Nvidia GTC October 2025 — Self-driving Ubers, 6G with Nokia and an AI supercomputer for the Department of Energy
This GTC was all about partnerships
Nvidia GTC is back in the US, and CEO Jensen Huang just finished delivering his keynote. Just like we saw in San Jose, Taiwan and Paris, Team Green are very much at the front of the Agentic AI explosion we're seeing, and a key part in America winning the AI arms race.
And its plan to do so is with a whole lot of exciting partnerships, including with Nokia to drive AI-native 6G connectivity, Foxconn to build an autonomous robot factory building GPUs in Texas, and even Uber with the tech to build a suite of self-driving robotaxis.
Nvidia GTC: every big announcement
- Nvidia is partnering up with Nokia: In a strategic partnership between the two, Nvidia is introducing a new Aerial RAN computer, which is a 6G-ready telecommunications platform that can support AI services directly infused into the connectivity. This means more stable, efficient connectivity, and far faster data transfers that could power the future of AI on cell networks.
- Nvidia and Oracle to build the U.S. Department of Energy's largest AI supercomputer: 110,000 Blackwell GPUs in total will lead the way to accelerate scientific discovery, and speed up research and development in America.
- Nvidia works with Uber to get ready for robotaxis: Nvidia's self-driving platform is coming to Stellantis, Lucid and Mercedes-Benz vehicles to give them robotaxi capabilities. But on top of that, Nvidia is working with Uber to launch an autonomous fleet of 100,000 robotaxis starting in 2027.
How to watch Nvidia GTC 2025
You can watch the keynote back here.
Nvidia GTC 2025: Latest updates
How to watch Nvidia's event
Welcome to the Nvidia GTC live blog! I'll be taking you through all the announcements Jensen has to share in his keynote — kicking off today at 9am PT / 12pm ET / 4pm GMT.
If you want to watchalong, the stream is ready and waiting for the show to begin. And if watching the keynote doesn't sound like how you want to spend your afternoon, follow along with our live blog.
It seriously is that easy. After all, Nvidia wants you to watch its event, so it's in the company's best interest to make watching as painless as possible.
The company's YouTube description sounds mightily similar to other GTCs: "Watch NVIDIA CEO Jensen Huang’s GTC keynote to catch all the announcements on AI advances that are shaping our future."
Current lowest prices on Nvidia RTX 50-series GPUs
While we're watching the pre-amble and interviews with key AI and infrastructure companies that rely on Nvidia, let's bring this back to you. If you've been looking to grab one of the best GPUs for your PC — be it for gaming or AI work — pricing has seemingly stabilized in the UK and US.
There's one particularly high-priced exception (looking at you, RTX 5090), but everything else seems OK at the moment.
Previously at GTC
Back in March, Jensen Huang really laid out the roadmap across AI, robotics and accelerated compute. Here's what was spoken about a few months ago.
- Nvidia Isaac Groot N1: A new model for robotics, Groot N1 is set to be "the world’s first open Humanoid Robot foundation model."
- "Blue," Robotics and AI collaboration: Powered by a new physics engine to simulate robotic movements, Newton was introduced, showcasing a little robot named Blue. It's a collaborative robotics project with Disney Research and Google DeepMind. An open-source model of Newton will arrive later in 2025.
- Blackwell Ultra AI chips: New chips to be released later this year, aimed at meeting the growing demand for computational power in AI.
- Vera Rubin Architecture: The next step beyond Blackwell, Vera Rubin will increase bandwidth and perform even faster. It's set for release in late 2026, with Vera Rubin Ultra expected in 2027.
- Nvidia Dynamo: Nvidia's new open-source software system designed to scale AI models effectively, and it's able to customize a data center far more effectively.
- Self-driving cars: Nvidia is teaming up with GM to create autonomous cars, developing custom AI systems for autonomous vehicles and putting "AI in the car."
Expect updates across all of these at the event today.
Have you tried Nvidia GeForce Now's upgrade yet?
