원문 보기
Let's dive into the earnings we got from Nvidia yesterday afternoon. Welcome in our next guest, Shai Boloor, chief market strategist over at Futurum Equities. Shai, great to have you back on the show here. I mean, we finally had the Super Bowl of earnings come our way, and the numbers were good. Uh we're just not seeing a whole lot of movement, and the movement we do see is slightly to the downside.
Yeah, I mean, just to quote Jensen, what he said this morning, the stock reaction is one of the biggest mysteries of the universe right now. And I think the stock is basically flat because expectations were already extremely high. I mean, that is the curse of being the best execution story in the market since the ChatGPT moment 3 years ago.
Uh that might sound ridiculous given how strong the numbers were at the scale that they're producing, but Nvidia is not just being judged as a normal company anymore. They The market clearly wanted a massive guide up or some kind of China surprise in the guidance or a major Rubin surprise or some kind of brand new catalyst to force the estimates meaningfully higher.
And to be frank, yesterday's surprise catalyst was $20 billion of CPU revenue for this year. And that sounds like a big number, but when you look at the run rate that Nvidia is producing, it's $400 billion a year now. So, it won't be enough to cut the um surprise factor for the next leg. But, I don't think the narrative should follow the price action with the stock because this is still an exceptionally run company.
I think that you can see in their earnings yesterday of how much they've graduated away from actually being a company that's not a GPU merchant to becoming a full AI factory supplier. And I think that the CPU was the biggest gem. I mean, that CPU expansion like $20 billion a year shouldn't be just uh glossed over.
Like, yes, that is just for this year, but Jensen talked about how this is a $200 billion TAM now in a market that the company has never really addressed before. Uh it doesn't suddenly mean that Nvidia is going to become a CPU story overnight. But, uh I do believe that it's showing proof that it Nvidia can continue expanding the platform uh as AI matures and we are seeing progress in that.
And I also think that physical AI still hasn't even begun playing yet and that is probably the next catalyst to really justify the next leg up in Nvidia's story. So, in every conversation that I've had about Nvidia, everyone has mentioned the CPU growth opportunity. They've mentioned the diversification, additional channels for growth here.
But, we're already seeing growth that doesn't have any even historical comps that I can find a basis for. As I look at it, has Nvidia just become a victim of its own success here? I mean, what is it going to take for the market to get back on the side of positive momentum when it has done nothing but beat and raise in recent quarters?
Yeah, that is probably what keeps Jensen up at night cuz he's seeing everything the way you would want it be as like the leader of the AI economy. But, the market's not rewarding the stock for whatever reason. It might be a six demand a six competition or you just don't know what the China unlocks going to be or if the Vera Rubin is going to be on track.
So, I do think that one of the biggest concerns entering this quarter was whether or not Rubin had timing issues or thermal design issues or some kind of road map risk. Luckily, Jensen directly tackled that issue on the call yesterday where Vera Rubin is going to be on track second half of this year.
Also, the CPU commentary extremely important that the market seemed to under appreciate. So, I think that this is just a classic example where the first stage of AI was all about training. It was all about GPUs, GPUs, GPUs and TSMC and Nvidia owned that lane. But, now AI has graduated into inference and inference is causing the expansion of opportunity in ways that the trained era didn't.
So now that the tension it for the market is almost like instant gratification. Like where's the next bottleneck? Because this game passed around from power power networking CPUs memory. So like everyone wants that next great bottleneck unlock and they're forgetting about who's actually laying down the groundwork for the AI economy and that's still in video.
So I think that Jensen should keep doing what he's doing. I think the market will reward it eventually cuz the data center demand keeps accelerating. I mean they just producing 92% year over year number and the market's going to keep asking whether or not hyperscale capex is going to sustain the pace.
We've seen no signs of slowing them down and also the second thing I want to talk about is the networking business. Like if networking keeps growing faster than compute the market will have a really difficult time re-evaluating Nvidia's just as GPU supplier. They're going to become a much bigger and greater AI factory company and I think that Vera Rubin's going to be that unlock cuz if it does maintain being on track for the second half of this year and that launch stays clean, that extends a road map and pushes out the peak earnings debate that I think the market's assigning the stock this year and I wouldn't be surprised if it gets a new all-time high once hyperscale release their next updated capex number to next cycle.
And I'm glad you brought up networking cuz that was up nearly 200% year over year. I mean that was absolutely one of the standout metrics if you were looking at the growth in terms of a percentage growth and also you know Nvidia's positioning to be able to offer the entire AI factory stack sort of as opposed to just the individual chips obviously a differentiator here.
So is Wall Street underestimating just how broad Nvidia's addressable market has become? Are we just overlooking a lot of these other avenues because of a historical story? I I uh mean the compute revenue is still the bulk of the revenue. Like, that's $60 billion growing I'm air quoting only 77% year-over-year, while the networking business is 15 billion growing almost 200% year-over-year.
