Welcome to the Risk Reversal Podcast. I am Dan Nathan. That is Dan Niles. He is the founder of Niles Investment Management. Dan, welcome back to the podcast.
Pleasure to be on again, Dan.
All right. Well, I just heard from Amanda. This is our lucky number 13 in nearly six years. I really, again, appreciate — Guy appreciates, our whole team appreciates — how you've just been that giving of your time over all those years. So, I appreciate it. And not only that, I think our listeners, I know our listeners really appreciate it, because you're a very, very unique guest for us. You were in the tech industry in the '90s. You went to the sell side. You were an II number one ranked computer hardware semiconductor analyst, and you've kind of worn different hats. And then you obviously went on to be an investor. Your approach and I think your opinions are very unique, and we really appreciate that. So, I'll just leave all those accolades right there.
Let's hit it — this is Thursday afternoon, and you're probably listening to this Friday. We had the Fed announcement and the presser Wednesday afternoon. We had, I think, somewhat of a muted reaction in the bond market and the equity markets. But today is a different story, and the market's kind of raging. We have a VIX at 15 and a half. We still have crude oil above 100. The stock market doesn't really care. So, let's talk a little macro, and then let's talk some tech. And you've had some really, really timely calls this year, nuanced calls within tech and within the AI kind of ecosystem. And I think it's really important for our listeners to remember, it's not just me talking about it, because you've been on — this is the second time this year. You are amazingly transparent. You put out very long tweets, I think on Sunday nights for the most part, kind of dictating or detailing where you are in the markets. And it's not just thematically, but you're also talking about individual names. So, let's just start on the macro, because I think, again, coming from where you sit, I think it's really helpful. You're not a big macro strategist, but it's a really important part of your thematic investing.
I always say you don't want to fight the tape. And seasonality works for a reason, right? Where if you look at the data and you go, 'Wow, this is the way it's acted over the last 60-70 years' — I tend to look at the S&P stats from 1957, which is the inception of the modern-day S&P, and then I'll look at shorter-term periods, and I'll go, 'Wow, if this is kind of the seasonality, or these are the patterns, then I don't want to fight that.' And I don't want to fight the Fed, right? That phrase exists for a reason. And I definitely don't want to fight the bond market. As the famous saying goes, when I get reincarnated, I want to come back to the bond market because I can bully anybody. That's the James Carville famous quote. And so right now I feel like you have all three of those things — between the Fed, the bond market, and seasonality, with September being the only month since 1957 that's down on average, and it's down more often than it's up. All that's a problem with bond yields hitting multi-decade highs across the globe in some cases.
For the listeners who might have to go Google who James Carville is — he was the campaign manager for Bill Clinton going back to 1992, and that quote has been famously used many, many times. But I think the way in which you are using it right now is really important, because we're entering this midterm election in about, what, five or six weeks. And let's be clear: what's going on as far as the economy is concerned is probably front and center for most voters, right? And so when you think about this whole notion of yields being at multi-decade highs and a president who's not on the ballot but really would like to see lower yields, which doesn't make sense to a lot of sort of macro strategists or economists — how are you thinking about a 5% 10-year? And this is something that even with Fed funds going back up a quarter basis point — this is obviously yesterday — and we're at 4%, I think the highs were 5.5% before Powell started cutting. What does it mean for the stock market, because the stock market is within 2% of all-time highs? The stock market has been going higher as yields basically have been much, much higher than they have on average since the financial crisis. So if the stock market is not particularly bothered by yields at 5% and going higher, what would be the thing other than seasonality that could make a deeper sort of drawdown from where we've been? We've had a couple big ones, but it was tariffs, and then it was the war, and we came back fairly quickly.
Well, the big one's data center, right? Because AI has really powered the markets this year. If you look at the semiconductor index for the year, it's up over 60%. That's somewhere in the low-teens percentage of the S&P 500. And so you take that out — that's a very small sector of the overall market — and you've got very minuscule returns for the rest of the market. And so what's going on in data center? Well, what really surprised me and started getting me more negative is several weeks ago, you may remember that the state of Texas, which is a big proponent of data centers, they went ahead and started to put a moratorium on data center expansion essentially, and connecting some of those data centers to the grid. And that's really unusual because it's a deeply red state and they've been very supportive. It's the second biggest data center market behind West Virginia out there. And so you had that kind of kick in several weeks ago, which took most of us by surprise.
