Today's number: 887 million. That's how many dollars Warner Brothers Discovery CEO David Zaslav is expected to receive when he sells the company to Paramount. During his tenure, Zaslav was responsible for instant classics such as Magic Mike's Last Dance and Joker Part Two, but he will always be remembered for his true crowning achievement: rebranding to HBO Max.
Welcome to Prof G Markets. I'm Ed Elson. It is March 18th. Let's check in on yesterday's market vitals. The major indices posted modest gains. The price of oil climbed higher as President Trump slammed allies for not helping him fight the war in Iran and reopen the Strait of Hormuz. And finally, bond yields fell ahead of the Federal Reserve's interest rate decision due later today. Okay, what else is happening?
Nvidia expects at least $1 trillion in revenue from Blackwell and Rubin chips through 2027. That is the headline news out of the company's GTC conference, also known as the Super Bowl of AI, which kicked off this week. In his keynote on Monday, CEO Jensen Huang pitched a future where Nvidia powers not just chatbots and data centers, but also robots, factories, and autonomous machines in the physical world. Nvidia's stock is up nearly 1.5% since his speech. Lots more to unpack from this conference. So, here to discuss all of it, we're speaking with our friend Gil Luria, head of technology research at DA Davidson. Gil, good to have you on the show. I think we have to start with this $1 trillion number that is making all the headlines. One trillion in chip sales through 2027. That's what Jensen Huang believes. Give us your views on that number. Is that a real number or is that just a headline?
No, it's a real number. What's really surprising is that investors shrugged it off as unimpressive. It really added half a trillion dollars of revenue in a year. And you would think investors would jump all over that, considering it implies upside to next calendar year's numbers for Nvidia. And yet investors shrugged it off, which tells us something really interesting. Investors don't believe the great data center build that will continue into next year. The market is now telling us that it is expecting 2026 to be the peak year in the data center buildout. That's indicating a little bit of split personality for the market, because as you discussed and we've discussed, the market is also saying mass unemployment is coming, software companies are worthless. So AI is going to happen, it's going to have a huge transformational effect, and yet at the same time the market's not believing that the great data center buildout will continue into next year. So somewhere along there, there's an inconsistency.
I guess something that I would empathize with the market's reaction here. I mean, when we hear Elon Musk, for example, talk about Optimus robots and sort of talk about these big numbers in the future, he'll often say something like, 'I think this is going to generate trillions of dollars worth of revenue.' And he seems to sort of just say it on the fly. He said similar things about robo-taxis as well. And then it never materialized, or if it did, it materialized in a much more insignificant way than a lot of people expected. And I guess the question remains here for Jensen Huang too. I mean, why is this one real when it seems to be also a little bit of a projection out into the future? Why is it something that we should really be taking seriously?
That's a fair question, because these numbers are fantastical. So we do have to always be skeptical when people give us fantastical numbers like that. But to be fair, two different people. Elon Musk has a long, illustrious track record of overpromising and compressing timelines. Even when he knows it's going to take a lot longer, even though he usually delivers on the promises, eventually the timeline gets extended a lot. Jensen Huang has actually been pretty consistent. He's actually been cautious not to guide too far into the future. Up until just six months ago, Nvidia was only guiding one quarter at a time. Six months ago, he took a departure to say, you know, I think there's going to be half a trillion dollars of Blackwell and Rubin by the end of '26 or in the next 18 months, which is slightly past '26. And he pointed to that and said, 'Look, here I am five months later. I'm telling you that number is too low. I need to take that to a trillion dollars if I extend it to 2027.' So he's being thoughtful, and he, at least in the way he's communicated and clarified that statement, feels like he's underpromising. He clarified this doesn't include CPUs or the Groq chips or the networking equipment or Rubin Ultra. So he actually said it's more than a trillion. So from his perspective, this is a number that he has high visibility into because he has these orders from his largest customers: Microsoft, Amazon, Google, Meta, etc. So he feels like that's actually underpromising, which is probably why he's a little surprised by the underreaction to such a high number.
