Nvidia Silences AI Skeptics (For Now)

0:01Bloomberg Audio Studios Podcast Radio News. Nvidia just reported earnings for the third quarter, and to say they were highly anticipated would be an understatement. Nvidia earnings was said to be the most important day of the year for Wall Street. Many speculating that these companies' valuations are overinflated, that the numbers have just... Wall Street has been worried about an AI bubble, and Nvidia, which is now the world's largest company by market cap, is seen as a bellwether for the entire industry. The biggest names in AI run on NVIDIA's powerful chips, its graphics processing units, or GPUs, including its Blackwell model.

0:45There's been a lot of talk about an AI bubble. From our vantage point, we see something very different. That's NVIDIA CEO Jensen Huang on a call with Wall Street analysts after the company reported earnings that roundly beat Wall Street's expectations. And the relief among investors was palpable. Shares of NVIDIA jumped by about 5 % in late trading. There was a lot more anxiety going into it. Ed Ludlow is the co-host of Bloomberg Tech. If you look on social media, which can be a precarious thing to do, there were lots of people saying either NVIDIA is going to save or crash the global economy overnight.

1:23The stakes weren't as high as that, but it was serious and they certainly delivered. NVIDIA says it expects to generate roughly$65 billion in revenue in the coming quarter. And when Ed sat down with Jensen Huang on Wednesday, the CEO said NVIDIA is well positioned. Sales are off the charts for Blackwell and NVIDIA GPUs in the cloud are sold out. We got plenty of Blackwells to sell you. Things seem to have even more momentum than the most bullish people on the street thought that NVIDIA had. And there is no evidence that at least the most important customers that NVIDIA has are reneging on or pulling back on their spending commitments.

2:07So NVIDIA has allayed some fear AI is a bubble, but there is still concern about the circular nature of AI mega deals, how quickly high-end chips depreciate, and about the stratospheric valuations of many AI companies, which is reminding some investors of the late 1990s. Ed says he's asked executives if there's a parallel. I asked the CEOs and here's their answer, that the balance sheets now are different to what they were in the dot-com bubble, that they see actually useful software being sold, generating actual revenues. This is existential for many American companies. They are investing wholeheartedly in AI because they believe it will define what those companies do going forward.

2:51The idea that there's a bubble is secondary. It is not true if these companies need to become AI companies. It's an absolutely necessary investment. The difference this time around, they have the cash to support that.

3:21about a potential AI bubble. But the company and its backers still face some big challenges as NVIDIA races to keep up with global demand for GPUs. It's not slowing down.

3:39Ed, you've interviewed Jensen Huang a handful of times now, and I'm curious what your impressions of him are. What's he like as a leader? What's his personality like? He's incredibly busy, obviously. So, you know, when we spoke to him in that interview, he'd spent the entire day in Washington, D.C. at the U.S.-Saudi Investment Forum and spent time with the president and managed to pull off an earnings call and an earnings interview in the same day, which I'm sure took a lot. But that kind of speaks to what he's like as NVIDIA has kind of ballooned and grown and become so big. He's taken that in his stride quite a lot.

4:12You know, he is the focus of attention. But I think first and foremost, you know, he would tell you that without speaking on his behalf, that he's an engineer. And so he is involved deeply in what's going on in the company. So, Ed, NVIDIA released its latest quarterly earnings report after the close on Wednesday. I'm curious, what stood out to you? You know, NVIDIA beat on the most important segment, which is data center revenues and compute. But what they said for the fiscal fourth quarter was really interesting. Which is the quarter ahead. The period that we're currently in. Yes, I should say.

4:45But sales will be$65 billion, plus or minus 2%. And that's pretty significantly beyond consensus. NVIDIA is poised to start shipping a new GPU model next year. It's called the Rubin CPX, and it's designed to handle even more demanding tasks, like generating video. NVIDIA does not give guidance. It doesn't give year-ahead forecasts or very long-term forecasts. Except it did, you know, on stage in Washington, D.C. Curiously, you and I were both in the city at the same time. Jensen was doing his PowerPoint presentation with the clicky thing. And this slide appears behind him. And the slide says, over the next six fiscal quarters, just from Blackwell Rubin products, excluding China, NVIDIA says it will do$500 billion of sales, half a trillion dollars of sales over six fiscal quarters.

