저자: DanIves · 외국인
요약 보기
Dan Ives는 빅테크 AI 설비투자(하이퍼스케일러 CapEx)가 수요를 기반으로 계속 확대되며, 2026~2030년 AI 투자 규모(JPMorgan 추정 5.5조 달러)와 수익화 진전으로 정당화된다고 주장. 메타·MS·알파벳·아마존의 AI 수익화 진전을 근거로 제시하며, 빅테크(MAG7)가 소외(penalty box) 상태이나 실적 시즌에서 검증될 것이라 봄. 메모리 가격 상승과 애플 가격 인상은 일부 수요 파괴 리스크지만 생태계 확장 대비 작은 세금 수준이라 평가. 향후 6~9개월 내 가격 안정화 필요. 중국 오픈소스 모델 확산에도 가치는 데이터와 엔터프라이즈 침투에 있다며 MS 승자론 유지.
다듬은 글
We're talking about hyperscalers in particular, which have been the huge losers until yesterday where you actually saw a little bit of a turnaround. Why do you reject the notion that people are pushing back against the spending plans of some of these tech behemoths? Look. The hyper is 700,000,000,000.
I mean, that's what's funding the AI revolution. I mean, when you talk about memory chips, NVIDIA, and everything else, but that's just the first phase. Because what the are doing is this is the build out. It's Vegas Strip Building in 1955. But, ultimately, the monetization now is gonna come. I mean, when you look at Meta, they're not just spending to spend.
You look at Microsoft. They essentially own the enterprise. Alphabet, 5% of their customers have gone to the AI path. Same thing with Amazon. So my whole point is you've had this tech rally, but MAG seven right now penalty box essentially. I think it significantly outperformed second half of the year.
I think earnings season, as we see in July, and there's been a huge validation moment for big tech. So the biggest trip maybe is a decent analogy on the just actuality of it, not in the scale. I mean, we're talking JPMorgan expecting 5 and a half trillion dollars to spend on CapEx spend for AI heading out to 2030.
We're talking about massive numbers that are accelerating, not decelerating. And capacity constrained companies. Right? I mean, some of these companies have said that we can't give you all the compute you're looking for. So how do you see them monetizing without raising prices enough to really trickle into the cost of what everybody is spending on an everyday basis?
Yeah. Well, first off, for every dollar spent on, let's say, an NVIDIA chip, there's an 8 to $10 multiplier across the rest of tech. So that's extremely important in terms of what you see in terms of you talked about some of the single stock moves, just the broadening out what we're seeing in terms of tech.
And you could argue for names like Caterpillar and others. But when it comes to hyperscalers, look. It's an arms race. That's why the question is, like, can they actually cut CapEx? They can't. In other words, like, you're diving in the deep end of the pool. Because if they cut back, then they go behind others in line that will clearly go ahead of them.
Anthropic OpenAI, ultimately, as they go public, they'll just have more and more cash. And I think that's the reality right now. And to to those, the bears are big. Oh, it's spending to nowhere. It's the opposite. I mean, you're essentially building out a new economy for consumers, enterprises. And for the first time in thirty years, The US is ahead of China when it comes to attack.
But what about demand destruction? Does that happen, especially when you see something like Apple come out and across the entire suite having to raise prices? Sure. Look. I mean, the there's gonna be negatives just like we see with memory because there's a limited amount of memory chip players. You can't just also So there is a breaking point, though?
There there is a breaking point to a point where, like, okay. If you continue at these prices, eventually, there's cannibalization. There's churn. For Apple at these levels, 100 bips of churn maybe. When it comes to iPhone, it'll be a 150, $200 price increase, but then there'll be subsidies from carriers and others.
But look. Now you actually you're seeing where it's a call a or what we're seeing in terms of memory prices. That is gonna have an impact, but I view it as it's paying a small tax on the broader ecosystem that's being built out by big tech. But if there is a little bit of a breaking point, when does it happen?
Look. I think in the next six to nine months is where you need to see some sort of stabilization when it comes to prices. You need to see scale. You need to see ultimately what we're seeing in terms on to on the token side where you need to see, like, prices start to come down. CapEx would be about trillion dollars for big tech next year, but then the monetization starts.
So, essentially, you start to actually now see the monetization piece, which is very important to just investors feeling like Meta, they're not just spending, like, nineteen eighties rock stars for no reason. There's one story that really caught my attention over the past couple of weeks. It's the idea of Microsoft shifting to Chinese large language models that are open source and that are cheaper.
It's the idea of OpenAI pushing back their, IPO and a question of how much OpenAI and how quickly the Anthropics of the world can monetize given the cheaper competition overseas. How much do you see that as a pressure point that hasn't fully been realized or even actualized in terms of people's ID ideas?
Look. My view and and I think you see with OpenAI and Anthropic, the models eventually will become commoditized. And it's not just OpenAI and Anthropic. You'll have hundreds of models across the world, vertical specific. The value will be in the data, how quickly that they could penetrate enterprise.
Microsoft, in terms of Nadella, Nadella is seen around the corner understanding, like, you have to ultimately diversify when it comes to miles and eventually as it becomes a global sort of game. But when it comes to Anthropic and OpenAI, they understand enterprise is golden goose. That's what they're going after.
Now to some extent, part of the reason that Microsoft do you see the stock the way where it is? Because there's a view like, okay. That becomes competitive. They had OpenAI, but now, like, where did they sit relative to enter my view is on enterprise, Microsoft will continue to be the ultimate winner, but it just shows it's a convergence what's happening across the AI ecosystem.
If that's a commoditized resource, the large language models, and it really is the data. The data providers are not open AI and anthropic. Well, then what's the monetization prospect for some of these companies that have potentially are gonna IPO at a trillion dollars? Well, for them, it's all about mean, if you look at growth, it's been unlike anything we've ever seen in terms of ARR.
Now for OpenAI, it's now it's a bull's eye on their back. They have to continue monetizing the enterprise because enterprise when you think of all the data centers being built and all the memory and everything, it's being built for what will be the use cases on enterprise, on the consumer. Software vendors, in my opinion, are the ones that it goes back to second, third derivative just like cybersecurity.
So I think part of what I think investors, and I don't think it's reflecting their saying the stocks is, like, they're underestimating the scale and scope of what this is all gonna look like. And I think we're just still we've talked about it. It's like, party start at 9PM, goes to 4AM But about 11PM.
Can you monetize a model that's currently restricted? No. And that's why the re when it comes to anthropic, that's why they're playing they're starting to play nice when it comes to government. You're seeing some sort of, you know, compromise or at least some moving of the goalpost. And that's important because anthropic right now, I mean, it's you're talking about the best model in the world.
You cannot be on relative to what we see on the DOD and the Pentagon. That's why it's an important time for Dario and Anthropic to get to some sort of compromise when it comes to US government.
Comments 0