidia N1X CPU and its newly announced Intel x86 RTX SoCs.

Finally! Nvidia's RTX 5060 Ti 16GB GPU is $50 cheaper than its launch price — a record low on my favorite graphics card

Jason England — Wed, 24 Sep 2025 09:42:29 +0000

The Nvidia GeForce RTX 5060 Ti has been one of my favorite GPUs this year, but I’ve only recommended it with two caveats — get the 16GB model (8GB isn’t enough) and only if it’s at (or below) retail price.

Now, those caveats are gone. For the first time, it’s dipped below MSRP: the PNY RTX 5060 Ti OC 16GB is just $379 at Walmart, $50 off its launch price.

Sitting pretty in the lower mid-range of Nvidia’s RTX 50-series of GPUs, the 5060 Ti with 16GB of VRAM gives a good balance of performance and headroom on video memory to be a beast of a 1440p graphics card. Plus, DLSS 4 and multi-frame gen massively expands the value of this GPU too with fast frame rates even at 4K.View Deal

Yes, Walmart may say you’re saving $219, but that’s on the overinflated price way over MSRP. The real news here is this is the first time in the history of this GPU that the price you pay has fallen below the launch price.

At this price, you’re getting a monster of a card. You’ll find stellar game performance at 1080p and 1440p through raw rendering, and with the AI trickery of DLSS 4 and multi-frame generation, you can see those frames start to skyrocket.

Future

Even better? You can use that AI arsenal to achieve 4K path-traced gameplay that is comfortably over 100 FPS — something that is seriously impressive in a $379 card. Now I know the initial concerns that come with that, and no, there is no telltale image ghosting that you normally see when the machine learning algorithm is trying to predict the next frames.

To those who have been waiting for the moment GPUs stop being so damn expensive, now is that time.

Intel promises Arc GPUs will survive the Nvidia deal — I'm not so sure about that

Jason England — Fri, 19 Sep 2025 10:22:21 +0000

So chances are you saw the seismic news yesterday that Intel and Nvidia announced a partnership to develop x86 chips together — including a CPU with RTX graphics built-in for PC gaming.

One of the biggest questions coming out from that is simple: does this mean that Arc is cooked? Is Intel’s own GPU efforts on the chopping block. Well, Intel told PCWorld that “the collaboration is complementary to Intel’s roadmap and Intel will continue to have GPU product offerings."

In one way, that’s a big old sigh of relief from me. But also, just because a company says one thing now doesn’t mean it won’t happen in the future, and I’m nervous.

Get on the Arc

(Image credit: Future)

For those uninitiated, Arc is Intel’s name for the company’s graphics work, and it’s been quite a journey to get to this point. Over years of rather middling GPU tech, we’re now at a point where Intel has been making some serious moves.

First is the big wins in discrete GPUs for desktop. The Intel Arc B580 is easily one of my favorite budget graphics cards — not giving you the feeling of being short-changed on video memory by giving you 12GB of VRAM for $249.

And on top of that, updates are continuing to roll in to improve XeSS: its AI frame generation and resolution scaling tech. Don’t get me wrong, it’s not close to what DLSS 4 is capable of, but it’s still strong.

Second is in handheld gaming. Only one exists at the moment: the MSI Claw 8 AI+, and at first the performance and power efficiency were a little out of whack, but the Arc graphics built into that Intel Core Ultra 7 258V have grown to be mightily impressive — giving AMD a run for its money in my own testing.

Plus, let's not forget these chips are used in laptops too, and the GPU capabilities on AAA games is surprisingly strong.

But this seems too convenient

(Image credit: Future)

Nvidia's RTX 50 SUPER GPUs looking to skip 2025 entirely — here's when they could launch

Darragh Murphy — Thu, 11 Sep 2025 10:16:57 +0000

Nvidia's RTX 50 SUPER GPUs may not arrive until 2026 after all, with a tipster noting a "delay" and reports suggesting the graphics cards were never planned to launch until 2026.

As per a post from known hardware leaker hongxing2020 on X, the rumored RTX 5080 SUPER, RTX 5070 Ti SUPER and RTX 5070 SUPER have been delayed. While no reason for this delay has been specified, if accurate, this would point to an expected reveal at CES 2026 — as with Nvidia's RTX 50-series GPUs at CES 2025.

This is despite a leak stating the SUPER GPUs may arrive before the end of 2025. Previously, tipster Moore's Law is Dead revealed that Nvidia would discontinue the base RTX 5080 and RTX 5070 Ti this October, suggesting Team Green would be making way for SUPER models to launch. If there is a delay, we may not see these graphics cards in 2025 after all.

09.10 update RTX50 SUPER series delay …🥲September 10, 2025

However, other reports state that Nvidia didn't plan on releasing its RTX 50 SUPER graphics cards before the end of the year. According to sources speaking to Wccftech, Nvidia's RTX 50-series GPUs are still in full-cycle, and SUPER variants were never planned to launch in Q4 2025.

Moreover, the report notes that the upgraded cards may not even show up at CES 2026. That said, a report from VideoCardz details that board partners expect RTX 50 SUPERs to be revealed at CES 2026.

Since Nvidia hasn't officially announced its RTX 50 SUPER GPUs, it's best to take this all with a pinch of salt. It's clear that rumors are scattered, but with CES 2026 coming up, it looks to be a likely stage for Team Green to announce its refreshed graphics cards.

When will RTX 50 SUPER GPUs arrive?

(Image credit: Future)

Following in the footsteps of its previous generations of graphics cards, it's almost certain that Nvidia will deliver an RTX 50 SUPER series of GPUs. It appears to be a matter of "when."

With the October launch rumored, it would be an early release for Nvidia's next SUPER series compared to past launches. Nvidia's SUPER models generally launch around 12 months after the release of the base models. Now, seeing as the original RTX 50-series were announced at CES 2025, it makes sense that the SUPER variants would take the spotlight at CES 2026 in January.

From there, we could see the launch of the RTX 5080 SUPER in Janaury, with the RTX 5070 Ti SUPER and RTX 5070 SUPER to follow. To note, the RTX 5080 launched on January 30, while the RTX 5070 Ti was released in February and the RTX 5070 came to shelves in early March.

Pricing and specs have also been rumored, with the big upgrade expected to be a boost in VRAM. Notably, the RTX 5080 SUPER and 5070 Ti SUPER may come with 24GB of GDDR7 video memory, and the 5070 SUPER with 18GB. This comes from leaker Kopite7kimi, with Moore's Law is Dead also leaking the same specs.

GPU	RTX 5080 SUPER	RTX 5080	RTX 5070 Ti SUPER	RTX 5070 Ti	RTX 5070 SUPER	RTX 5070
Price	$999-1,199	$999	$749-$799	$749	$549-$599	$549
CUDA Cores	10752	10752	8960	8960	6400	6144
Video Memory	24GB GDDR7	16GB GDDR7	24GB GDDR7	16GB GDDR7	18GB GDDR7	12GB GDDR7
TGP	415 Watts	360 Watts	350 Watts	300 Watts	275 Watts	250 Watts

Major RTX 50 SUPER GPU leak points to October launch — and it confirms key VRAM upgrades

Darragh Murphy — Thu, 04 Sep 2025 10:07:39 +0000

The wait for Nvidia's rumored RTX 50 SUPER GPUs may not be as long as we first thought. A new leak just revealed major details on the upgraded graphics cards — with expected pricing included.