5:38And the street was a bit like, huh? That's way beyond what we've modeled for. This is a basic question, but why is NVIDIA viewed as a bellwether for AI? Why is it such an important company for Wall Street to track? Really simply, GPUs, graphics processing units, are chips that can handle multiple processes at the same time. They're very good for training AI models, and they're very good for running the AI models, what we call inference. There are many more options now today than they were, say, a year ago or two or three years ago. but nvidia still has a monopoly a technical monopoly for that market and the vast majority of workloads be they training workloads or inference workloads are done using nvidia gear so nvidia is a bellwether simply because if it's telling us this thing is happening it has the best lens into it happening the slightly different answer that nvidia would give you is that they are in their mind the only technology company in history in fact the only company in history, irrespective of what field they're in, that will say to everyone, their customers, their competitors, their suppliers, the analysts, here's what we're doing for the next five years from a technology perspective.

6:54And to their credit, they've done that. And so their argument is that the world that's around them in supply chain and their customers and everything that's being built around data center has had a pretty good heads up of what's coming to market. And that's one reason why Wall Street at least believes what they have to say. Ed, what are the indicators that are leading some investors and analysts to say that if we're not in a bubble, we're in bubble territory? There are two. Depreciation, basically the idea that any generation of GPU or AI accelerator has a limited shelf life. And therefore, because of the depreciation and its value, those big customers will have to mark down that value on their balance sheet at some point, noting that up front they made massive commitments to spend on this technology.

7:43And the other is the circular financing thing, which isn't going away. The idea that NVIDIA is the biggest player in the technology and it gives billions of dollars to an AI company, a software company, and that software company takes that money and spends it on NVIDIA technology and thus the circular financing argument. there's a concern that that is not a strong and long-term underpinning of a market that will keep growing. NVIDIA says software updates extend the life of its GPUs. But those circular deals are giving some investors pause. In September, the company announced a$100 billion investment in OpenAI to support new data centers and AI infrastructure, which will be equipped with NVIDIA's advanced chips.

8:29Since the start of the year, NVIDIA also inked deals with dozens of other companies, including Microsoft and Anthropic. You asked him about the circular nature of these deals as well. You had a wonderful way of putting the question to him. You asked about all that OpenAI is committed to buying. And you said, effectively, are they good for it? How do you know they're good for it? And what was his response to that question? Yeah, no, I literally said, how do you know that they're good for it? The way that Jensen described it in the interview is that their partnership with OpenAI is what he called disciplined.

8:58Well, we're thoughtful, along with OpenAI, thoughtful in aligning on and taking into consideration visibility of demand and their financing capabilities. All of that has to be in accordance, has to be aligned, has to be coherent before we start to build out. And so I think the ambition is large, but the execution is disciplined. Previously, in countenance to the circular financing discussion or debate, Jensen has said that nothing is a quid pro quo. When they make an investment in OpenAI, for example, there is no mandatory requirement that OpenAI takes that money and uses it on NVIDIA GPUs. They could use it to buy someone else's.

9:47But there are plenty of case studies where that just simply isn't true, right? This week alone, you saw the NVIDIA investment in Anthropic, a big rival of OpenAI, which also included one of the other cloud providers. And in that case, NVIDIA makes an investment in Anthropic, Anthropic gets money, and it does use that money to invest back into capacity that relies on NVIDIA chips. He didn't really answer the question as it relates to OpenAI. He just said that they're not concerned about it. In the run-up to the release of these earnings, these NVIDIA earnings, you saw some big shareholders offload a sizable amount of stock.

10:26You saw SoftBank do that. Peter Thiel's hedge fund did it. You've seen bets against this company as well in the options market. What does that tell us, those kind of sizable moves we've seen from some pretty key investors? What it speaks to probably is where people think the next opportunity is. NVIDIA does face increasing competition. There are other players that are working on GPU specifically for use in AI data center. But we also have ASICs and custom chips like Google's TPU. There are more options now. And part of the argument that Jensen Wang and the other NVIDIA executives make is like, this is broadening out.

11:02They love to see it. But it also reflects the anxiety that we had going into this earnings. After the break, more from Bloomberg's interview with NVIDIA CEO Jensen Huang. Now the company is navigating the ongoing trade tensions between the US and China.

11:27Ed, the two big global competitors in the SEI race are the US and China, and NVIDIA plays a central role in that competition. NVIDIA is now allowed by the U.S. government to sell some of its chips to China. But interestingly, China doesn't seem to be buying those chips from NVIDIA. Here's what the CEO of NVIDIA, Jensen Huang, told you about that. We would love the opportunity to be able to re-engage the Chinese market with excellent products that we deliver and to be able to compete globally. The Chinese market is very large. This year, my guess is probably about$50 billion. It's great for the American people that we're able to compete in the Chinese market.