The second thing is very recently — you saw this last weekend — you saw Sam Altman of OpenAI, Dario of Anthropic, and Elon Musk all talk about the fact that they needed to slow down the pace of innovation. And so you go, 'Whoa, okay, well that's not something you expected to see in a broader sense.' And we can talk about that a little bit, but that's not great either.
And then the final piece of it, which so far hasn't been a problem, but I think it could get interesting. Part of the reason I put out a piece on June 20th when I said, 'Hey, I think we could see a speed bump in the AI trade' — and it turned out semiconductors went down 29% from June 22nd to July 29th, so it was one heck of a speed bump. But the problem back then was you had customers saying, 'Hey, we've been using a Ferrari to go to the corner store to get milk, and we should be using a Ford, right?' And so people started to switch from the expensive Anthropic, OpenAI models to the cheap open-source, open-weight models. And you've seen the cost per token being charged on average go down 50% since the end of May till today. The good news is that the number of tokens being produced quadrupled. And so that wasn't necessarily an issue.
But within that, what you're seeing is that the open-source, open-weight group of models — and that's mostly Chinese, with some US stuff in there like Neotron and what Meta just launched — you've got the volumes really accruing to that group, and you've got some profitability metrics that seem to be moving around. Where I saw something that supposedly OpenAI's profitability actually got worse. They cut, as you may be aware, they cut their leading model's price by quite a bit. Looks like they're taking a lot of share as a result from Anthropic, but as a result of that, it seems like the profitability metrics are getting worse. So now — these are all private companies, so you're kind of picking up tidbits that you hear floating around when they're trying to raise money or they're speaking somewhere.
So, it's hard to figure out the exact data, but there's some questions around some of that. And then there's some architectural shifts going on on top of that, such as separating out the cache from the inference piece. And anyway, I don't want to get long, but there's a whole bunch of other stuff going on that you go — the AI trade, between the politics of the data center pushback, and then the leading guys, Anthropic and OpenAI, talking about slowing down the pace of innovation, and then some of the architectural shifts going on underneath it, which maybe reduce the need for some levels of compute. You better hope that that Jevons paradox, which everybody keeps quoting, continues.
And part of this, which is getting interesting, is if you want to really dig deep: the top 1% of users actually saw a 10% reduction in spend from July to August if you look at the Ramp index.
And that's been pretty much straight up every month since ChatGPT showed up. And that's only the 1%, right? The other 99% of us, our spend keeps going up. But you always want to look for leading indicators, and you go, 'Wow, that means that guys are really starting to use the cheaper models from ChatGPT, as well as more importantly open source in general.' And at the end of the day, it comes down to profits and revenue growth for the AI space.
By the way, how cool is that Ramp index? I mean, talk about a company that is absolutely killing it on so many different fronts — but coming up with that index, and obviously pulling customer data but doing it in anonymous fashion. And we know that a lot of their customers are kind of cutting-edge tech firms when you think about it. So very cool. Good on you, Eric Lyman. Let's go — we'll come back to the AI thing, because I think we got there because of its contribution to growth, right? And that's it. But let's go back to this notion of not fighting the Fed and not fighting the stock market. And let's start with the Fed, because I think your point that you came out with about the stock market — you ended up being very right throughout the summer, and you're right because of the leadership sort of failing, or at least kind of, I don't know, like the passing of the baton a little bit. But let's start with a 5% 10-year, and then Fed funds, which feels like it's going to be, I don't know, four and a half percent or something — at least that's what the CME Fed funds tracker is indicating over the next six months or so. We're not going to get a raise in the October meeting a couple weeks or so before the election, but it sounds like once again it could be that December is live, because there's not a meeting in November, I think.