According to reputable leaker Moore's Law is Dead (via TweakTown), RTX 5080 SUPER, 5070 Ti SUPER and 5070 SUPER GPUs are set to launch in Q4 2025. More importantly, the leak confirms video memory and performance upgrades, with up to 24GB GDDR7 VRAM on the RTX 5080 and 5070 Ti SUPERs and 18GB GDDR7 on the RTX 5070 SUPER.

These are the VRAM gains many have been expecting, which line up with a previous rumor from reliable tipster Kopite7kimi. This includes the upgrade in TGP, too, with MLID stating to expect a 9-16% performance uplift with the RTX 5080 SUPER, around 7-11% with the 5070 TI SUPER and somewhere between 8-12% in the 5070 SUPER.

Interestingly, the leak suggests Nvidia is planning to discontinue the base RTX 5080 and RTX 5070 Ti this October in favor of the SUPER variant. If this is the case, it's likely we'll see upgraded GPU models this October.

This may be due to Nvidia looking to now offer more VRAM in its higher-end GPUs, just as PC users have been wanting. As we've seen, the base models have been set at premium prices with lower-than-expected video memory. Fortunately, the leaker claims pricing may stay the same.

GPU	RTX 5080 SUPER	RTX 5080	RTX 5070 Ti SUPER	RTX 5070 Ti	RTX 5070 SUPER	RTX 5070
Price	$999-1,199	$999	$749-$799	$749	$549-$599	$549
CUDA Cores	10752	10752	8960	8960	6400	6144
Video Memory	24GB GDDR7	16GB GDDR7	24GB GDDR7	16GB GDDR7	18GB GDDR7	12GB GDDR7
TGP	415 Watts	360 Watts	350 Watts	300 Watts	275 Watts	250 Watts

As per MLID's leak, the RTX 5080 SUPER may be priced at the same $999 cost as the regular RTX 5080, and the same goes for the RTX 5070 Ti SUPER at $749 and RTX 5070 SUPER at $549. That said, the upgraded 5080 may be $200 higher, while the other two may only differ by around $50.

Nvidia has yet to officially reveal any RTX 50 SUPER models, so take this all with a pinch of salt. But it's yet another sign that Team Green's next GPUs are coming, and we may only have to wait until next month o check them out — and wave goodbye to the base RTX 5080 and RTX 5070 Ti (which may see bigger discounts with the SUPERs arriving).

RTX 5080 SUPER may come later

(Image credit: Monica J. White)

While the leak suggests an October launch at the earliest, the RTX 5080 SUPER may not be released until early 2026 instead.

As per MLID, the RTX 5070 Ti and 5070 SUPER models should arrive this year, but the 5080 SUPER may wait until Q1 2026 (between January and March) for a bigger boost in memory speeds. If it does arrive in 2026, it may come with up to 36Gbps speeds compared to the 32Gbps it's expected to have for its rumored launch.

This would fall more in line with the release dates of Nvidia's previous SUPER GPUs, which usually launch around 12 months after their base counterparts. However, if these SUPER graphics cards end up coming out in the next month or so, this would be an incredibly fast turnaround for Nvidia.

We won't know for sure until these cards are officially announced. In the meantime, check out our thoughts on Nvidia's other GPUs, including the RTX 5060 Ti.

Nvidia RTX 50 SUPER GPU rumors: Everything we know so far

Darragh Murphy — Mon, 25 Aug 2025 05:45:00 +0000

Nvidia's GeForce RTX 50-series graphics cards are now well and truly in the wild, and while it hasn't even been a year, they look to be outstaged by a new offering: RTX 50 SUPER GPUs.

Rumors of RTX 5080 SUPER, RTX 5070 Ti SUPER and RTX 5070 SUPER GPUs are popping up, signaling an upcoming release of an all-new array of graphics cards. We've been impressed by Team Green's latest GPU offerings, especially the RTX 5070 Ti, but upgraded specs for its RTX 5080 and RTX 5070 would be hugely welcome.

Already, it's almost a given that Nvidia will release a SUPER series of GPUs, following in the footsteps of its previous generations of graphics card — harking back to the RTX 20-series. We've seen the performance gains the SUPER cards deliver, but this time, we expect a crucial upgrade: video memory (VRAM).

While DLSS 4 and neural rendering do a lot of the legwork to boost frame rates well past 120 FPS at 4K, these can only go so far when cranking up demanding AAA games to max settings. That's where the extra VRAM would come in, and thankfully, Nvidia appears to be addressing this issue with its rumored RTX 50 SUPER GPUs.

But when will they arrive, and what can we expect? Let's dive into what we know so far about Nvidia GeForce RTX 50 SUPER GPUs.

Nvidia RTX 50 SUPER GPUs: rumored price and release date

(Image credit: Future)

Despite Nvidia's RTX 50-series launching at the start of 2025, rumor has it that Nvidia RTX 50 SUPER GPUs may arrive before the year ends. Or, at the very least, be announced. However, this may not be the case anymore, as rumors appear to be scattered.

Back in July, news outlet TweakTown had sources claim that the SUPER series is set to launch in Q4 2025, putting the release window sometime between October and December. That's only a few months away, and would mean the graphics cards would arrive before CES 2026, set to take place at the start of January 2026.

Now, this would be an interesting release, as Nvidia's SUPER models generally launch around 12 months after the release of the base models. Seeing as the original RTX 50-series were announced at CES 2025, and then launched at the end of January this year, the latest SUPERs would launch in a much tighter timeframe.

That's especially true of their counterparts. The RTX 5080 launched on January 30, while the RTX 5070 Ti was released in February and the RTX 5070 came to shelves in early March.

However, another major RTX 50 SUPER leak has pointed out that they will arrive in Q4 2025. According to Moore's Law is Dead, Nvidia is planning to discontinue the base RTX 5080 and 5070 Ti in favor of the SUPER models. Apparently, this is taking place in October, so it only makes sense the upgraded models arrive for then, if all of this is accurate.

However, the RTX 5080 SUPER may arrive a tad later than its siblings, which MLID states could be in early 2026.

Now, apparently, Nvidia's RTX 50 SUPER GPUs may not arrive until 2026 after all. According to hardware leaker hongxing2020 on X, the rumored RTX 5080 SUPER, RTX 5070 Ti SUPER and RTX 5070 SUPER have been delayed. While no reason for this delay has been specified, if accurate, this would point to an expected reveal at CES 2026.

09.10 update RTX50 SUPER series delay …🥲September 10, 2025

Other reports state that Nvidia didn't plan on releasing its RTX 50 SUPER graphics cards before the end of the year. According to sources speaking to Wccftech, Nvidia's RTX 50-series GPUs are still in full-cycle, and SUPER variants were never planned to launch in Q4 2025.