12:05It's great for the China market that we're able to provide NVIDIA's technology to them. It's great for the rest of the world as Chinese software companies and Chinese open source models leave China and are used all over the world. And so I think it's fantastic that we're able, it would be fantastic if we're able to participate in the China market. But for now, we should just assume NVIDIA's forecast for China market is zero. Jensen Huang, very focused on the way China is growing its AI capacity and capabilities. What is China's approach to the build out and adoption of AI in complement or contrast to what's happening in the US?

12:45China wants to win the AI race as well. That much is clear. Let's start with the basics of what Jensen reiterated in the interview, which is base case scenario right now is NVIDIA does not derive any revenues from China. What was interesting to part of your question, on the earnings call, the CFO CollectCrest had to address the China issue, right? Because, you know, everyone knows it's going to be asked by the analysts anyway. and your interpretation of and I mean that with respect the word interpretation of what she said was they were allowed to sell H20 in the quarter but there wasn't any demand for it what they're not allowed to do is sell Blackwell the kind of current generation architecture even a deprecated version into China so those distinctions are important and where we left it with Jensen in the interview was they want to be able to to do that in China and they will work with both the United States government and Chinese government to find a solution, but base case is zero.

13:43There is a wider political discussion. Jensen Huang is in very close proximity to this administration, right? He was in DC when you talked to him, and he was in DC when you talked to him the time before that. And in both of those occasions, the president of the United States was part of the news flow, commenting very positively about Jensen Huang and what NVIDIA is doing. And In the context of the day of the interview in the US-Saudi Investment Forum, the president asked Jensen to stand up to a standing ovation. But the reality for NVIDIA is that they right now have no access to a market that at one time was 25 % of revenues and is the world's biggest end market for data centers probably.

14:25But more academically speaking, Jensen Wang has been very consistent that more than half of the world's AI engineers are either in China or they're Chinese national citizens. And so America should want its leading technology to be the gold standard, the default that those engineers use to develop artificial intelligence, noting fully that there is, of course, geopolitical and national security considerations that come with that. Ed, if you talk to our colleagues who cover tech in China, do they say that there is as much of a fear of there being an AI bubble there or the circumstance is just entirely different?

15:06The circumstances are different insofar as there is less red tape in China and more direct government support or action in favor of something. It's not the same level of anxiety in China because they have a different system of government input or support. Their domestic industry on the infrastructure side is nascent. What if Jensen Huang is wrong and we are in a bubble? and that bubble bursts, are we likely to see a crash? Are we likely to see consolidation? As you talk to your sources, how do they think about the way in which that might play out, given the rapid breakneck growth we've seen, the size of these valuations, the money that's been poured into this and taken from one pocket to another?

15:58What would a bubble bursting look like when it comes to AI? Well, in order to look at what a bubble bursting looks like, I guess you have to set the state of play and the state of play is that you know in 2025 nvidia is up about 40 and of course nvidia accounts for eight percent of the s &p 500 anyway so in equity markets there's a lot at stake the leverage side of this is really fascinating as well debt oracle for example is very very involved in the ai infrastructure build out in the united states and it has been able to raise massive amounts of debt and also get investors to raise debt for some of its projects where Oracle doesn't hold the debt on its balance sheet.

16:39It's just able to get someone to do it on its behalf. But the worrying indicator there is that Oracle swung to negative free cash flow for the first time since 1992 this year. And so the confidence that investors had was that this is early innings. The AI infrastructure build out is early. The figure that is common between, say Jensen Wang and other chip CEOs is there's a$4 trillion addressable market for AI data centers in the next say three to five years. And that is all built on capital expenditures of the big hyperscalers, cloud computing companies, and also like Meta, Elon Musk's XAI, etc.

17:18And everyone is working in good faith that those capital expenditure commitments that are forecast, they go through with them. If they don't, you can kind of see the trickle down effect that I'm painting here. That's why going into this Inverdia earnings print, the anxiety was high. You know, if you look at the options market or whatever your indicator is, there was potential for this to go either way. And at the time you and I are speaking, it went positively in the market.

17:59slash podcast offer. If you liked this episode, make sure to follow and review The Big Take wherever you listen to podcasts. It helps people find the show. Thanks for listening. We'll be back tomorrow.