Yeah, that's correct. And I doubt the Fed's going to want to raise right in front of the elections, because the amount of blowback they're going to get from the White House would be ridiculous, right? It's like something like five days before the midterms. The next Fed meeting. So then I think the next one after that is December 10th, at which one I think they will in fact raise, because a quarter-point raise doesn't do anything, right? You have to actually... And the thing I thought was interesting is one Fed official talked about getting inflation down in a timely manner, and inflation's been running above the 2% target now for 65 months actually — now it might be 66. And so we're nowhere close, and so you should expect a series of rate hikes. The market just, based on history, underperforms when the Fed is hiking. That's why the phrase 'don't fight the Fed' comes along.
And part of that also ties back to the AI trade, because when you've got 10-years sitting up at 5% — don't forget these companies that didn't used to ever need to raise money. Google went cash flow negative for the first time since they went public. They're raising debt, they're raising equity — all of these companies are. And so they're now competing with the federal government that's running a 6% deficit relative to GDP, which by the way is the highest deficit in a post — without being in a major war — that we've ever seen. And so none of this is good when you look at it all together.
And you know, investors have sort of — I think unless you lived through and invested through the global financial crisis, which at this point is back in 2007, so 20 years ago — and you had money at play where you lost a bunch of money. Everybody seems to believe it's always buy the dip, the market can never stay down for a long period of time. But that's just because you've been able to throw money at the problem during COVID, and so you've powered through this. But central banks and governments, which are running out of ammunition, which is part of the reason why the bond market's doing what it's doing, that creates an issue. And then throw on top of it: if you're sitting in Japan and you're a saver, you haven't had a good yield in forever. And right now you have yields the highest since the 1990s that they can invest in. And in the US, you're going back to that 2007-08 period in terms of where yields are here. And a guaranteed 5% return over 10 years at these valuation levels — which on a GDP versus market cap basis, you're tracking to one of the highest levels since 2000 — it's not that inconceivable that that's a good alternative.
All right. So, let's take a beat here because I think this is super important, right? So, when you think about — you started with valuations, you talked about a five-year, a 10-year, a 5% 10-year Treasury, right? And all of a sudden you say to yourself, well, you talk about Japan having the highest yields since the '90s. Well, you have two situations here. You have an equity risk premium here in the US which is basically non-existent, right? And we just went through this a little bit — a lot of the powering of our economy, but also returns in the S&P 500 and earnings growth, has had to do with the AI trade. But if investors had this carry trade on, where they're borrowing in yen and they're buying long-duration assets here in the US, if that is coming unwound, that could be a real big problem for the equity market. Does that make some sense to you? Because yes, I know that you made a tactical call about your equity allocation, right? But is there some sort of risk over the next — and you just said this — we have a rate hiking cycle, like that seems pretty clear at least from what we heard from the Fed. It was a unanimous decision to raise 25.
That shocked me. It was unanimous. I thought there'd be at least one dissenter — and you know who didn't want to do that. But the fact that it was unanimous I thought was very interesting. Also, it's a rate hiking cycle not just here but in Japan, at the ECB, in the UK — so you have this going on across the globe. And people sort of have it's like, 'Ah, no need to worry, it'll all be good,' and that makes me very nervous. Because to your point, my thought had been — and it's sort of starting to evolve, and we'll see — is that after the midterms we should be okay. But then now one stat, and I've written about this, that you have to take into account is from July 31st through November 9th, since 1990 — which has the Gulf War in it — the average drawdown from July 31st to November 9th is about 10% from peak to trough in the S&P. That's double what it is in non-midterm years.
So we're nowhere near that either. And I think about what's going on — what we've seen politically. Right now you've seen the Socialist Democrats pick up a lot of seats in different states, etc. And if you look at their platform, by definition they are anti-big business, and they're really anti-data centers, right? And so I sit there and I go, well, do things get better after the midterms, or if you have a whole bunch of seats picked up by the Socialist Democrats and Democrats in general, does that actually get worse for data centers?