Moreover, the report notes that the upgraded cards may not even show up at CES 2026. That said, a report from VideoCardz details that board partners expect RTX 50 SUPERs to be revealed at CES 2026.

If I put money on it, I'd predict the RTX 50 SUPER series will be announced at CES 2026 (along with a possible Nvidia N1X CPU). But seeing as a previous rumor suggesting RTX 5080 SUPER and 5070 SUPER GPUs would arrive in 2025, as per Moore's Law is Dead, I may lose out on that cash.

Excitingly, as per MLID's leak, the RTX 5080 SUPER may be priced at the same $999 cost as the regular RTX 5080, and the same goes for the RTX 5070 Ti SUPER at $749 and RTX 5070 SUPER at $549. That said, the upgraded 5080 may be $200 higher, while the other two may only differ by around $50.

The previous RTX 40 SUPER series launched at a cheaper price than their base alternatives, so hopefully, Nvidia follows that tradition. That said, seeing as the price of RTX 50-series GPUs has gone beyond MSRP, and with the rumored release date coming soon, I doubt this will be the case.

Regardless, take this all in with a pinch of salt, as we won't know for sure until Nvidia makes the big RTX 50 SUPER GPU announcement,

Nvidia RTX 50 SUPER GPUs: specs

(Image credit: Future)

Not only do we have a rumored release date, but we also have a general idea of what to expect from the RTX 5080, RTX 5070 Ti and RTX 5070 SUPER GPUs. And all the right specs appear to be getting a boost.

As per trusted leaker Kopite7kimi, we now have a look at the rumored specs for the RTX 5080, 5070 Ti and 5070 SUPER. While there's a notable bump up in CUDA Cores (processing units designed for parallel computing), especially with the RTX 5070 Ti, and TGP, the real kicker here is the VRAM upgrade.

Many have complained about Nvidia's recent graphics cards' lack of VRAM, and the 12GB GDDR7 in the RTX 5070 says it all. With the SUPER series, however, that may no longer be an issue.

Here's what the RTX 5080, 5070 Ti, and 5070 SUPER are set to arrive with, along with a comparison with their base counterparts:

GPU	RTX 5080 SUPER	RTX 5080	RTX 5070 Ti SUPER	RTX 5070 Ti	RTX 5070 SUPER	RTX 5070
Price	$999-1,199	$999	$749-$799	$749	$549-$599	$549
CUDA Cores	10752	10752	8960	8960	6400	6144
Video Memory	24GB GDDR7	16GB GDDR7	24GB GDDR7	16GB GDDR7	18GB GDDR7	12GB GDDR7
TGP	415 Watts	360 Watts	350 Watts	300 Watts	275 Watts	250 Watts

Fortunately, the leak from Moore's Law is Dead also details these same specs, which only further proves we'll see a VRAM upgrade. So, expect up to 24GB GDDR7 VRAM on the RTX 5080 and 5070 Ti SUPERs and 18GB GDDR7 on the RTX 5070 SUPER.

To put this into perspective, many of the best PC games require demanding specs, especially when it comes to higher resolutions (4K gaming is quickly becoming more common). Cyberpunk 2077, with Path Tracing in overdrive mode, can push 16GB, while Alan Wake 2 and Indiana Jones and the Great Circle can go north of 12GB.

While DLSS 4 helps out in a wide array of supported games, and an RTX 5090 can handle whatever is thrown at it, having more options with more VRAM would definitely give gamers more options for their next GPU upgrade.

Nvidia RTX 50 SUPER GPUs: worth the wait?

(Image credit: Future / Tom's Guide)

By simply adding more VRAM, grabbing an RTX 50 SUPER GPU would be well worth the wait for those who have yet to upgrade to pick up an RTX 50-series GPU. If rumors are accurate, it will only be a few months until we see a more powerful range of graphics cards on shelves — and they should give PC users that extra inch when it comes to gaming.

Of course, it all comes down to pricing. With RTX 50-series graphics cards still being priced high, there's reason to believe the next-gen SUPER cards won't be so affordable, either. Only time will tell, but here's hoping they won't be overly past their MSRP.

So, if you've been looking to get your hands on an RTX 5080, RTX 5070 Ti or RTX 5070, it's looking like it's best to wait for their SUPER siblings to arrive. We'll keep you updated on the latest.

Here's how to update your Nvidia GPU drivers in 5 easy steps

Monica J. White — Wed, 20 Aug 2025 07:30:00 +0000

If you want smooth frame rates, stable performance, and access to all the latest features, you're going to need more than just a tip-top GPU, like the RTX 5090. You're also going to need the latest drivers.

Graphics cards get driver updates more often than most other components; it's not unusual to get at least one update per month. This is why it's important to learn how to update yours, and also how to roll them back if you run into issues.

Fortunately, updating Nvidia GPU drivers is quick and easy. We'll walk you through the entire process below. First, check your PC specs to make sure you're searching for updates for the right GPU.

How to update Nvidia GPU drivers

Whether you want to install your first Nvidia GPU driver or update your current one, the easiest way to do this is through the Nvidia App. This is a standalone program that lets you manage your GPU settings, check for driver updates, and more.

1. Download and install the Nvidia app

You can get the app for free from Nvidia's website. You'll have to make an account once you set up the app.

Once you've downloaded it, run the program and allow it to make changes to your PC. Follow the steps during the installation.

2. Find the "Drivers" tab

Launch the app and register if required. Next, in the main window, look for the Drivers tab. You'll find it on the left-hand side menu, second from the top.

3. Check for driver updates

In the Drivers tab, you can check for driver updates. The app will notify you if a new driver is available.

You can also scroll down to see the current driver or roll back to a previous version — more on that later.

For now, simply click on Download.

4. Install the driver

We recommend first closing all other programs.

The software will first download the driver. Once that's done, click on Install. Allow the app to make changes to your device when prompted by Windows.

You can now choose between Express installation and Custom installation.

We recommend using Express installation. However, Custom lets you perform a clean installation, which means restoring all driver settings to default. This is worth doing if your PC is having problems. In this option, you can also choose whether you want to install additional drivers or software.

For most users, Express installation will do just fine, so go ahead and choose that unless you have a specific reason not to.

Your screen may go black during the installation. Don't worry about it and just wait it out. Once the installation is over, you're done! You've got the latest Nvidia driver.

5. Alternative way to install your Nvidia driver

If, for any reason, you'd rather not download the Nvidia App, you can always download the latest drivers manually from Nvidia's website.

Once you're there, choose the lineup your graphics card belongs to. For example, an RTX 4070 would belong to the GeForce RTX 40 Series family. Next, pick your operating system and the language you wish to download the driver in.

Download the file and run it. The rest of the process remains the same as outlined above.

How to reinstall an older Nvidia driver

If your current driver is misbehaving, it could help to get rid of it and go back to an older version. There are a few ways to do this, but we'll outline the most reliable option below.

Keep in mind that Nvidia will let you do this in its own software.

Option 1: Use Nvidia software

In the Nvidia app, navigate to Drivers, then scroll down to the bottom. Pick the older driver and choose Reinstall.

