Invest Like the Best with Patrick O'Shaughnessy
· Gavin Baker
And then you get a bubble. That bubble funds the buildout of this new technology, but supply gets ahead of demand and you get a crash. It's a particularly severe crash if it's a debt-fueled buildout, like the year 2000. One thing that's really good about the current buildout is that it's still overwhelmingly funded out of operating cash flows, which is a really important fundamental difference versus the year 2000. As is valuation. As is the fact that every GPU is running at 100% utilization, when 99% of fiber was unutilized. There are all these fundamental differences. History doesn't repeat, but it rhymes. As an investor, we have to be very cognizant of it and recognize that based on the last 200 or 300 years — forget the internet bubble; we had a railroad bubble, a canal bubble, every kind of bubble, the South Sea bubble — we should expect a bubble. That's terrifying. Nobody wants a bubble. The reason it's terrible is, if you're valuation sensitive, you massively underperform. You get fired by probably all your clients. George Vanderheiden, who is no longer with us — a great Fidelity portfolio manager — fought the bubble in '99 and retired in early 2000 because I think he just couldn't take it. He knew it was wrong. His clients were deeply skeptical. "George, you're out of step." He had white hair. He's a truly great man. I only overlapped with him briefly, but he was a very important mentor and friend to my good friend and mentor, Jennifer Uhrig. So I have a lot of Vanderheiden DNA through her. He was the same person who said, "Being early is the same thing as being wrong." George retires because he can't take the underperformance and he can't take clients saying, "What's wrong with you? You don't get it." He has 40% of his fund in tobacco, 40% in home builders — and literally, he probably outperformed the NASDAQ by 20 or 30x over the next 3 years. I have been optimistic that this fundamental shortage of wafers, which today is controlled by Taiwan Semi, will prevent one. If Taiwan Semi did what Jensen wanted, I think Nvidia could sell $2 trillion of GPUs in '26 or '27, maybe $2.5 trillion, maybe $3 trillion. But there is a limit where consumers would consume so much, that you probably would be in an overbuild. So Taiwan Semi — if we don't get a bubble, we need to throw a party for them, because they will have single-handedly prevented a bubble.
· Gavin Baker, Patrick O'Shaughnessy
You are starting to see companies go to Intel and Samsung. [00:00:00] S0: Let's just assume TSM stays super supply-constrained versus the latent demand. What happens? [00:00:00] S1: Well, the history of markets is — I don't know who, but one of Intel or Samsung — they're not going to stay disciplined. They will break, and then at some level that will force everyone else to break. A lot of this may come down to the degree to which Taiwan Semi can maintain a lead over Intel and Samsung. You've got to remember, it's whatever it is — 9, 12, 15 months. [00:00:00] S0: The leading node edge, you mean? [00:00:00] S1: Exactly. The pace at which they expand capacity. If I were to watch one thing to understand whether there's a bubble, it's Taiwan Semi's capacity decisions. There's a Goldilocks zone where they expand enough to make it hard for Intel or Samsung to really truly emerge as an at-scale second source with something well north of 30% market share, and yet they also keep this fundamental constraint on wafers that helps us avoid a bubble. And then obviously the Terafab is going to play into this too. [00:00:00] S0: Say more about that.
· Gavin Baker
[00:00:00] S1: It's a SpaceX — I believe Tesla's involved as well — joint venture to build the world's largest fab, here in America. I think they're going to be successful. One, they have a partnership with Intel, which is very important because they're getting access to 50 years of institutional knowledge that's just 9 months, a few quarters, 12 months — 3 to 5 quarters behind the front. That's an advantage. It's also an advantage that the Terafab is going to get attention from the A-teams of all the semi-cap equipment companies. One big reason Taiwan Semi caught up is ASML and KLA-Tencor and Lam Research and Applied Materials — they wanted them to catch up. They don't like having a monopsony. The A-teams were in Taiwan working. Intel made some mistakes, and presto. So the A-teams will be here because of Elon's reputation in hardware engineering. To a degree that I think is maybe hard for people to imagine in America, where politics has replaced religion — because Elon had his foray into politics, that makes it hard for some people in America to see him clearly — which is sad because I do think he's probably doing more for America than any other American. He's single-handedly bringing manufacturing back to America, he's revived defense tech. SpaceX is in some ways the most important defense contractor in America. He's doing Starlink, which is amazing for the world. He's creating all these blue-collar manufacturing jobs, which is a goal of a lot of liberals and good for America. He's done more than any living human to decarbonize the world. And if you are upset about data centers on Earth for environmental reasons, well, here you go.