And so, you know, it's something I'm trying to think through, combined with the carry trade, which — they called it the widowmaker trade, right? Trying to short Japan, because it just hasn't worked. And now it's starting to really work over the last year or so. If that becomes a much bigger trend, then you have a much bigger problem going on, especially with valuations sitting where they are. And to be clear, what I'm talking about is 10-year valuations, market cap to GDP. If you look at some of the individual names like an Nvidia, you go, well, if the estimates are right, this thing is ridiculously cheap, trading at a mid-teens multiple for calendar '27 earnings. But you also go, well, are a lot of companies over-earning, such as in memory is a great example. And I don't think this is the top, but there's obviously a lot of people who do focus on the circular financing and everything going on, and over-investing by all of these companies, which is definitely happening, and go, 'Well, we've seen the top, not just a speed bump.' I'm in the speed bump category, but I want to keep my mind open that I might be wrong. But I think we're just so early in the agentic AI trade, which really started only on January 30th with OpenClaw, that I feel like we still have at least a year or two to run.
Let's move back to the data center build and the moratorium, and I think the doom sort of stuff. Do you think that Musk and Dario and Sam Altman really believe what they said over the last week or so — that they want to slow the pace, or they're happy to slow the pace of innovation, because of the concerns around what this technology might do?
Well, remember two things can be true at once. You can be concerned about what this technology can do, and at the same time you can be wanting to have a government-sanctioned duopoly if you can — and both of those statements can be true. And then don't forget both OpenAI and Anthropic are trying to come public, and so if they can show better cash flow because they're spending less, that's going to help them being able to come public. So I think you've got three separate motivations all at the same time. Which is: Elon Musk has been concerned about AI wiping out humanity for, I think, a decade, if you're not careful. Very true. You can also obviously — if you could have a duopoly between OpenAI and Anthropic being that far ahead and choke off open-source, open-weight, they would love that. So that can also be true. And then finally, you want to get a great valuation, and so you want to cut back on your spending by saying, 'Hey, we're doing this for safety reasons.' That can also be true. So you can have multiple motivations all for the same thing.
Yeah. And I think that's such a great point. I mean, if OpenAI and even Meta, obviously Anthropic, but if a lot of these hyperscalers or these huge frontier labs, if it was kind of some sort of phony sort of — I don't know what you call it — slowing down the pace, and they stop spending to the degree in which they have been, it might be something that investors might think about somewhat favorably. And I think Meta, and I know you have...
They won't be thinking about this favorably, let's be very clear about this, because what's going to happen is it's going to start showing up in different pockets of the AI trade. Because don't forget, capex last year was up like 75%. This year it's up like 100%. If that slows — and the analogy I would take is 2001-02. It's not like the internet stopped growing in 2001-2002; it doubled in each of those years. It's just it wasn't doubling every three to six months. So you don't need it to contract; you just need the growth rates to slow. And when you've got companies saying, 'Oh, you know, we think we're going to be great through 2030-31,' for me, I remember those same forecasts being made in 2000, and we all know how that ended up.
Well, that is part of the reason why I said in a CNBC interview — I think it was on Monday — like, the market can keep pitching, but you don't have to swing. And to me, the downside risk, if something occurs, is much greater than the upside risk. And you can trade the volatility, because right now it's terrific — you've got stocks moving up and down 10% every day. But I think you want to be a lot more picky, because there are things that can go wrong. Like one example I will give: everybody's been very bullish on, 'Hey, we can't overbuild because we have power constraints and land constraints.' Okay, if you're telling me power and land is what is dictating how much you can spend — right? — then how is bipartisan pushback on data center spend and connecting to the grid? That means by definition you're going to have less spend, right? You can't have it both ways. So there's a lot of nuances in here that are all happening at the same time that you have to pay attention to.
Let's talk about a deceleration of the spend. And so we've seen it reaccelerate to your point. We've seen Amazon, for instance, double their capex over the last year or so, right? And we've seen Meta — they keep growing at a pace. You've seen a deceleration as far as Google and Microsoft to some degree, but they're still growing a whole heck of a lot. If you do see some sort of hiccup — and maybe it's power, or maybe it's land, or maybe it's policy, whatever it is, maybe it's demand, maybe it's a bit of all of them — what gets hit first? And we're already seeing semis, for instance, they're down 25% from their all-time highs made in July, but the SOX doubled from April to June, so you could say to yourself we're still doing pretty well.