Option 2: Use the Device Manager

Open up the Device Manager by right-clicking on the Windows icon on your Taskbar, then selecting Device Manager.

Next, click on the arrow next to Display adapters. Select your graphics card.

Navigate to the Driver tab and click on Roll Back Driver. This will bring back a previously installed version of your graphics driver.

You can also uninstall the driver first and then install a new one of your choice from Nvidia.

Option 3: Use Display Driver Uninstaller (DDU)

DDU is a free tool that lets you completely remove your existing driver and perform a truly fresh install. However, the process is a lot more complicated than the two options described above, so we recommend this one for more advanced users — or simply those who are struggling with crashes and poor performance.

Download the older driver from Nvidia directly.

Download DDU from WagnardSoft.

Disconnect from the internet to prevent Windows from auto-installing new drivers.

Boot into Safe Mode.

Run DDU in Safe Mode.

Select your GPU type, then click Clean and restart.

In regular Windows, run the driver installer you previously downloaded.

Reconnect your internet and reboot your PC.

It is possible to do this without disabling your internet connection, but chances are that Windows will automatically download drivers and potentially mess with your whole clean install.

That's it — hopefully, now you have a good grasp on how to keep your Nvidia drivers up to date. If you want to go even further, you can follow my guide on replacing your GPU directly right here.

How to update your AMD GPU drivers — follow this step-by-step guide

Monica J. White — Wed, 20 Aug 2025 06:30:00 +0000

If you're looking to get the best performance and the most rock solid frame rates from your gaming, you'll need to know how to keep your drivers updated. No matter what AMD GPU you're using, keeping them updated is crucial.

Unlike other areas of hardware, graphics cards often get pretty frequent driver updates. You can expect to see a new one roll around each month. Knowing where to look to find them, as well as how to roll back an update when you need to, is a key part of PC ownership.

Fortunately, updating AMD GPU drivers is very simple. I'll walk you through the entire process below. First, check your PC specs to make sure you're searching for updates for the right GPU.

How to update AMD GPU drivers

AMD makes some of the best processors, but it also makes superb graphics cards. The good news is that you can update drivers for both in the exact same way, and it's a super simple process. You'll be using AMD's proprietary software to get your new drivers.

1. Download AMD Software: Adrenalin Edition

Your first step is to download AMD Software: Adrenalin Edition. You might already have it on your PC if you've ever installed AMD drivers, but if not, you can download it directly from AMD here.

Click on Download. You'll be taken to a page where you have two choices: either search for your specific driver manually or download AMD Adrenalin. For example, typing in "RX 7800 XT" will show you drivers for that particular card.

To download it, click on Download Windows Drivers (or Linux, if that's what you're using).

If you'd rather just get the driver without the software, type in the name of your AMD product in the search bar below, then click Search. Pick the version of your operating system that you use and download either the Auto-Detect and Install, which is an app that'll help you install the driver, or the AMD Adrenalin, which is a more robust software.

Next, proceed with our installation instructions below.

2. Find the Drivers section in AMD Adrenalin (optional)

If you already have AMD Adrenalin installed, you should first check for driver updates directly in the app.

To do this, open up AMD Software: Adrenalin Edition, and then click on the cogwheel in the top-right corner. You'll see driver updates on the left side. Check for updates and then proceed with the installation.

3. Go through the available options

Whether you picked the Auto-Detect tool or AMD Adrenalin, the process is very similar.

AMD will show you which drivers are available to you. You can click on Additional Options to choose where to install the program and/or the driver.

You can also tick Factory Reset, although we don't recommend doing this unless you have any issues with your existing drivers. This option will pause your Windows Updates (temporarily) and delete all existing AMD drivers, then let you install them anew without Windows attempting to find a driver on its own.

4. Install your driver

Pick between Accept & Express Install and Custom Install. For most users, we recommend choosing Express, but clicking on Custom lets you decide which additional drivers you can get at the same time, such as a new chipset driver.

Once you've made your picks, click on Accept & Install.

This will either install just the drivers (in the case of the Auto-Detect tool) or the drivers in addition to AMD Software: Adrenalin Edition.

Sit tight and let your new drivers get installed. You're all done!

How to reinstall an older AMD driver

If your current driver is misbehaving, it could help to get rid of it and go back to an older version. There are a few ways to do this, but we'll outline the most reliable option below.

Keep in mind that both AMD lets you do this in its own software.

Option 1: Use AMD software

In the case of AMD, your best bet is to head to the website (as outlined above), pick the driver you want to download, and then tick Factory Reset during the installation.

Option 2: Use the Device Manager

Open up the Device Manager by right-clicking on the Windows icon on your Taskbar, then selecting Device Manager.

Next, click on the arrow next to Display adapters. Select your graphics card.

Navigate to the Driver tab and click on Roll Back Driver. This will bring back a previously installed version of your graphics driver.

You can also uninstall the driver first and then install a new one of your choice from AMD.

Option 3: Use Display Driver Uninstaller (DDU)

Download the older driver from AMD directly.

Download DDU from WagnardSoft.

Disconnect from the internet to prevent Windows from auto-installing new drivers.

Boot into Safe Mode.

Run DDU in Safe Mode.

Select your GPU type, then click Clean and restart.

In regular Windows, run the driver installer you previously downloaded.

Reconnect your internet and reboot your PC.

It is possible to do this without disabling your internet connection, but chances are that Windows will automatically download drivers and potentially mess with your whole clean install.

That's it — hopefully, now you have a good grasp on how to keep your AMD drivers up to date. If you want to go even further, you can follow my guide on replacing your GPU directly right here.

Nvidia RTX 50 SUPER GPUs may arrive sooner than you think — even before CES 2026

Darragh Murphy — Tue, 29 Jul 2025 10:13:13 +0000

Nvidia RTX 50 SUPER GPUs are rumored to be in the works, which will hopefully (and finally) fix the lack of video memory on its latest graphics cards. Now, it appears we may see them arrive soon — even before the start of 2026.

According to news outlet TweakTown, sources claim the Nvidia GeForce RTX 50 Super series is set to launch in Q4 2025, or sometime during the holiday season. This means we could see an RTX 5080 SUPER, 5070 Ti SUPER and 5070 SUPER arrive before their expected announcement during CES 2026.

This puts the release windows between October and December, despite CES 2026 kicking off at the start of January 2026. Team Green's SUPER models generally launch around 12 months after the release of the base models, so this launch window does line up.

However, at under 12 months, it would be the earliest launch of an RTX SUPER GPU, seeing as the RTX 5080 launched on January 30, while the RTX 5070 Ti was released in February and the RTX 5070 came to shelves in early March.

If accurate, we'll see stronger RTX 50 SUPER cards before the end of the year, with the RTX 5080 SUPER expected to be the first to arrive. This also falls in line with a previous rumor suggesting RTX 5080 SUPER and 5070 SUPER GPUs would arrive in 2025, as per Moore's Law is Dead.

SUPER VRAM upgrades

(Image credit: Future)

Thanks to reliable leaker Kopite7kimi, we have a hint of the rumored specs the RTX 5080, 5070 Ti and 5070 SUPER will deliver. While there's an expected increase in CUDA Cores (only with the RTX 5070 SUPER) and TGP, the real draw is the boost in video memory (VRAM).

