The "AI Super Bowl" Previews the Future
19:02 Share

Today on the AI Daily Brief, NVIDIA hosts the AI Super Bowl. So why are markets so unimpressed? The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. To join the conversation, follow the Discord link in our show notes. Hello, friends. Today we have a big fun topic with the GTC event going on in California. And it is one of those topics that's big enough that we are going to skip the headlines today and just focus on the main topic.

We will be back tomorrow with our normal format, but for now, let's talk about AI's Super Bowl.

Today, we are talking about the AI Super Bowl, by which, of course, I mean NVIDIA's annual developer conference, GTC. I actually don't even know if the Super Bowl is the right analogy. Other people have pointed out that at this point, tracking the company's rise over the past few years, the event is more like a rock concert than a tech conference. There were literally lines around the block as crowds crammed into the SAP Center in San Jose, which again, is usually used to house professional sports games or Taylor Swift concerts.

Tae Kim, the author of the NVIDIA way, called it an absolute madhouse and reported that 25,000 people were in attendance. That Super Bowl analogy actually came from CEO Jensen Huang. And to really torture and extend this analogy even further, there was even sports-style coverage from a roundtable of analysts throughout the event, including over here, ex-Intel CEO Pat Gelsinger.

Now, NVIDIA had a jam-packed slate of announcements, and of course, we're going to go through all the big ones. We're going to kick it off by looking at Jensen Huang's keynote speech. But before we do, I want to set a little bit of context for the reaction. Depending on what part of the world you're coming in the door from, you might have very different perceptions of what's going on over there. For a look at the mainstream and Wall Street side of things,

Let's turn it over to friend of the show, Bloomberg host Caroline Hyde, who posted this on TikTok in the wake of the keynote.

Where's the AI hype gone? Where's Nvidia, Jensen Wang's ability to rally the stock, rally the crowd at GTC, which is basically his AI Super Bowl? Because it's really fizzling at the moment, even though we have a whole host of announcements coming from that particular event. Whether it's the latest Blackwell Ultra, which of course is the next leapfrog of his AI platform currently and the chipset that he offers.

Ultimately being able to offer more compute for reasoning models for agentic AI, the agents everyone keeps talking about, or indeed even physical AI, robotics and cars. He's talking up the next Vera Rubin AI platform, the chips that are going to be three times as powerful in terms of compute than the Blackwell iteration. Talking up how they're now going to use light, photons instead of electricity in chips and what that's going to do to galvanize growth, let alone energy.

what's happening for 6G networks, how he's working with telecoms on having AI integrated into our compute at the edge, whether it's also doing a deal with GM on self-driving cars, self-driving manufacturing too. There's a whole host of news and the shares stay lower. Let me know what you think about perhaps this fizzling AI vibe when even Jensen Huang can't manage the crowd.

All right, so I think that's a great setup for our conversation. And before we get in and try to answer that question, let's look at what was actually announced. First and foremost, Jensen is sticking to his guns and forecasting much, much more demand for AI chips into the future. He declared, "...the amount of computation we need at this point as a result of agentic AI and reasoning is easily 100 times more than we thought we needed at this point last year."

And of course, right here, we are getting into perhaps one of the big challenges for the market analysts around NVIDIA. We are still living in the question most poignantly brought up by DeepSeek around just how much compute we're going to need in the future. On the one hand are those in the AI industry who are squawking as I have so often squawked about Jevon's paradox and the fact that the cost coming down means that usage will go up.

But it's clear that market participants are looking for reasons to doubt NVIDIA's continued growth, and so this lingers. Makes it not surprising that this was something important addressed in this keynote speech. Now, Jensen's new metaphor for how he's thinking about the era of supersized data centers was this. He said, The future of software requires capital investment. Now, this is a very big idea. Whereas in the past, we wrote the software and we ran it on computers,

In the future, the computer is going to generate the tokens for the software. And so the computer has become a generator of tokens, not a retrieval of files. From retrieval-based computing to generative-based computing, from the old way of doing data centers to a new way of building these infrastructure, and I call them AI factories.