And you also just mentioned that we're seeing these stocks, from a trading perspective, maybe it's a great environment, because you have up 10, down 10%. Intel, as we're talking right now, is up 10% now; it was down 35% from its recent highs to the recent lows. But on a day like today, it's up 10%. Why? Because SK Hynix might be giving them a contract for their Ohio fab — you know what I mean? That actually won't be built out until 2030 or something like that. Why would you bid up this stock, that market cap right now, based on that sort of headline? Am I framing that correctly in a way? Because to me, it just feels like investors are going to just trade off every headline that's out there. I mean, help me think about that, because if there is a pullback in spend, if you are AMD, if you are Intel, right, and you're not Nvidia right now, like you're dead. If you're Oracle, if you're one of these neoclouds, right, you're dead. If you're one of these memory names, it's going to be really hard for you to continue obviously to get the orders and for folks to kind of think about what the backlog is. If I hear about these backlogs one more time, and we're talking about hundreds of billions of dollars, there's a good chance that those don't get realized at some point over the next five years. Am I just thinking about this incorrectly?
No. I forget the exact thing — I remembered this and I've forgotten the exact figures, but if you go and look at the May 2001 Cisco press release when they released earnings — and you can find it online — in the commentary they go, 'We went from' — I'm going to get this wrong — but it's something like they went from orders being up 50% year-over-year to down 30% in several months. It's something like that. I've probably got the exact percentages off a little bit, but I think that's what it is. I remember when the press release was though: it was May of 2001. You can find it on their website.
And for those viewers, Cisco was Nvidia back then. It had the biggest market cap in the world. And so for them to come out and say that was a big deal. And so this backlog stuff is ridiculous, because it can change very quickly. And when you can't get what you want anyway, you're going to double and triple order. And that's what's going on. Because if you fundamentally believe — and I do — that AI is the most transformational technology we've ever seen since the internet, then by definition you're overbuilding. I mean, Microsoft said it themselves, right? The CEO Satya said, 'The biggest risk is underinvesting, not overinvesting.' And so all these companies are overinvesting, knowing that if at the end of this they end up being the dominant company, they're going to make a fortune. But you're also going to have train wrecks along the way, right? How did AOL do? Not so good. Google came out of that. Similarly with Meta coming out in social media, right? They're the only...
Dominant person there, right? And you can go through that list. How about e-commerce? Well, WebVan certainly didn't make it. Neither did Pets.com, but Amazon stock went down 95%. But ultimately, it came out of that. So, that's what you have going on right now. What I'm trying to say is, I still think we have another year or two before we hit the top.
Yeah, but it doesn't mean that — you know, there is a camp of people that believe we've seen the top, right? That circular investments coming back to roost, and bond yields going up where they are — when you have to raise debt now, it's a lot more expensive than it used to be. This talk about pacing by the big, you know, Anthropic, OpenAI, certainly not helping, and some of the architectural shifts. But thankfully, the one metric underneath it, which is revenue growth, seems to be working. The bad news is a lot of that seems to be accruing to open source. So, a lot of things to be balanced.
You know, you talk about Google — it was born out of the, you know, the .com bust. Amazon went down 90-some percent from its all-time highs, but it's the dominant e-commerce company now. So we could go probably a handful of others, right? Do you think this is going to be a winner-take-most like we've seen in digital, you know, in search? You do. And right now it's just a two-horse race as far as Anthropic and OpenAI, and that's it.