Is cloud gaming going to be talked about at Nvidia GTC? No. This is going to be more focused on AI and robotics, but I have to admit, GeForce Now with RTX 5080 is a real showcase of what Nvidia's cloud computing platform is capable of.
With up to 5K 120FPS gameplay and latency that is lower than your PS5 Pro in your living room, it seemingly solves a lot of the key issues of cloud gaming. Take a look at my video review.
Does the teaser trailer reveal anything?
It's only a brief teaser, but we can glean a couple of things from this. There's a lot of showcases of the ways Nvidia's GPU power is being used, so expect plenty of demonstrations around this.
Interestingly, there's also peaks at both physical and digital AI too — both in virtual chatbots and robotics. These are two key areas that Nvidia has been focusing its infrastructure on.
Are we in an AI bubble?
If you ask CNBC's Kristina Partsinevelos as she walks the show floor at Nvidia GTC, the answer is seemingly "no." But she does discuss some of the other narratives around AI such as job displacement (Amazon just cut 14,000 jobs in favor of AI today).
But when you take a look at the virtuous circles of vendor financing happening between the likes of Nvidia, OpenAI, AMD, Oracle and more — all with outside investment continuing to pour in on the promise of AI.
It made me think one day while watching Stephen Hawking pull off a sick 900 in a half-pipe via Sora 2, that maybe...just maybe this seems a little bubble-ish. After living through the dot-com bubble, I'm starting to feel nervous.
Jensen just interrupted to answer this question!
Well this is unexpected! Usually we have to wait for the keynote for Jensen Huang to appear. But he's just walked into an interview happening where the conversation had steered to job displacement. Translation: will we lose our jobs to AI?
In talking about how AI can increase productivity, which makes a company more money to reinvest and hire more people, he just said something that I feel is going to be important to keep in mind going forward: “You’re not going to lose your job to a robot, you’re going to lose your job to somebody who uses a robot.”
Nvidia's AI factories are at the center of this strategy — taking into account plans to make AI token generation more efficient and more performant from its chips.
However, in my mind, this skips over one sticking point that I don't have the answer to. A company can make money in two ways: being more productive and streamlining the outgoing costs. If AI gives companies the ability to do both, why would they hire more people when they can just make more money?
Huang's all over the shop!
The CEO of Nvidia has just moved over to the panel discussing quantum computing and is currently reminding the panelists to hydrate while providing water to everyone!
He's talking about how quantum computing can help solve a lot of sticky problems, and what the future looks like when quantum becomes more of a reality.
One of my prediction looks set to come true
As I said, computing in America is become a much more patriotic talking point, and that much is obvious in what Jensen just said to the panel. He told a story of how Nvidia brought GTC to the Government's front yard in Washington D.C., but timings for Trump to attend couldn't line up because of his tour across Asia.
That being said, Huang said that Nvidia is committed to supporting the president and help America win. As is clear, this nation is in an AI race, and Team Green is the engine to it. Expect a lot of "Made in America" talk.
Stock price check
This is just over today, so not representative of the massive growth seen over the past year. But as you can see, the stock market is in a bit of a holding pattern at the moment to see what is announced at GTC.
For example, if some big Government work is confirmed or trade restrictions on selling GPUs to China are ready to go again, expect this to start climbing again.
A reminder of where AI is going in Nvidia's mind
This graph was shown at Nvidia's CES 2025 keynote and at almost every GTC since. This gives you a little context on where the path of AI is proceeding according to Nvidia.
Currently, we're heading hovering around Agentic AI (looking at you, Mico). So expect more happening in this space along with talk of Physical AI.
What can the schedule tell us about the keynote?
GTC's pre-show has always been a teaser for what to expect, and the key topics seem to be clear. We'll see how AI applies to science and quantum computing, robotics, and agentic AI for everyday industry.
The stream has started
The familiar graphics are back on screen, and I do feel a nice warm familiarity to seeing them. GTC kicks off soon!
Will we hear more about DGX Spark?