They're anti-factories because it has one job and one job only: generating these incredible tokens that we then reconstitute into music, into words, into videos, into research, into chemicals or proteins. We reconstitute it into all kinds of information of different types.

Jensen went on to predict that the data center buildout will reach a trillion dollars and said, I'm fairly sure we're going to reach that soon. So clearly you are getting a very consistent message that this line is still nothing but up and to the right. Another big overhang facing the company is the rollout of the new Blackwell chips. After numerous delays, the shipments do seem to be ramping up, but we're yet to see a full data center buildout running the new chips. Jensen insists the wait will be worth it.

Comparing them to the current Hopper H100s, he said, Quipping that his team might not love those remarks, he added, And so here again, we have another challenge for Wall Street.

Wall Street is already concerned about what they believe could be out-of-control AI capex, and yet the technology is proceeding so fast that companies that aren't keeping up face some real threats.

Selling the need for chip upgrades even further, and really crisply defining the problem, Jensen commented, every single data center in the future will be power limited. Your revenues are power limited. You can figure out what your revenues are going to be based on the power you have to work with. We are now a power limited industry. Based on that, you want to make sure you have the most energy efficient compute architecture you can possibly get. So big theme one, big takeaway one.

Demand for compute, need for compute, need for upgrades to the latest high-performance compute. Not going anywhere anytime soon, according to NVIDIA.

Now, if you haven't really been paying attention, you still might have seen a cute little robot that NVIDIA is calling Blue with a design reminiscent of Star Wars. Moving on past the base case to other products, Jensen announced that NVIDIA would be partnering with Disney to release these next-generation entertainment robots designed for use in theme parks. The robots are being trained using a new world model called Newton, developed in collaboration with Google DeepMind.

The world model allows the robot's AI model to be trained in a simulated environment, ensuring it can handle real-world situations. Chubby remarked that when Disney is in, it's probably time to pay attention, stating, This was the wow moment for me. Robots are coming. This time it's real. And I am all in for it. The other big robotic announcement was a new open-source foundation model designed to power humanoids called, quite perfectly, Groot N1.

The model was shown off via X1's NeoGamma robot, which was displayed autonomously vacuuming a mocked-up lounge room. Jensen said, the age of generalist robotics is here. With GROOT N1 and new data generation and robot learning frameworks, robotics developers everywhere will open the next frontier in the age of AI.

X1 Technologies CEO Brett Bornick added, The future of humanoids is about adaptability and learning. NVIDIA's GRUT N1 model provides a major breakthrough for robot reasoning and skills. With a minimal amount of post-training and data, we were able to fully deploy on Neo-Gamma, furthering our mission of creating robots that are not tools, but companions that can assist humans in meaningful, measurable ways. Today's episode is brought to you by Vanta. Trust isn't just earned, it's demanded.

Whether you're a startup founder navigating your first audit or a seasoned security professional scaling your GRC program, proving your commitment to security has never been more critical or more complex. That's where Vanta comes in. Businesses use Vanta to establish trust by automating compliance needs across over 35 frameworks like SOC 2 and ISO 27001. Centralized security workflows complete questionnaires up to 5x faster and proactively manage vendor risk.

Vanta can help you start or scale up your security program by connecting you with auditors and experts to conduct your audit and set up your security program quickly. Plus, with automation and AI throughout the platform, Vanta gives you time back so you can focus on building your company. Join over 9,000 global companies like Atlassian, Quora, and Factory who use Vanta to manage risk and prove security in real time.

For a limited time, this audience gets $1,000 off Vanta at vanta.com slash NLW. That's V-A-N-T-A dot com slash NLW for $1,000 off. Hey listeners, want to supercharge your business with AI?

In our fast-paced world, having a solid AI plan can make all the difference. Enabling organizations to create new value, grow, and stay ahead of the competition is what it's all about. KPMG is here to help you create an AI strategy that really works. Don't wait. Now's the time to get ahead. Check out real stories from KPMG of how AI is driving success with its clients at kpmg.us slash AI. Again, that's www.kpmg.us slash AI.