No, I don't think that they're going to be the winners. I think Anthropic will do very well in enterprise because that's what they focused on from day one. Enterprises will pay, right? OpenAI focused on consumers from day one. Consumers don't pay. They're getting stuff for free, right? You put in a search for Google, do you pay? No. And that's when they made the switch to enterprise. That's why Anthropic is bigger in revenues today than OpenAI is, right? The last figures we have is Anthropic's at 65 billion in revenues and OpenAI is at 40 billion. We're not sure exactly how they're counting the revenues, right? In terms of, you know, the infrastructure costs, etc. But by most accepted measures, Anthropic is now much bigger than OpenAI — earlier it was the other way around and OpenAI was much bigger than Anthropic. But it's that focus on enterprise. So I think Anthropic will do well. They also call it profitable adjusted operating profits, and I hate that term, but supposedly they got to adjusted profitability in Q2. Supposedly that's improving in Q3. And so that's good for them. I think Google is going to be ultimately one of the big winners in consumer. And one dark horse that I wrote about a few weeks ago that continues to be gaining momentum is Meta.
And why? Well, if you go to the app store, you'll see that the Meta agent is the number two downloaded free app behind ChatGPT. And Meta's valuation is at a market multiple right now. It's gone up a lot. Now it's gone back to a market multiple, but their peers are sitting in the low 20 times P/E ratios. And they introduced — like, their big issue obviously is that originally, all the spending on capex, the only way they were able to monetize it was improving their own business. They had 3.6 billion daily active users. If all of a sudden people say, 'Hey, I want to access and do a lot of agentic things using the Meta application,' they could be a dark horse winner.
So, I think it's early to figure out how this shakes out, but people get into habits, right? If you're used to using a certain social media app, that's probably what you're going to do. If you're used to using Twitter, that's what you'll do. Instagram, that's what you'll do, etc. And I think with the chatbots or agentic AI or whatever, people will settle on, 'Yeah, I'm a ChatGPT user or an Anthropic user, or maybe I end up using the Meta Muse agent.' I think this will end up being like every other major technological shift, which is ultimately it coalesces down to a few players. And if you look at the internet, as we just talked about, you have great examples of that, which is Google dominant in search, Amazon dominant in e-commerce, Netflix dominant in streaming, you know, Meta dominant in social. Like, who's the number two? How far behind are they? So even if I say I'm going to have three winners, that's still more than we had post internet meltdown.
You know, the Meta thing, and I know you wrote about this, and you know, talk about catching a move — you know, when I think you probably looked at it from a valuation standpoint.
Yeah, from a product standpoint, it's like Google, right? It's exactly like Google. If you go back, Google was struggling, then the judge came out in that ruling and basically slapped them on the wrist. They introduced a model that was great, and then the stock took off. The same thing with Meta: they settled with the attorneys general on this whole, you know, safety thing with youth. Then they introduced an API. So that then got the move going a couple of weeks ago. Then last week they introduced their agent for consumers. That kept the move going. And now we're hearing how well it's doing in terms of people adopting this thing, and that's keeping the move going. And it's still got a low valuation. So, it's sort of got the holy grail of it can be a value stock, a growth stock, and an AI play all at the same time.
And the stock's up like 1% year to date. It's not because they have been struggling for a while. And then Watermelon coming out, which is supposedly a model that's going to challenge at the frontier, and that's going to be out at the latest in October. And they have this event, I believe, next week with the Meta Connect conference, and you might get a preview of that. So, those are all things that make, you know, it's important when you catch these stocks when there's a fundamental shift. So it wasn't a valuation thing I was looking at for Meta. What I was looking at was they can do what Google did a couple of years ago, right? In 20 — I think it was 2025 actually.
And you settle a legal overhang, and then you introduce good products, you can rerate very quickly.
All right. Let's talk about Google for a second here, because I think one of the narratives was the vertical integration that they had, right? The TPU was another leg of that story, right? So you have GCP, you have the TPU, you have, you know, a lot of Nvidia's customers looking to diversify away from them, right? And so that became an opportunity for them to sell that outside of what they were using. Um, you know, Gemini — you talk about Meta and you talk about that user base. I mean, you think about what, uh, Chrome, you know, it's like dominant market share as it relates to browser. You know, we talk about YouTube, you talk about all those products that they have over two billion daily actives also, right? And you say to yourself, okay — have you looked at this Dream Beans? It's really cool. Um, you know, they're, they're, uh — have you seen this thing yet or no?
No, I haven't seen that one.