While we wait for Jensen to take the stage, I'm gonna talk about DGX Spark for a second. The AI mini supercomputer capable of generating a whole lot of tokens and running models locally at peak speed.
I think we may see news on software updates to the Linux-based OS coming to this. Maybe more LLM support?
The show is beginning
We're getting a trip down the memory lane of computing in America — spanning Apple, Microsoft and other key developments.
"Invented in America. Built in America"
The AI arms race is well and truly on, and Nvidia is keen to put America at the front of it.
Jensen is here!
Well after quite the patriotic start, Jensen has just taken the stage.
Moore's Law died, so it was time for another processor
We're getting a history lesson in how important the GPU and accelerated computing is moving things forward. The number of transistors may continue to grow, but the performance acceleration has slowed.
So it's time for new algorithms, new accelerated computing.
And here are the models powering accelerated computing
Nvidia has the hook-up of libraries to accelerate deep learning, quantum computing, and so much more.
RTX is looking GOOD
Yes, I know these are demos from CES, but it's still seriously impressive to see what a little bit of neural trickery can bring to gaming fidelity.
And here's how Nvidia GPU tech is influencing the planet
From robotics and self-driving, to weather prediction, healthcare and much more. All is simulated within the deep computer science.
Lock in. Here's what Jensen is going to be talking about!
We're covering AI, 6G, Quantum computing, models, enterprise, robotics and factories.
BREAKING: Nvidia is partnering with Nokia
Nvidia and Nokia are coming hand in hand to bring the Aerial Ran Computer (ARC) to kickstart the transition to 6G networking.
Aerial is the wireless communications system running on top of CUDA-X. So essentially, a model that runs efficiently while learning and improving itself with AI. Nokia seems like the perfect partner for it!
How does 6G and AI come together?
Beyond AI learning and making the spectral range of 6G more efficient — taking into account any limitations like weather conditions — you'll also see "AI on RAN."
What this means is that while in the current state of the internet, a lot of it runs (and went down last week) on AWS, Nvidia's creating a cloud computing platform on top of its 6G connectivity. That shows a promise of insanely fast AI that could fuel self-driving cars and the like.
Also, nobody laughed, but I did. Shout-out to the Nokia ringtone easter egg there.
The future of quantum computing
Currently, all states of quantum computing is strong but sensitive to environmental noise with limited usage. And that's where a GPU-based supercomputer comes in — to take that load off the quantum processor. This is NVQLink build on CUDA-Q cores.
This isn't here to replace quantum computers, but rather work hand-in-hand to accelerate quantum computing.
BREAKING: Department of Energy partnering with Nvidia to build 7 new AI supercomputers
Alongside the quantum computing jump, Nvidia has just dropped its first big governmental announcement. Team Green is teaming up with the Department of Energy to build 7 new AI supercomputers.
These will be at the Argonne National Laboratory and Los Alamos National Laboratory.
Pro-energy growth
With a new AI industrial revolution around the corner, there's a need for a whole lot of energy. Jensen Huang just praised President Trump for his pro energy stance to help America win.
Nokia stock check
Just like what happened to Intel's stock after the Nvidia partnership was announced, Nokia's taken a 20% leap.
"AI is not a tool. AI is work."
We're moving onto talking about AI, and about how agentic AI is not a tool, but rather an assistant in everything you do.
"The list of opportunities go on and on," Huang commented.
Here's Nvidia's vision
Nvidia's plan is to build factories dedicated to AI — packed with "mountains and mountains" of chips.
How to scale AI
Three key scaling laws are putting a lot of pressure on computation. Chief amongst them is the ability to think. This is the most GPU-intensive part of the puzzle!
"These AI models are now good enough that they are worth paying for"
Jensen proclaims we are now at a point where AI is smart enough that people should pay for it. And that creates intense pressure on compute resources with AI getting smarter and more people using it.
What's the answer? This is the virtuous cycle Jensen talks about — the more people use it, the more profitable it becomes. The cycle is spinning, but now its about cost-cutting.