· Patrick O'Shaughnessy, Gavin Baker
[00:00:00] S0: Put them in space! [00:00:00] S1: It's sad, but he is a living deity in China, Taiwan, South Korea, and Japan. Having watched him for a long time, what he's going to do is recruit the best people, because the best engineers want to work for Elon, especially in hardware engineering. Next to the Terafab, there'll be a Taiwantown. "Oh, these are your favorite restaurants. I'm going to move them and their whole staff from Taiwan to Texas, and we're going to make everything the way they like it." And then we'll have Japantown — same thing. We're going to have Koreatown. We're going to have all these things exactly dialed to recruit the best engineers. That's just not the way that the people who run Intel and Samsung think. So he's going to have the best talent. He's going to have the A-teams at the wafer fab equipment companies. He has Intel, which is important. It's so good for any administration's political goals. And I think it's different enough that it will not alienate Taiwan Semi. [00:00:00] S0: And these have long lead times, right? Terafab is going to be pumping out whatever GPUs, whatever chips — quite a long time from now. [00:00:00] S1: We'll see. Elon tends to do things differently. Everybody else has taken 3 years to build a data center; he built one in 122 days. Samsung had to give him an office in their fab in Texas because he was so unhappy about the pace at which they were expanding and building.
· Gavin Baker, Patrick O'Shaughnessy
[00:00:00] S1: We'll see. Elon tends to do things differently. Everybody else has taken 3 years to build a data center; he built one in 122 days. Samsung had to give him an office in their fab in Texas because he was so unhappy about the pace at which they were expanding and building. [00:00:00] S0: Are you surprised by — you mentioned DeepSeek earlier. The simple reaction to that was, "OK, these models are just going to get 95% as effective for some tiny fraction of the cost. Distilled Chinese open-source models. We'll be able to use these for most of what we want to do. Fast-forward a little bit of time, 2 years from now, there's no reason I have to spend $1 million a year in my small little firm on tokens or something." But the actual reality seems quite different than this, and I'm curious why there's that dissonance in your mind.
· Gavin Baker
[00:00:00] S1: I do think it's fascinating — the returns to the frontier. All the economic returns to AI at the model layer — not all of them, but an overwhelming amount of them have been at the frontier, which is surprising to me. I think it's been surprising to a lot of people. This is one of the most important questions to be answered, and you need to have a hypothesis on it as an investor — are frontier tokens going to continue capturing the overwhelming majority of economic value created at the model layer? It is surprising. I remember when Gemini 3.1 Pro came out. It was mind-blowing to me. It was so good. Today, it's intolerable. There's probably a little bit of a dynamic where companies prototype with frontiers and then when they put something into production, you're hearing a lot of people use Vertex or open source. But still, it is a fact today that the overwhelming majority of these economic returns come from frontier tokens, and that's surprising. Whether or not it continues is a very interesting question. I'm much more open-minded to that, having had the experience I've had with Gemini 3.1, and then Opus. I do use Grok 4.3 a lot. It is on the Pareto frontier. The companies that are on the Pareto frontier — and this is, by the way, a big change and a consequence of what we talked about last time — Google losing their per-token cost leadership as a result of making very conservative design decisions with TPU v8 to try to take it away partially from Broadcom, and Nvidia continuing to make aggressive choices. But Google dominated the Pareto frontier — the Pareto frontier being intelligence versus cost. This is the most important thing to look at to analyze AI labs. Google dominated that 9 months ago. Every point on the Pareto frontier — OpenAI, xAI, and Anthropic were inside of them. Now the Pareto frontier is dominated by Anthropic and OpenAI, and then Grok 4.3 is on the Pareto frontier. It's clearly the best, lowest-cost 500-billion-parameter model. And then Gemini 3.1 is hanging onto the Pareto frontier. If I were to bet, I would bet that they're subsidizing that out of pride. A violation of Richard Sutton's bitter lesson is for sure the biggest risk to this trade, to all of AI. The closer someone is to AI, the more skeptical they are this will occur.