Today's episode is brought to you by Super Intelligent and more specifically, Super's Agent Readiness Audits. If you've been listening for a while, you have probably heard me talk about this. But basically, the idea of the Agent Readiness Audit is that this is a system that we've created to help you benchmark and map opportunities in your organizations where agents could

specifically help you solve your problems, create new opportunities in a way that, again, is completely customized to you. When you do one of these audits, what you're going to do is a voice-based agent interview where we work with some number of your leadership and employees to map what's going on inside the organization and to figure out where you are in your agent journey.

That's going to produce an agent readiness score that comes with a deep set of explanations, strength, weaknesses, key findings, and of course, a set of very specific recommendations that then we have the ability to help you go find the right partners to actually fulfill. So if you are looking for a way to jumpstart your agent strategy, send us an email at agent at besuper.ai, and let's get you plugged into the agentic era.

All right, so big theme two, robots and physical AI. Next, NVIDIA previewed the next few generations of their AI chips, including the full product line for the Blackwell architecture. Like the Hopper, the new architecture will be used to power a variety of different chips of varying capacity. Jensen unveiled the flagship model for this generation called the Blackwell Ultra GB300. The Ultra version of the chip will have the same GPU but feature additional memory. He said the top-of-the-line unit would be delivered in the second half of this year.

Their next GPU architecture, called the Rubin, is expected in 2026, and will feature another step up in power, networking capability, and a new CPU called the Vera. A full rack of the Vera Rubin chips is expected to be able to deliver more than three times the performance of a full rack of Blackwell Ultra chips. The final architecture revealed at the event was called the Feynman. It will feature the same Vera CPU-T and another step up in networking technology. The 2028 release date is still a little far off in the future for more solid descriptions. Touch

Touching quickly on the production run for Blackwell, Jensen said that NVIDIA had already booked $11 billion worth of Blackwell revenue and that the top four buyers alone have purchased 1.8 million chips so far this year.

The event also featured an update on the desktop AI supercomputers that were previewed in January. Then known as Project Digits, the expanded product lineup now includes a pair of personal AI supercomputers with the updated name DGX Spark and DGX Station. The Spark is on pre-order for $3,000 already, while the Station is coming later this year and has no price tag for now. The Spark is the model that we've already seen, a Mac mini-style profile computer around the size of a large fantasy novel.

It contains a Blackwell GB10 core and 128GB of memory. NVIDIA claim the Spark will be capable of running models up to 200 billion parameters locally and is suitable for prototyping, fine-tuning, and inference. The station is a larger workstation-sized machine designed for a full-size Blackwell Ultra GB300 chip and boasts 784GB of memory. NVIDIA claim the station will be capable of delivering 20 petaflops of AI performance.

The station will also be delivered through manufacturing partners with Asus, Box, Dell, HP, and Lenovo, all producing their own versions of the product. Jensen said, This is the computer of the age of AI. This is what computers should look like, and this is what computers will run in the future. And we have a whole lineup for enterprise now, from little tiny ones to workstation ones. Drawing a fine point on the use cases for these ultra-powerful machines, he continued, AI agents will be everywhere. How they run, what enterprises run, and how we run it will be fundamentally different.

And so we need a new line of computers, and this is it. And so we have a third point here. Not only do we have a never-ending and ever-growing demand for compute, not only are we going to have totally new arenas for that compute to be delivered, such as the embodied AI of robotics, even when it comes to more traditional business use cases, the nature of computing itself is changing so significantly that we're going to need different types of computers. So here we have another pillar of Jensen's arguments.

And yet still we weren't done. Going a little bit more quickly through the rest of the announcements, there was a large section of the event that focused on NVIDIA's work on self-driving vehicles. The company has signed a partnership with General Motors to deploy the technology and much more. Jensen said, "...we're looking forward to building with GMAI in all three areas. AI for manufacturing, so they can revolutionize the way they manufacture. AI for enterprise, so they can revolutionize the way they work to design cars and simulate cars. And then also AI for in-the-car."

