Why Nvidia's Grip on China Won't Last Forever | Gil Luria

Welcome back to the Rundown interview edition. In today's episode, we are talking to Gil Loria, the head of tech research at the investment bank, D.A. Davidson. Our conversation today was mostly about NVIDIA and their earnings. Gil has been covering NVIDIA for a while now, that he just raised the price target of their stock. Thanks for having me. Well, today we got to talk about NVIDIA. It was the story of the week, story of the year, honestly. Their earnings dropped last night and their stock reaction, not so great, dropped 3% after hours. It's recovering a little bit right now, but the numbers are pretty solid, you know, beat on top line, beat on bottom line. So Gil, tell me, are we cooked or is NVIDIA going to be okay? NVIDIA is going to be okay. Okay. We're cooked in many other ways. This isn't one of them, right? We are, big picture, we are using AI a lot more. The models have gotten so good and they're so compute intensive because of how powerful they are that we need more and more compute capacity. And NVIDIA... Not the only game in town, but it's the main game in town. And so you're big, the company's building all this capacity out. Microsoft Azure, Amazon Web Services, Google Cloud are mostly still relying on NVIDIA. And then companies that are building AI capabilities. He's... Either because it helps their internal business or because they're trying to achieve superintelligence. And let's put Meta and Elon Inc. on top of that list. Also have insatiable demand for more compute capacity. And again, as long as NVIDIA is the main game in town, they're going to be just fine. So that's okay. I see what you're saying. And that's the bull case right now for NVIDIA. See ya. So what was your takeaway on the earnings? They missed slightly depending on which publication you look at on their data center revenue. It came in at 41.1 billion, still grew 56%. So what was your takeaway on that slight miss? And is this just a matter of our expectations are so high that even a slight miss like that can cause a stock to drop? It's really all about China. China is a lot of noise. And with NVIDIA, China is a messy story to begin with because they have this quirky way of reporting geography where they report geography by where the invoice came from, which really doesn't help us know where the chips are going to. And so we actually believe that at least a quarter of their chips Used to go to China and still a pretty substantial amount of their chips go to China. Now, either into China, to a Chinese company, or to a reseller in Southeast Asia that ends up selling into China. And I wish they were more transparent about it. They're not. And this is when it comes back to bite them, because not only are they dealing with a dynamic Chinese market that has internal competition, that they're not allowed to sell the advanced chips into, but the disputes between the U.S. government and China are making it even more complicated. So they were they sold H20s for a while. That's the degraded hopper chip. They sold them into China for a while. It was a really big part of the business. Then abruptly in April, they were told by the government they can't sell them. And then the government said, well, maybe you can sell them, but we're going to put tracker chips and we're going to charge you 15 percent of the revenue. And so now they're not even sure they can sell them into China. So they're selling them not into China. And then the Chinese are saying, what do you mean you're putting tracker chips into the servers? We don't want any of your chips because you have tracker chips in them. So NVIDIA is caught in the middle. It's creating a lot of noise. That's why their guidance was less than expectation. That will continue to be the source of noise, especially in the short term in NVIDIA. I would draw a parallel to Apple, right? Remember Apple up until five years ago, 20% of their revenue came from China and then China got more complicated. They got into the middle of the trade dispute and now it's closer to 10%. This is the trajectory NVIDIA is on. They'll keep selling in China, but it'll be less and less over time. And that's the reaction today in the stock is that guidance was less than expected because they can't sell into China probably this quarter. If they do, they said that's an additional two to five billion. If that's the case, estimates are fine. If not, they will have a little bit of a short-term headwind. So you see this playing out to where, at some point, it's just going to be an inconsequential amount for their revenue. The Chinese market is essentially going to be shut off from Nvidia As they just, as they get, you know. Pushed out by either the Chinese government or homegrown players in China like Huawei. There's a couple others that are trying to make chips that rival NVIDIA. Do you see that playing out or do you see maybe something get, I mean, Jensen Huang, very charismatic guy, you know, does he broker a deal where he's allowed to sell his chips back in China? We don't know. The short answer is we don't know. We think that it is an eventuality that the Chinese market will be taken over by Chinese incumbents. But Jensen Wang may be able to negotiate directly with the president, by the way, a way to continue to sell some chips for some time. And again, let's not forget the advanced chips that they're selling to resellers in Southeast Asia are still getting to China. So they are still selling to China. They're not supposed to. People are getting arrested. But it's happening. Let's face it. It's happening. And that can continue for a while longer. But again, the eventuality, to your point, Huawei is going to sell these chips. The Chinese