Extreme Co-design
This is what's been at the core of Nvidia's meteoric rise over the past year. Working past Moore's Law by scaling up the chip itself, scaling it out across many systems, and scaling it across the planet in an interconnected way.
The NVL72 shield is back!
This is part of the new GPU supercomputer we saw back at Computex (albeit blown up in size), but it shows how these GPU cores come together in 72 GPUs in one rack. This is a monster of AI power.
Exponential growth
By better splitting out the thinking across more GPUs in the scaling, the GB200 NVL72 AI supercomputer is able to perform 10x faster, while also running it at 10x lower cost too.
What this means for you
So with NVL72 being manufactured and deploying to companies as you can see by these astronomical numbers on the graph behind Jensen, what does this mean for you?
Because I can talk about cost-per-token going down, but that could all very well be gibberish. What it means is AI is getting faster and cheaper to do. This could be an exciting inflection point for AI compute that makes it all the more accessible for everyone.
That's a lot of money
Nvidia has signed plenty of deals on the back of this, and is on for a 5X growth over its past generation GPU — totaling up to over half a trillion dollars of revenue through 2026 (so far).
And now we're seeing how these transitions are made on the wafer — and its happening in Arizona, Texas and California. It's a fascinating process!
And here's the timeline
Nvidia's got a lot of work to do through 2028 to continue growing that performance exponentially year-over-year.
DON'T FALL FOR THE DEEPFAKE!
As Tom's Hardware reports, there's a fake Nvidia GTC stream that shows a deepfake Jensen Huang shilling a crypto scam. It's currently garnering 5x more views than the real event. Don't fall for it!
And here's what will be in the racks of the future
Nvidia's new wave of chips purpose-built for AI supercomputing are here, and Jensen's showing off how they'll look in the data center racks. As someone who always thought car engines look awesome, this is peaking my nerd curiosity!
And here's the rack!
Two tonnes. 1.5 million parts — set to be used in AI factories. As Jensen proves, you can play a sick game of hide and seek amongst these beasts.
So much computing power it can build itself
No, not literally. Rather, the Omniverse model is capable of helping maximize the space of the AI factory you have to maximize the amount of AI tokens created — and even create a digital twin to optimize power consumption.
These are popping up in Virginia, Texas and all over the world.
Time for open source models
Open source AI models have become the "lifeblood of startups," as they are important for moving the world forward. "America needs to lead on this," and Nvidia is the world leader in providing open source models.
Here's the work behind it
"Science needs it, researchers need it, startups need it and companies need it." Looks like we're set for a big push on open source here.
BREAKING: Nvidia teams up with crowdstrike for an AI cybersecurity model
Agentic AI is coming to cybersecurity, aiming to take on the evergrowing list of breaches and tackle them head on with growing intelligence.
Time for physical AI — robotics time!
Here's how it works for robotics. Nvidia's training scenario is based on a digital twin in which the robot can train itself. At that point, those skills are then adapted to the real world. The physical AI of factories is almost here, and the robots are here to reindustrialize the US.
BREAKING: new Blackwell factory coming to Texas
To help with building all those Blackwell GPUs in the US, Nvidia is working with Foxconn to build a robotic factory in Texas to manufacture.
The Star Wars robot is back!
We've seen this a few times, but this is starting to grow a little more into reality here, as Nvidia and Disney Research continue to work on advanced robot learning using a digital twin to help its adorable sidekick to be aware of scenarios in real-time via training in the omniverse.
Nvidia DRIVE Hyperion is here to make cars "robotaxi ready"
Using all the advanced sensors you'd need like LIDAR, Drive Hyperion is the tech needed to build a standard platform for autonomous driving that could create robotaxis of the future.
And as you can see, a lot of companies have jumped on.
BREAKING: Uber is partnering up with Nvidia
Uber drivers, beware! Nvidia Drive Hyperion cars will have the option to be connected into Uber's network.