So, as time goes on, you're going to see networking become a bigger bigger component of their overall business, and I think that's going to help them sell the story that they're not just selling accelerators, but selling the full AI factory stack. And the way that AI is moving where AI clusters are getting larger and larger, the bottleneck's going to become more than just the raw GPU compute.
It's going to become GPUs need to start communicating with other GPUs, CPUs, memory, storage, networking systems at massive scale. So, I think that if networking slows down, then the whole system underperforms it. Right now, we're in a prisoner's dilemma with the biggest companies in the world with the deepest pockets.
They can't afford to have any system underperform because they're in an arms race right now. So, the market can keep debating if Nvidia is just a GPU company, but the business is moving deeper into every layer of the AI system, and I think time will provide the solution on what the market is looking for that this is everything you want to be in as an AI winner.
And I think it's going to be the first 10-year that our company the world's ever seen. And a big wild card yesterday was China. And if we were going to get any commentary on China and sales to China and any inclusion in in potential guidance here, we didn't really get anything. You know, one, was that your expectation?
And two, how much upside optionality still exists if we do really see the restrictions ease, we can't actually start selling perhaps those H200 chips. How much of a growth driver is that actually going to be of a market that they once held the predominant majority share in? It would be a massive unlock.
I mean, this is the number two market in the world. I think that um China's basically, just like you said, upset optionality. Like, if you own Nvidia right now, that you have to treat China as the call free call option. Uh and that's why Nvidia kind of their guidance assumed no data center compute revenue from China.
I mean, if that market reopens, great, but I think Jensen's realizing how little control he has in opening that market and I think there is a little too much attention with something that's that small in the Nvidia story. So, I think he's just moving on to focus on everything else that's obviously humming in the organization.
Like, yeah, the numbers can move higher if that market opens, and that could be the catalyst it needs before the hyperscalers report their CapEx number. But if it doesn't, I mean, the business is still growing at a scale almost no company in history has ever delivered, and I think that it's going to be just fine regardless of what China market's going to open or not.
And so much of the reaction we see to Nvidia, whether it's earnings or a report or an investor day, comes from what Jensen Huang has to say. I mean, he speaks and everybody listens. So, in terms of the commentary that we got from him, both in the release and but more importantly on the call, you know, what stuck out to you the most?
What did you think was the most impactful thing that he said or the most telling thing that he said? He was like a little more revealing this earnings call. I mean, I Jensen said the demand has gone parabolic uh was one of the words he said because Jensen AI has arrived. AI can do productive work, tokens are becoming profitable.
I think that that last line I just mentioned right there is the most important part because if tokens are profitable, then the model builders have a reason to produce more of them. And if model builders need to produce more tokens, what does that mean? It means more compute. That is the loop. More agents mean means more inference, more inference means more networking, memory, CPUs, power, cooling, data center infrastructure.
That's why I just don't think the AI data AI infrastructure trade is anywhere close to being over. The cycle is broadening out, and I think that Jensen is highlighting the demand is insatiable, and it's not going away anytime soon. So, you can assign a peak earnings multiple if you want for this stock, but we're not seeing any evidence of that at all.
So, what's the biggest risk then to the Nvidia thesis right now? I mean, here's some things I heard mentioned to me today. Potential competition, regulation, AI spending, energy bottlenecks, execution, or is it just Nvidia itself right now? Just being a victim of its own success. I actually think it's uh it's going to be more policy driven.
I think this is my tin hat theory. I have no evidence of it. This is my gut, but I do believe that the midterms are going to have a net negative narrative on the data centers. I think that, universally, uh the Americans view AI as the bad guy. Like, means the rich getting richer, deep displacement of jobs, blue and white-collar, and I think that it is not going in the favor as the Trump administration would like to say on the midterms in a couple months, that there might be a break on the gas pedal on the midterm AI data center narrative, and that could obviously hurt names like Nvidia because they remain the center of that cycle because uh they essentially fund all the the whole ecosystem.
So, I think maybe possibly that's the reason it's not experiencing any kind of multiple expansion, but also you don't need the greatest chips anymore in this stage of AI. Like, inference is just 70% is good enough, and you're seeing a ton of momentum right now in Trainium. You saw what TPU did to take all the demand out of the oxygen out of the room when Blackwell was getting ramped up.
So, I think that there is finally new signs that there is competition, but Okay, so maybe 95% market share of Nvidia is in right now goes to 85%. They still have the majority of the pie, but the market is focusing on that slight decrease on market share, but in my eyes, 85% is a monopoly. But, I do believe that the risk is going to be more policy driven just on midterms.
But, even if that happens, it'll be a temporary bump along the journey of, again, them being a $10 trillion company in a couple years. I don't even think we have to put that in tin foil hat territory. I mean, I up until a few months ago, I had never said NIMBYism, and now I'm talking about NIMBY and NIMBYism, not in my backyard, relatively regularly.
So, there there could be some proof in that pudding. We shall see as we get further into the year. But, Shai, I really appreciate you breaking down Nvidia with us today and the reaction we're seeing, down about 1.4%. That was Shai Belore from Futurum Equity.
댓글 0