Check it out. There was an article in the Wall Street Journal just a couple weeks ago, and it's like a really cool application. And I mean, if you're okay with it listening to you, looking at your email, all that sort of stuff, it's creating some really, really cool stuff. But my point is that they've been able to demonstrate, despite Gemini not scoring the way, you know, some of these other models have — I mean, to me, I just think that's an unusual value. I think it's unusual distribution, or not unusual, but you know, something that you can make the case not too different than Meta. But you know what? Meta's gains have kind of been Google's losses over the last, call it, two months or so, really since earnings season going back to, you know, the Q2 print. So I'm just curious how you think of, you know, Google right now, because I feel like they have a lot of the attributes, let's say, like a Meta does right now, but it's not being rewarded for it.
Oh, I love Google because, you know, if you look at the fundamentals of their business, they saw the most acceleration of any of the big hyperscalers, right? Amazon went from growing 28% to — AWS, excuse me, went from growing 28% to 37% sequentially. Azure went from 39% to growing 43% sequentially. And GCP, it went from growing 63% to 82%. The reason the stock didn't act well was they reported before situational awareness got unwound in their public portfolio — they had really massive acceleration in their cloud numbers.
So Google, and this kind of goes back to even Meta, right? Meta used to have the leading open-source model back a couple of years ago, and then DeepSeek came out. Meta ran into some troubles. Then Mark Zuckerberg went out and acquired Alexander Wang and his company, and now they've come out with a new model. It wouldn't surprise me to see Google back in that top-tier rank soon as well, because don't forget, these guys invented the transformer technology that OpenAI went out and then commercialized first. They already had it.
So this is going to be a continuation of leapfrogging back and forth. And I go back to what makes a model smart: it's the amount of data you have to train it with. Google, because they have all of these free products that we then give up the ability so that they can train on our data — that's what gives them a fundamental edge going forward. And I see this as temporary, that they're not on the frontier. And it's not going to surprise me if in a few months we're talking about Watermelon from Meta at the frontier, and then Google introducing whatever they have coming next, because Sergey and Larry, the two co-founders of Google, are still very intimately involved, and they've been pushing pretty hard on making sure they're at the frontier. And I certainly would not count them out because of that fundamental advantage with training data that they have.
All right, let's talk about the circular financing thing. And you and I haven't talked about this in a while, but you know, just think — I mean, we can go on and on about this. Microsoft's the largest investor in OpenAI. OpenAI is one of Azure's biggest customers. Uh, Nvidia obviously, um, you know, not obviously, but they invested in OpenAI, and OpenAI is one of their biggest customers. And then OpenAI, you know, has like thrown out these contracts all over the place. You know, one of them was Oracle, right? I mean, the list just goes on and on and on, right? Like you think about this, so—
Quite honestly, it doesn't matter until the spending starts to slow down. You had circular financing back in the 1990s as well, and you had some, you know, some nasty dips along the way for Cisco. You can go back and look — there are multiple, or at least two, 30%-plus drawdowns in Cisco along the way. And so the circular financing stuff's only going to matter when things start to slow. And that's why this whole thing of is OpenAI and Anthropic talking about pacing — is that something that's actually going to happen? I don't think it can, by the way, and we didn't finish flushing that out.
Because if you slow down and Google keeps going, you're spooked. So, I don't think any of these companies is actually going to slow down, I think. But if you're getting ready to go public, that introduces a new wrinkle where, you know, do you slow down for a quarter or two just to make your cash flow look bigger, right? To make your expenses look smaller? And then does that cause some, you know, issues?
Let's talk about Microsoft for a second, because you just mentioned Azure and the reacceleration that they had in growth there wasn't nearly the magnitude of what you just mentioned about GCP, right? And Amazon saw a really nice reacceleration. You know, Microsoft was trading very poorly, and the stock after that print, you know, it went up, I want to say, 20% in a straight line over like a week or two, right?
Well, that's because they didn't raise capex. Yeah, had numbers go up and capex not go up — that's the big difference. And also when they report it. So I think, because I'm pretty sure they reported after situational awareness got, um, you know, got —