One general complaint in Nvidia's latest graphics cards is that they lack enough VRAM, especially when it comes to the 12GB GDDR7 VRAM in the RTX 5070. However, the SUPER series is expected to deliver a big jump in video memory,

Here's a look at the rumored specs of the RTX 5080, 5070 Ti and 5070 SUPER, and how much of a boost these GPUs offer over the base RTX 50-series counterparts.

GPU	RTX 5080 SUPER	RTX 5080	RTX 5070 Ti SUPER	RTX 5070 Ti	RTX 5070 SUPER	RTX 5070
CUDA Cores	10752	10752	8960	8960	6400	6144
Video Memory	24GB GDDR7	16GB GDDR7	24GB GDDR7	16GB GDDR7	18GB GDDR7	12GB GDDR7
TGP	415 Watts	360 Watts	350 Watts	300 Watts	275 Watts	250 Watts

Nvidia wants to make 8GB GPUs great again with AI texture compression — but I'm not convinced

Jason England — Thu, 17 Jul 2025 11:35:18 +0000

If you’re annoyed by just getting 8GB of video memory (VRAM) on your Nvidia RTX 5060 Ti, RTX 5060 or RTX 5050 GPU, there may be a fix coming. And just like a lot of Team Green’s work, it’s all about AI.

In 2025, when plenty of games are requiring more than this from the jump, it’s simply not enough (and PC gamers are letting Nvidia and AMD know with their wallets). Which is why Nvidia is looking to neural trickery — it’s bread and butter with the likes of DLSS 4 and multi-frame gen.

You may already know of Neural Texture Compression (or NTC), which is exactly what it says on the tin: taking those detailed in-game textures and compressing them for efficiency of loading and frame rate.

As WCCFTech reports, NTC has seemingly taken another giant step forward by taking advantage of Microsoft’s new Cooperative Vector in DirectX Raytracing 1.2 — resulting in one test showing an up-to-90% reduction in VRAM consumption for textures.

To someone who is always wanting to make sure people get the best PC gaming bang for their buck, this sounds amazing. But I’m a little weary for three key reasons.

Only in Nvidia’s house

As you can see in tests run by Osvaldo Pinali Doederlein on X (using a prerelease driver), this update to make the pipeline of loading textures more efficient with AI is significant. Texture size dropped from 79 MB all the way to just 9 MB — dropping the VRAM consumption by nearly 90%.

How does it perform? Disabling v-sync, RTX 5080, demo at the startup position: (explained next tweet)Default: 2,350fps / 9.20MBNo FP8: 2,160fps / 9.20MBNo Int8: 2,350fps / 9.20MBDP4A: 1,030fps / 9.14MBTranscoded: 2,600fps / 79.38MBJuly 15, 2025

Just like DLSS 4 and other technologies extracting a higher frame rate and better graphical fidelity out of RTX 50-series GPUs, NTC requires developers to code it in.

And while Nvidia is one of the better companies in terms of game support for its AI magic (so far over 125 games support DLSS 4), it’s still a relatively small number when you think of the many thousands of PC titles that launch every year.

Tested very specifically

(Image credit: Future)

Of course, this is not a burn on Doederlein here. This testing is great! But it is one example that doesn’t take into account the broader landscape of challenges that are faced in a game — a test scene of a mask with several different textures isn’t the same as rendering an entire level.

So while this near-90% number is impressive nonetheless, when put to a far bigger challenge, I anticipate that number will be much lower on average. But when it comes to 8GB GPUs, every little bit helps!

8GB still has an expiry date

(Image credit: Future)

So yes, on paper, Nvidia’s NTC could be the savior of 8GB GPUs, and it could extract more value from your budget graphics card.

But let’s address the elephant in the room — graphics cards with this low amount of video memory have been around for years, games in 2025 have proven that it’s not enough and neural texture compression looks to me like a sticking plaster.

I don’t want to ignore the benefits here, though, because any chance to make budget tech even better through software and AI is always going to be a big win for me.

But with the ever-increasing demands of developers (especially with Unreal Engine 5 bringing ever-more demanding visual masterpieces like The Witcher 4 to the front), how far can AI compression really go?

AMD Ryzen Z2 Extreme benchmarks revealed — here's how the MSI Claw A8 and ROG Xbox Ally X may perform

Darragh Murphy — Wed, 16 Jul 2025 10:13:04 +0000

Gaming handhelds are about to see a major boost in gaming performance thanks to AMD's new flagship chip, the Ryzen Z2 Extreme, and we now have a taste of what to expect thanks to new benchmarks.

The first AMD Ryzen Z2 Extreme benchmarks are in on Geekbench, as spotted by X user Olrak29_, showcasing the performance we can expect in the upcoming MSI Claw A8 — both in CPU and GPU performance.

In the Geekbench 6 CPU results, the Ryzen Z2 Extreme-equipped MSI Claw A8 delivered a 2,748 single-core score, along with a 12,182 multi-core score. This is a strong improvement over its predecessor, the Ryzen Z1 Extreme, with around a 27% boost overall compared to other Geekbench benchmarks (via Wccftech).

(Image credit: Geekbench)

We expected a major boost in performance, with AMD's Ryzen Z2 Extreme coming with 8 cores and 16 threads based on Zen 5 architecture for faster multitasking, 15W to 35W power output and 16 GPU cores from its Radeon 890M iGPU with AMD's RDNA 3.5 to deliver FSR and frame-gen tech to boost frame rates.

Speaking of graphics, the GPU performance is also shown on Geekbench, with a Vulkan score of 45,064 and an OpenCL score of 37,970. As per the Geekbench results, this pushes past the AMD Ryzen 9 HX 370 (44,399/37,297) and crushes the Z1 Extreme (34,715/29,632).

According to the results shown on Geekbench, here's how it measures up against the previous Ryzen Z1 Extreme APU in the Asus ROG Ally X, along with the laptop equivalent AMD Ryzen AI 9 HX 370 in the Asus TUF Gaming A14 in our testing.

Devices	Geekbench 6 Single core	Geekbench 6 multi-core
MSI Claw A8 (AMD Ryzen Z2 Extreme)	2748	12182
Asus ROG Ally X (AMD Ryzen Z1 Extreme)	2565	11085
Asus TUF Gaming A14 (AMD Ryzen AI 9 HX 370)	2863	12729

As you can see, this shows a smaller boost in CPU performance (a 6-9% boost), but given the Z2 Extreme APU is made for handheld gaming, this is still a solid uplift, and the GPU power it appears to offer will make this chip a beast for gaming on the move. Plus, it isn't certain if this ran on 15W or 30W power, and the MSI Claw A8 tested appears to use 16GB of DDR5 RAM as opposed to the 24GB it can go up to.

Regardless, the improvements are clear, and this also gives us a look at what the ROG Xbox Ally X will deliver, seeing as Microsoft's upcoming handheld comes with the same Ryzen Z2 Extreme APU.