government prefers that it's Huawei chips. Huawei's closed the gap. The performance gap, it's maybe a year's worth of performance gap. So the more NVIDIA is required to sell degraded chips, the less their chips are even better than the Chinese chips. So yeah, over time, the Chinese market will go away from NVIDIA. The question is, how quickly does it happen? The short term, does it happen gradually? That depends really on the negotiation between Jensen Wang and the president. Well, there's a lot of demand, supposedly, here in the US, right? I think Jensen said in the earnings call that $600 billion in CapEx AI spending from the big tech companies over the next couple of years, and he said trillions of dollars by the end of 2030. Do you really see that playing out? Or the skeptical person would say, well, like, okay. Is it actually going to continue? Are we going to actually see massive amounts of spending? Or are we going to have a situation where we get into an earnings call and Satya Nadella gets on and says, you know what? Kind of scale them back a little bit. That causes a domino effect. I mean, are investors going to put up with these big tech hyperscalers spending so much money without any sort of payoff? Because we haven't seen an ROI just yet. Yeah. I'd say that up until six months ago, we were in the camp that that's inevitable and it could happen as soon as this year. Because we were doing the ROI math and we could walk into the, we could walk through the ROI math a little bit more. And we thought that could happen as early as this year. Having said that, and remember Deep Seek was the moment of kind of... in terms of thinking models are going to be as good as they're going to get, and now they're just going to get smaller and cheaper. And that was, you know, early this year, kind of in that January through April timeframe. What's actually happened, though, is the models have gotten a lot better. So if Chad GPT was more of a curiosity and a little bit of a parlor trick up until Until even six months ago. If you're using the model now to its fullest capability, and more and more people are, it's actually quite powerful. And so that's why we still think Microsoft, Amazon, Google's customers will continue to come to them for more capacity. Now, let's talk about ROI for a second. And here, the analogy is the picks and shovels analogy, right? NVIDIA sells picks and shovels. Microsoft, Azure, Amazon Web Services, Google Cloud, buy picks and shovels and rent them out to their customers. That business is a good business. It's not great. The returns there are OK, maybe low single-digit returns. They're positive. They're not great. But the growth there is so substantial that Microsoft, Amazon, Google, who have So much capital that they don't have anything else to do with will continue to pursue it. The question is about the gold diggers themselves, right? Those companies are coming to Microsoft, Amazon, and Google and renting as much capacity as they can get because they're so eager to defend their business from the disruption that's coming that they're throwing stuff at the wall. As long as that's the case, Microsoft, Amazon will continue to buy the picks and shovels and rent them out. Only if we really see that people aren't getting value and there's no prospect of people getting value from AI, will we think that that's going to dry up. And again, we thought that could happen this year, but based on how good the model they're getting, we don't think that's happening in the next year or two now. We think that we are going to be so much more productive if we use these AI tools. The phrase that comes to mind is something an economist named Richard Baldwin said back in 2023, which is, your job isn't going to be replaced by AI, your job is going to be replaced by somebody who uses AI better than you. That's where we're at at this point in time. The tools are so good, the people that use them well are going to be so much more productive that they'll continue to drive that demand. That's where we're at on this right now. What do you think about the sentiment now of this AI bubble? Even Sam Altman himself, CEO of OpenAI, he's talking about it. There's this meme going around talking about how there's these AI apps that are running at negative gross margins. That money is going to Anthropic. Anthropic is sending that money to Amazon. Amazon is using that money to buy NVIDIA. It's all built on this pyramid, but it's all negative gross margins. Uh, being, um, financed by VC money. What do you think about that? I mean, even, even, even this, this week I saw. Microsoft CEO Satya Nadella, he did a Twitter thread talking about his favorite use cases of AI, and I'm like, that's a negative sign to me. So Gil, where is your head at about this whole narrative of AI? A lot of that stuff's valid. So who had it in their bingo card that Satya Nadella would turn into a thread boy, right? I didn't have that in my bingo card. Thank you, Lord. Yes, that's true. These are heavily negative gross margin businesses. This is OpenAI. OpenAI is a hugely negative gross margin business that can't last forever. And what's interesting there is, you know, people draw the analogy to Uber, right? Uber was a hugely negative gross margin business for a while until we only used Uber and then they raised prices. OpenAI is not necessarily in the same boat because... We could use Google Gemini, we could use Anthropoclod, so it's not clear, or Grok, right? Grok's in my car now, all I have to do is press the voice button for a second and I can talk to Grok, which is totally creepy for another day. That's true. Very negative gross margin, right? Right. I get that. But again, I use the tools all the time. Like I was one of the