NVIDIA will leverage Omniverse and Cosmos models to train the AI underpinning GM self-driving cars, with the ambitious goal to deliver advanced self-driving exclusively using synthetic data, with NVIDIA producing a digital twin of the real world. On the production line, these same models will be able to produce digital twins of the entire facility to test and refine procedures without needing to shut anything down.

Finally, NVIDIA announced a new version of their open-source Nemotron reasoning models. The models were first announced in January and are designed to be business-ready to deliver reasoning for agentic deployments. All right, and so with all of that, the question comes back to why did this fall flat for Wall Street? Certainly it didn't fall flat for others. Ishwar Srinivasan writes, This morning I sat in awe during Jensen's keynote at NVIDIA.

What struck me most is how these innovations are set to change everything, from healthcare to robotics to enterprise solutions. We're moving from an era where AI simply retrieves data to one where it generates insights, solves intricate problems, and becomes a true partner in innovation. Steph Sheplik writes, The most insane thing about NVIDIA is what they're selling now is essentially two to three years behind what they're preparing to deploy. It's like a weapons manufacturer who knows they have to keep some secrets locked down until the market is ready.

Those with NVIDIA chips can build the fastest, perform the best globally. There's no competitor that can touch them right now. You can read all the FUD you want, but we haven't even scratched the surface yet of what's coming and what's possible. Sending it all up, Fired Up Wealth says, if this made you sell NVIDIA stock, you have no idea what you're doing. So what's going on? In short, I think that the market's reaction to this NVIDIA keynote has less to do with NVIDIA and more to do with the market right now.

I've made this point before, but for about two and a half years now, AI has had this really weird role in markets where at many points it has effectively just been asked to bolster them and be the entire counterweight narrative to everything bad happening outside of AI. Jerome Powell and the Fed underwent the fastest rate hiking cycle in 40 years. And the thing keeping markets afloat and not totally just descending off into despondency was NVIDIA and the rise of AI and the excitement around it.

Indeed, it was AI versus everything for basically all of 2023 and a lot of 2024. Now, ever since the hiking cycle ended, and especially since the cutting cycle began, investors have been looking for opportunities to reprice these tech stocks, both AI and otherwise, and there's been a constant never-ending drum of potential FUD around them. One could be forgiven for thinking that the deep-seek moment has really been the straw that broke the camel's back,

and crystallize these fears of overspending on AI capex. And while I don't think that that's totally inaccurate, I think it would be putting too much on markets' reaction to AI and underappreciating how much the general volatility of this new administration is actually the force that's driving most market action right now.

Markets are not only trying to make sense of tariffs and a global economic realignment. They're dealing with the fact that in Trump, too, there is no Trump put. That the administration has Treasury Secretary Scott Besson out here on the morning shows every day telling the markets that there's going to be pain. It's completely unlike things that Wall Street has experienced for a very long time. And it's causing a broader volatility.

And so with all of that, basically, to the extent that you're trying to draw inferences from the market's response to NVIDIA, I think it's fair to say that even Jensen Huang couldn't get people excited over all those other factors. But I don't think that you can say that anything that he did or that NVIDIA is saying is in and of itself a negative signal. It's just not positive enough right now to beat out everything else.

Ultimately, when you take a step back, this image of a previous Jensen keynote, where he shows AI moving from perception AI to generative AI, into where we are now with agentic AI, into the future, which is physical AI with self-driving cars and general robotics, all pointing aggressively up to the right, this is the story of the time. This is the moment we're actually living through.

There are going to be market moves in both directions throughout this, but ultimately, none of those moves are going to change the fundamental underlying fact that progress is screaming forward and the world is being remade around it. So that is from where I'm sitting, the story of the AI Super Bowl this year. Let me know what you thought about the keynote. Hit me up in the comments, either on Spotify or YouTube. For now, that is going to do it for today's AI Daily Brief. Thanks for listening or watching as always. And until next time, peace.