Top-tier performance at a price

(Image credit: Future)

The RTX 5090 is the best graphics card I've ever owned — but there's a catch for living room PC gamers

dave.meikleham@futurenet.com (Dave Meikleham) — Mon, 14 Jul 2025 07:30:00 +0000

Just as Terminator 2: Judgment Day predicted back in ye olden days of 1991, the future belongs to AI. That could be a problem for Nvidia RTX 50-series GPUs, even when it comes to the best consumer graphics card money can buy.

I was ‘fortunate’ enough to pick up an Nvidia GeForce RTX 5090 a couple of months ago. I use those semi-joking apostrophes because I merely had to pay $650 over MSRP for the new overlord of GPUs. Lucky me.

Before you factor in the 5090’s frame-generating AI voodoo (which I’ll get to), it’s important to give credit to Team Green for assembling an utter beastly piece of silicon. Around 30% more powerful than the RTX 4090 — the previous graphics card champ — there’s no denying it’s an astonishing piece of kit.

Whether you’re gaming on one of the best TVs at 120 FPS or one of the best gaming monitors at 240 fps and above, the RTX 5090 has been designed for the most ludicrously committed hardcore gamers. And wouldn't you know it? I just happen to fall into this aforementioned, horribly clichéd category.

My setup

(Image credit: Future)

So I have a PC similar to the build our lab tester Matt Murray constructed (he even posted a handy how-to on building a PC) — packing the 5090, AMD Ryzen 7 9800X3D, and 64GB DDR5 RAM on a Gigabyte X870 Aorus motherboard.

In terms of the screens I play on, I have two. For the desk, I've got an Samsung Odyssey G9 OLED with a max 240Hz refresh rate, but most of the time, I'll be in living room mode with my LG G3 OLED’s max 120Hz refresh rate.

Frame game

(Image credit: Future)

The main selling point of Nvidia’s latest flagship product is DLSS 4’s Multi Frame Generation tech. Taking advantage of sophisticated AI features, Nvidia’s RTX 50 cards are capable of serving up blistering frame rates that simply can’t be achieved through brute force hardware horsepower.

Multi Frame Generation — and I promise that’s the last time I capitalize Team Green’s latest buzz phrase — feels like the biggest (and most contentious) development to hit the PC gaming scene in ages. The tech has only been out for a few months and there are already over 100 titles that support Nvidia’s ambitious AI wizardry.

How does it work? Depending on the setting you choose, an additional 1-3 AI-driven frames of gameplay will be rendered for every native frame your GPU draws. This can lead to colossal onscreen FPS counts, even in the most demanding games.

Doom: The Dark Ages, Cyberpunk 2077, Indiana Jones and the Great Circle and Half-Life 2 RTX — some of the most graphically intense titles around can now be played at incredibly high frame rates with full ray tracing engaged. That’s mainly thanks to multi frame generation.

So I got into the games (turn off Vsync for the best results). For more specific context, these figures were taken from Doom’s Forsaken Plain level, Indy’s Marshall College section during a particularly challenging path traced scene, driving around downtown Night City in Cyberpunk, and Gorden’s mesmerizing new take on Ravenholm.

All games tested at 4K (Max settings, DLSS Balanced)	Cyberpunk 2077	Doom: The Dark Ages	Indiana Jones and the Great Circle	Half-Life 2 RTX demo
Frame gen off (Average frame rate / latency)	58 FPS / 36-47 ms	95 FPS / 37-48 ms	85 FPS / 33-40 ms	75 FPS / 26-3 ms
Frame gen x2 (Average frame rate / latency)	130 FPS / 29-42 ms	160 FPS / 51-58 ms	140 FPS / 35-46 ms	130 FPS / 29-42 ms
Frame gen x3 (Average frame rate / latency)	195 FPS / 37-52 ms	225 FPS / 54-78 ms	197 FPS / 43-53 ms	195 FPS / 37-52 ms
Frame gen x4 (Average frame rate / latency)	240 FPS / 41-60 ms	270 FPS / 56-92 ms	243 FPS / 44-57 ms	240 FPS / 41-60 ms

These are ludicrous frame rates — limited only by either my LG G3 OLED’s max 120Hz refresh rate, or even the sky high 240Hz on my Samsung Odyssey G9 OLED in a couple circumstances.

There is a catch, though, which goes back to the ways that I play.

A mixed lag

(Image credit: Future)

Despite my frame rate counter showing seriously impressive numbers, the in-game experiences often don’t feel as smooth as I expected.

As much as I’ve tried to resist, I’ve become increasingly obsessed with the excellent Nvidia app (and more specifically) its statistics overlay while messing around with multi frame gen of late. These stats let you monitor FPS, GPU and CPU usage, and most crucially for me, latency.

Also known as input lag, latency measures the time it takes a game to register the press of a button on one of the best PC game controllers or the click of a key/mouse in milliseconds. If your latency is high, movement is going to feel sluggish, regardless of how lofty your frame rate is.

And that situation is compounded on my TV. The high frame rate is glorious on my monitor, but when locked to 120Hz, you don't get the perceived smoother motion of those additional frames — creating a disconnect that makes that latency a bit more noticeable.

My advice to TV gamers

(Image credit: CDPR)

If you own one of the best gaming PCs and want to enjoy a rich ray traced experience with acceptable input lag at responsive frame rates on your TV, my advice would be to aim for the frame gen level that is as close to your maximum refresh rate as possible.

For all the games I tested, that would be 2x. At this level, I find latency hovers around the mid 30s but never exceeds 60 ms, which feels as snappy in that kind of living room gaming setup.

Crank up the multi frame gen set to either x4 or x3 setting, and there's a depreciation of what you get here, as the latency becomes more visibly prevalent at the restricted refresh rate using one of the best gaming mice.

Flip to a 240Hz monitor, however, and the difference is night and day, as the latency remains at a responsive level alongside those AI-injected frames for a buttery smooth experience.

Path life

(Image credit: Future/Machine Games/Bethesda)

And now, we've got to talk about path tracing — it's already blowing minds in Doom: The Dark Ages, and it's prevalent in the likes of Cyberpunk and Doctor Jones' enjoyable romp. It's essentially the ‘pro level’ form of ray tracing, this lighting algorithm can produce in-game scenes that look staggeringly authentic.

Given the demands of this tech on your GPU, the most graphically exciting development in PC gaming for years will most likely demand you use DLSS 4’s x4 or x3 AI frame-generating settings to maintain high frame rates in future implementations.

I wasn’t surprised that path tracing floored me in CD Projekt Red’s seedy yet sensational open-world — I was messing with its path traced photo mode long before DLSS 4 arrived. The quality of the effect cranked to the max in The Great Circle knocked my socks off, though.

That stunning screenshot a few paragraphs above is from the game’s second level, set in Indy’s Marshall College. During a segment where Jones and his vexed buddy Marcus search for clues following a robbery, path tracing gets to really flex its muscles in a sun-dappled room full of antiquities. Dropping down to the highest form of more traditional ray tracing, I was genuinely shocked at just how much more convincing the path traced equivalent looked.