first users of ChatGPT back in December of 2022, right after it came out, blew my mind, right? I thought I didn't believe that in my lifetime I'd see something that passes the Turing test. But pretty quickly, I got to a point where, yeah, this is kind of a curiosity. I don't see how it changes my life. Every time you use it now, You realize that it's gotten more and more powerful. And that's, again, I think that's the only thing that matters is that we think about entire professions that are now going to be very different. I'm not saying there won't be any paralegals or customer service reps or software coders. There will be just a lot less. And we in the office are going to be a lot more productive. You know, all of us, part of our job is reading and summarizing and synthesizing. Those things are going to collapse in terms of the time they take. And as long as that's the case, as long as this can make us more useful, it's worth pursuing. But I really don't want to gloss over the fact that there's not enough revenue being generated on AI right now. There's not enough cost savings being generated to justify anywhere near the investment we're making. So we will have to become more productive in order to continue to justify this. So that's one part of it. The other part is, there's this whole pursuit of super intelligence. That's really... That's driving a lot of this, especially if you're thinking about, why is Duck buying so many data centers? Why is Elon buying so many data centers? They really believe we are a couple of years away from superintelligence. Superintelligence being like a model that does everything we do. Not helps us be more productive, but does everything we do. And they will continue, and both of those guys have very deep pockets. And both of them are hell-bent on getting the superintelligence first. So they're investing that way. Again, Microsoft, Amazon, Google don't have a better use for their capital. So they're continuing to invest. So we still have some runway. But again, I'll concede that if this doesn't start making us productive, then at some point this well dries out because you can only be negative gross margin for so long. You know, $600 billion in investment just to send each other AI-generated emails. I mean, you know, I get the use case, but it's like, you know, how much can you spend on that? But I concede as well that, like, yeah, this stuff is really awesome. The technology is awesome. Maybe it's all going to be part of everyone's workflow one day. But again, the investment numbers, the CapEx numbers are so eye-watering that it just makes everyone pause. Investors are cool with it right now because we're just, the vibes are good. As soon as the vibes turn a little bit sour, that's what makes me nervous, is like, as As soon as the vibes turn. Then how much appetite is everyone going to have to just keep spending the hundreds and hundreds of billions of dollars? Are these are these leaderships of these companies, even even, you know, someone like Zuck can probably get away with it because he's the founder and he owns super majority voting shares. But others like is Andy Jassy going to be OK with spending 300 billion dollars if all of a sudden Amazon stock is not performing well and investors aren't happy? So that's what makes me nervous about the whole thing. Yeah, no. We should absolutely be nervous. Again, if we were just to go back, not that long, just to January of this year, when DeepSeek happened, and again, we thought, hey, bottles have just got as good as they're going to get, and now they're just going to get cheaper and smaller. That's all it took to cut NVIDIA's price in half, right? And to cause, by the way, to your point, Andy, Jesse, Satya, Nadella, and Sudhar Pichai to reconsider. They reconsidered, but again, they have a front row view to where the technology is and said, we're not going to panic. The technology is great. It's going to develop. And again, I can't emphasize enough. We do not have a better use for the capital. We can't do M&A. We're not going to be allowed to. We're already buying back our stock. We've plowed as much money back into the business as we need to without degrading our margins. This really is the best and highest use of our capital because we believe, those three guys, believe that this technology is going to accelerate GDP growth and generate tremendous returns for the economy. As long as they believe that, they'll keep doing it. Investor sentiment It's going to have to turn for a lot worse for them to not believe that. But I agree with you. Investor sentiment is fickle. That's why I point back to that January episode, right? We took a trillion dollars off NVIDIA's market cap because the VINs were bad. Oh, Those guys have a little bit of a longer time frame. That's an excellent point, and you make a good point. These companies have so much money, they literally don't know what to do with it, so they're like, ah, we'll just throw in some NVIDIA chips, God. which is why NVIDIA has been the biggest beneficiary of that. Gil, I know you got to run. Thank you so much for hopping on. It was a pleasure chatting, and hopefully we can do this again for next earnings season. Sounds good. Thank you. Thank you. Well, all right, guys, that's the interview for this week. Let me know in the comments on what you thought about the interview. And if you have any feedback for me, you know, I'm trying to get better as an interviewer because we're going to have a new interview every Sunday. So keep an eye on your podcast feed for that. We got some great guests coming up. Thank you guys so much for listening and watching. Shout out to Mike and Connor for all the help behind the scenes. And we'll see you guys back here tomorrow.