So I guess what then do we make of the fact that the market doesn't seem to be so hyped about the next couple of years in data center buildout? Why do investors believe this? Why do they, it seems like you're saying, think that we've kind of reached peak data center?
Well, I don't think we've reached peak data center. I'm saying the valuations for these stocks imply that. And it's not an unreasonable thing to say, look, these big companies are investing hundreds of billions of dollars. They do not yet have the revenue and the returns to justify that. Until we see those returns come in, we're going to be skeptical about these companies continuing to invest at these rates, because these companies are using all their cash flow and then some to build data centers. So they can't keep doing this forever. So we, the investors, the market as a collective, is view is let's wait until we see the returns come in a little better. Then we'll be able to make that leap of faith that the data center construction is going to continue at these rates. It is our opinion, it is my opinion, that we will get these returns. That the growth rates we're seeing at OpenAI and Anthropic and in usage of AI by consumers and enterprises is so significant, and the technology is still developing, that we will actually need a lot more data centers. But the skepticism is always warranted. It just creates opportunities for investments.
One of the big themes in his keynote was physical AI. So robots, factories, autonomous systems, etc. Does that play a big role in your view on how this AI buildout will unfold? Are you thinking about physical AI? Was that story compelling to you?
It's compelling, but the timeline is different. You pointed earlier to Elon Musk and his promises about Optimus enrolling millions of Optimuses into every home, and we know that's a few years away. I would think of that as extending the AI cycle, right? If the AI cycle is about increasing productivity of white-collar workers, physical AI and robotics are about extending the productivity of blue-collar workers. For those of you that didn't see the keynote, they even animated Olaf from Frozen and had him come on stage to talk to Jensen Huang. And so it is coming, and we've seen some demos of Optimus and Figure robots, etc. We just know that that's several years out. So that's probably behind the big AI wave. We're going to have a big humanoid robot wave that'll help automate the other types of jobs, from white-collar jobs to blue-collar jobs.
What would you say was the single most important takeaway from GTC so far?
The incorporation of the Groq technology that Nvidia acquired. So, if we've been talking about Broadcom and AMD trying to catch up to Nvidia, AMD has a new generation of GPUs, some custom GPUs they're doing for Meta, for OpenAI. Broadcom does the A6, which is custom accelerator chips that they do for other large companies like Google, etc. And those companies appeared to be catching up to Nvidia by incorporating Groq technology into their data center systems. Nvidia is extending its lead from those companies. They're talking about the fact that they can generate more tokens faster and cheaper than any of those alternatives. And it was very important for them to bring the Groq technology in in order to accomplish that. So it's Nvidia extending its lead over the competition and making the argument that the total cost of ownership, even for inference, is lower with Nvidia systems, and the key to that was adding the Groq technology.
One other thing that has struck me about this conference is just how big it is. I mean, people are going Super Bowl of AI, Woodstock of AI. I mean, you see these images of these gigantic crowds literally filling out a stadium. I saw someone online saying that Jensen Huang is the Taylor Swift of men, which I think is weirdly actually not that inaccurate. What do you make of just how big of an event this has become, and I guess the stardom of Jensen Huang in 2026? What does that say about society and where we are as a market right now?
Yeah, this is the big event of the year for AI, and it's not a coincidence. Jensen Huang foresaw the future of AI before anybody else did, right? He was meeting Sam Altman in 2017 to give him chips. So he saw this coming. He's anticipated several steps forward. So this is really well deserved in the sense that he is very much, and Nvidia is very much, responsible for the AI revolution, and it's not a coincidence that they also capture most of the profit dollars in AI and will continue to do so for a while. So it's his vision that has been responsible for a lot of the progress. It's made Nvidia GTC the main event for AI right now. And that's why it's such a big celebration. And plus, he has the iconic black leather jacket. That always helps that you have something like that. It looks pretty good.