So while the technology matures, I hope Nvidia continues to work to reduce latency at these middle-of-the-road frame rates too, so that this AI trickery really hits the spot when maxed out.

(Image credit: Future)

To be clear to those on the ropes about buying an RTX 5090 — just as we've said in our reviews of the RTX 5060 Ti, 5070 and 5070 Ti, if you own a 40 series-equivalent GPU, you should stick with your current card.

You may not get that multi-frame gen goodness, but with DLSS 4 running through its veins, you still get the benefits of Nvidia’s latest form of supersampling and its new Transformer model — delivering considerably better anti-aliasing while being less power-hungry than the existing Legacy edition.

I don’t want to end on a total downer though, so I’ll give credit where its due. If you’re on a monitor with a blisteringly refresh rate though, I’ll admit multi frame generation might be a good suit for your setup.

My fondness for the RTX 5090 is only matched by Hannibal Lecter’s delight in chowing down on human livers. But for those who hot switch between the desk and the couch like I do, make sure you tweak those settings reflective of your refresh rate.

How to replace your PC GPU — 16 easy steps to upgrade your desktop quickly and easily

Monica J. White — Sun, 06 Jul 2025 12:00:00 +0000

If you own a PC, the day will eventually come when you'll need to replace your graphics card (GPU), Maybe you just got your hands on Nvidia's beastly RTX 5090, or you made a more mid-range upgrade to the likes of the RTX 5060 Ti or the AMD Radeon RX 9060 XT.

One way or another, knowing how to take out your GPU and install a new one will come in handy, but it can be an intimidating process if you've never had to do it before.

It's been nearly 20 years since I built my first PC, so suffice it to say I've installed my fair share of GPUs. Below, I'll tell you everything you need to know and walk you through the process of replacing a graphics card quickly and easily.

Things to do before replacing your GPU

There are a few things to be mindful of before you're elbow-deep inside your PC case. Ideally, before you take out your old GPU, check out the following.

1. Check your power supply

Before you uninstall your existing GPU, make sure that your power supply (PSU) can support your new graphics card. If you’ve already researched compatibility while shopping for your GPU, you’re probably good to go — but it never hurts to double-check.

Your PSU needs to meet the wattage requirements of your graphics card. The easiest way to check whether it does is to look up your exact GPU model online, as manufacturers list recommended PSU wattage based on the card's power draw. Keep in mind that the rest of your PC needs to draw power from the same PSU, too, so GPU power requirements don't equal PSU wattage — you'll need headroom to account for your CPU, motherboard, fans, storage, and more.

Next, confirm that your PSU supports the correct power connectors for your new GPU. These are determined by the number of pins on the connector:

6-pin and 8-pin (6 + 2), common in older GPUs
Some GPUs require dual 8-pin (2 x 8), dual 6-pin, or a combination
Newer Nvidia GPUs use the 12VHPWR (16-pin) or 12V-2x6 (12-pin) connector, both of which may require an adapter (often bundled with the GPU)

Not sure which PSU is installed in your system? The easiest way is to power everything down, open up your case, and look for the label on your power supply, as well as scoping out the connectors. I'll talk about how to safely open your PC case below.

2. Other important checks

You've most likely checked this before shopping, but again, before you get rid of your existing GPU, check that the new one will fit with the rest of your system.

This includes:

Case size and clearance, including power connector (it should fit without bending).
Monitor compatibility, including whether you have the right cables and/or adapters.
PCIe compatibility — this is less of an issue, as PCIe is backward compatible, so don't worry if you have an older motherboard.

3. Things to prepare beforehand

Before you open up your PC case and start meddling with its contents, it's important to set the stage and prepare a few things:

Clear some room on your desk or on an otherwise clean, dust-free, non-static surface. Avoid setting your PC down on carpet; if you have no other option, put it down on a cardboard box instead.
If you're installing a big, heavy GPU, make sure that you have an anti-sag brace or support bracket.
Buy an anti-static wrist strap, if possible.
Grab a Philips screwdriver, ideally a magnetic one to avoid dropping screws into your case.
Prepare an anti-static bag to store your old graphics card.

With all that out of the way, it's time to remove your old GPU.

How to remove your old GPU

With all the pre-flight checks complete, it's time to roll up your sleeves and remove your old GPU. But first, drivers.

Hint: We also have a complete guide to building a PC if you need help picking your components as well as putting the build together.

1. Uninstall your drivers

Start by uninstalling the drivers for your current graphics card. This gives your new GPU a fresh start and avoids driver conflicts.

You can do this by going into the Device Manager (right-click on the Windows icon on the taskbar and select it). Locate Display adapters, find your GPU in the list, and right-click on it. Next, select Uninstall device. If prompted, delete the driver.

Alternatively, I recommend using Display Driver Uninstaller (DDU) to ensure a completely clean uninstall. For a more thorough cleanup, use it in Safe Mode.

2. Power down your computer and unplug everything

Turn off your PC. Next, flip the power switch on your PSU at the back of your PC to power it down completely.

Unplug all the cables, as well as things like USB hubs or Bluetooth adapters. Be gentle and remember that some of them can't just be yanked out; for example, the Ethernet cable needs to be pressed down on.

3. Open up the side panel

Place your PC on its side on a flat, dust-free, non-static surface. As mentioned previously, the rug won't do the trick — make sure there's something in between your PC and the carpet if you're using one.

Ground yourself, either with an anti-static wrist strap or by touching the metal part of your case with both your hands.

Time to remove the side panel. Depending on your case, this involves unscrewing regular screws or thumbscrews, or in newer cases, simply pressing down on a latch and pulling the side panel off.

4. Check your power connectors

This is just a callback to the previous section.

If you need to verify whether your power connectors align with what your GPU needs, take a look at your power supply, seated somewhere at the bottom of your case. In the above image, the PSU is hidden away in an enclosure at the bottom of the chassis.

Verify whether you've got all the connectors you need for your new GPU. You might need to use an extra PCIe power connector if not, or an adapter if your PSU is older and your GPU is a new Nvidia model from the RTX 40-series or 50-series.

5. Locate your GPU

You won't need your glasses to locate the GPU — it's the most prominent component attached to your motherboard. It's attached to the PCIe slot and screwed in near your PC's back panel.

Take note of where all the screws are, whether there's an additional anti-sag brace, and where the PCIe latch is.

The PCIe latch is the easiest part to miss, but it's really important. It's a small, usually black, piece of plastic underneath your GPU.

6. Disconnect and unscrew the GPU

Ground yourself again (never hurts to be extra sure) and disconnect everything from your GPU. This mainly means the power connector that leads to the power supply. Press down on the clip (don't force it) and pull the cable straight out.

There might be some more cabling, such as ARGB headers for lighting. Disconnect all of those.

Next, grab your screwdriver and unscrew the GPU from the bracket that it sits in. You'll find all the screws near the back of the case (or in some cases, outside it), close to your display ports.

If you have a magnetic screwdriver, this won't be a problem. If not, a good idea is to start with your screwdriver and finish with your thumb in order to avoid dropping the screws inside the case, which is supremely annoying.

If you're using a large, high‑end GPU like an RTX 5080 or RTX 5090, it may include an anti‑sag bracket or GPU brace. Unscrew this, too — but usually, it can stay in your case. Your next GPU will likely fit into the same bracket.

7. Remove the graphics card

All you need to do now is go right back to that PCIe latch that you found before, somewhere beneath the GPU. It's a small, black piece of plastic.

You'll either need to press it or slide it, depending on the motherboard.

Remember: Don't force it. If the latch isn’t fully released, the GPU won’t budge — and that’s by design to prevent damage.

Be ready to hold up your GPU with your other hand as you'll release it from the PCIe slot.

8. Store your GPU safely

Now is your chance to gently clean off any dust from your old GPU (if you want to). If using compressed air, make sure to hold the fan blades to stop them from spinning, which can damage the bearings.

Put your GPU in an anti-static bag and store it somewhere safe.

How to install your new GPU

We're almost done! It's time to give your new GPU a new home. Here's how to install your GPU.

1. Plug your new GPU into the PCIe x16 slot

You know what I'm going to say: Ground yourself, yet again, by touching the metal part of your case.

Next, grab your new GPU and line it up with the topmost PCIe x16 slot, and then gently slide it in. You should hear or feel a click once the PCIe latch locks the card firmly in place.

Make sure that you have plenty of room around the graphics card for things like M.2 ports or the power connector. Remember: You shouldn't bend it, as that can end in disaster on some GPUs; the RTX 4090 had a lot of reported cases of the power connector melting due to excessive bending.

2. Screw the GPU in at the rear bracket

With one hand supporting the GPU, begin securing it to the case with the screws at the rear bracket, near the display outputs.

This is the same process as before, but in reverse, where you'll attach the GPU to the motherboard. You'll often have to use one, two, or even more screws, so keep supporting the GPU until it sits safely in the bracket.

If your model came with a dedicated anti-sag brace, now is the time to install it according to the manufacturer's instructions. You'll always do this after installing the graphics card into the PCIe slot first to make sure that the bracket aligns with its position and height.

There are a few different types of support brackets, but they're all fairly similar to install. Slide the bracket in from beneath the back end, opposite the PCIe slot.

The contact point should be under the backplate, not under the fans or PCB components. Some might have adjustable height, so adjust if needed, and secure it properly.

3. Plug the power connectors in

Depending on the graphics card, you might need one to four PCIe power connectors. In the case of newer Nvidia cards, you might also need an adapter to connect your PSU to the 12VHPWR or 12V-2x6 connector on the GPU.

Carefully plug all the required connectors into your graphics card. Be firm but gentle — you should hear or feel a click when each one is fully seated. If using an adapter, plug it into the GPU first, then connect all required PCIe power cables from the PSU into the adapter.

Warning: Avoid bending the adapter cable as much as possible.

4. Close your side panel and plug everything back in

With your GPU firmly in place, it's time to reattach the side panel. Depending on your case, you can just slot it in and leave it like that once it clicks into place, or you may need to deal with screws or thumbscrews.

Next, plug everything back in. Remember to plug your monitor(s) into the GPU and not the motherboard, and don't forget the small bits and bobs, such as USB hubs.

Flip the switch at the back of your PC, and then power your computer back on.

5. Install your new drivers

We're at the finish line. It's time to install your drivers.

The easiest way to do this is to use the proprietary software provided by each GPU manufacturer:

Alternatively, you can download the drivers manually. This makes for a less streamlined installation process, though. Here are the direct links:

Once you've got your drivers, it's time to power up your favorite game and enjoy your new GPU! Next step: Benchmark your GPU. Have fun!

Latest from Tom's Guide AU in Gpus

I gamed for 250 hours on AMD Radeon RX 9070 GRE — it’s a $549 nightmare for Nvidia’s RTX 5060 Ti

AMD Radeon RX 9070 GRE: Cheat Sheet

AMD Radeon RX 9070 GRE: Specs

AMD Radeon RX 9070 GRE: 3DMark tests

AMD Radeon RX 9070 GRE: The ups

Challenging the gaming throne

FSR Redstone is getting better

AMD Radeon RX 9070 GRE: The downs

Nvidia keeps the compatibility advantage

Productivity performance is lacking

AMD Radeon RX 9070 GRE: Verdict

Update your Nvidia GPU drivers now to protect your PC from 9 "high-severity" vulnerabilities — here's what's at risk

Keep your PC safe from malware

More from Tom's Guide

The end of unlimited AI: Why Google’s Gemini leak is a warning for every power user

What actually changed with Gemini

The real reason this is happening

The takeaway

More from Tom’s Guide

I stress-tested Pragmata’s path tracing — it looks like a dream and Nvidia’s DLSS 4 makes it play like one (provided you use it right)

What settings should you use?

By the numbers

Latency checker

Path traced sci-fi fun

More from Tom's Guide

I tested DLSS 4.5 Dynamic Multi Frame Generation: It’s not just AI trickery, it’s Nvidia’s cheat code for 4K 240Hz and a true console killer

What is Dynamic Multi Frame Generation?

By the numbers

To my eyes

TV time

Verdict: a console killer?

More from Tom's Guide

Is the RAM price crisis finally ending? I’m starting to see cracks

Uncle Sam owes Big Tech $130 billion — and you might get the change

🤔 Skeptical take: Lower ceiling, but higher floor?

The AI Tax is getting a divorce from your gaming rig

🤔 Skeptical take: A technicality?

The DeepSeek effect continues

🤔 Skeptical take: Jevons paradox

OK, time for Plan B

🤔 Skeptical take: The quality lottery

An omen, not a guarantee

More from Tom's Guide

What is Nvidia DLSS? Explaining the AI tech transforming PC gaming — and why it’s getting controversial

What is DLSS?

How to use DLSS the right way

My Super Resolution preferences

My Frame Generation preferences

DLSS 5 controversy explained

Outlook

More from Tom's Guide

I pixel-peeped DLSS 5 — and now I can’t tell if Nvidia just changed gaming or broke it with AI

What is DLSS 5? Let’s explain it with pizza

Bridging the uncanny valley

I’ve got some questions

DLSS5 outlook

More from Tom's Guide

Nvidia says PC gaming will ‘look like a film’ — GTC 2026 could herald 1 million times better path tracing for GPUs

What is Path Tracing?

The death of brute force

Neural rendering is the whole thing

GTC 2026 is the starting pistol

More from Tom's Guide

Nvidia is bringing a killer app to the Apple Vision Pro — and it's good news for sim racers

What is Nvidia’s CloudXR?

The Vision Pro’s killer app?

Not the only XR upgrade

More from Tom's Guide

I went eyes-on with Nvidia’s DLSS 4.5 Dynamic Multi-Frame Generation launching March 31 — and it’s a game-changer

How does Dynamic Multi-Frame Generation work?

Gamesmaxxing

Outlook

More from Tom's Guide

Nvidia GTC 2026: The biggest reveals we expect to see

Nvidia’s mystery chip

The Arm-based N1X CPU

Next-gen GPU tease

Physical AI and robotics

Agentic AI