Watch the Video
Our conversation began with SoftBank’s sudden 40% drop from mid-October levels. Lee and I both agreed that few investors fully appreciate how central SoftBank has become to the global AI capital stack. The company owns stakes in OpenAI, has backed emerging infrastructure projects like the Trump-aligned Stargate initiative, and effectively repositioned itself as the Berkshire Hathaway of AI by reallocating its Nvidia profits into higher-risk frontier plays. Yet unlike traditional cash-rich tech giants, SoftBank must borrow to fund its commitments. That leverage profile is now front and center.
Institutional Exits From Nvidia
Lee pointed out that the real shock wasn’t just SoftBank’s liquidation of its Nvidia stake, but the fact that respected insiders like Peter Thiel also dumped every share they held. As I told him, this wasn’t a typical multi-quarter trimming that shows up gradually on 13Fs. These were full-scale exits, executed quickly and all at once. When prominent AI insiders remove exposure simultaneously, it raises the question of whether the long AI rush has started to hit valuation limits.
A Competitive Turn for Chips
I highlighted that Nvidia’s stock weakness goes beyond profit-taking. Alphabet recently introduced a high-end chip already deployed internally, and Meta has reportedly purchased large volumes. For the first time, cloud operators may have credible alternatives to Nvidia’s 90 percent data-center GPU share. If AMD, Amazon, Microsoft, and others introduce similarly competitive hardware, what was once a near-monopoly could turn into a multi-player dogfight.
Lee noted that Google’s chip has already proven itself in production environments. This alone destabilizes the assumption that Nvidia can maintain its lead across multiple AI cycles.
Pressure on OpenAI and the Model Layer
We also reviewed early press commentary that Google’s newest AI model appears to surpass OpenAI in certain applications. The combination of hardware and software pressure on the two largest AI winners arrived within days, suggesting that the competitive landscape may tighten far faster than previously expected.
Debt and the Data Center Bottleneck
I emphasized that the real systemic risk may lie in AI infrastructure debt. The assumption that data centers can be spun up quickly is running into reality: transformer shortages, multi-year timelines for power upgrades, political resistance to electricity load increases, and utilities reporting that developers are shopping multiple providers but committing to none. If data center construction is the rate-limiting step, then the AI buildout may face multi-year constraints.
Lee added that established players like Digital Realty and Equinix have a significant head start. But even they face power-supply bottlenecks when utilities cannot double electrical throughput fast enough to satisfy both cloud and AI demand.
The Trillion-Dollar Guess
We closed by acknowledging that every AI participant is making a capacity guess, often years into the future, and the cumulative investment could reach two to three trillion dollars. If demand falls short or infrastructure capacity lags too far behind, investors may see the first signs of overbuild risk across the ecosystem.
Transcript:
[00:00:00] Doug McIntyre: Lee this morning, SoftBank was down 40% from where it was in mid-October. Now for people who don’t know what SoftBank is, it’s the Berkshire Hathaway (NYSE: BRK.A) | BRK.A Price Prediction of ai. It owned a bunch Nvidia stock. It sold that to put money into open ai.
[00:00:16] Lee Jackson: Right?
[00:00:17] Doug McIntyre: Some of you may remember that, president Trump and the head of Oracle and the head of open ai were with the.
[00:00:23] Doug McIntyre: CEO of , when they had the big con grab SoftBank when they announced they were gonna spend $500 billion into AI infrastructure. So as far as I’m concerned, if you wanna know what’s going on with AI in the world, you look at SoftBank, not everybody’s familiar with it, but if I were everybody and I was in any AI, stocks, bonds, whatever, watch what happens at SoftBank, they have more skin in the game.
[00:00:52] Doug McIntyre: For AI than any company in the world. And they’re not as well funded. They’re not this huge cash machine. Like no Alpha powder or micro. No, not at all. So they’re gonna have to borrow money to do their piece of this.
[00:01:04] Lee Jackson: Well, that, that certainly shouldn’t be difficult given what we’ve seen in these circular financing world.
[00:01:09] Doug McIntyre: It it is, it is true. But I, I just want people to understand that if SoftBank is down 40%. You see people keep saying, is there an AI bubble? When are these stocks gonna go down? What I’m trying to tell for everybody is they are going down. , they’re SoftBank is SoftBank is just the first guy in the shooting gallery and 40% is a lot.
[00:01:33] Doug McIntyre: I mean, it’s like that’s, that’s, that’s means that’s dotcom. . 40% is dotcom numbers, so .
[00:01:41] Doug McIntyre: ,
[00:01:41] Doug McIntyre: It really is. I’m very worried. Very, very worried. As far as everybody here is concerned that you see that kind of hit in SoftBank, to me, that would be like seeing a 20 or 30% hit in, in Nvidia. absolutely.
[00:01:55] Doug McIntyre: It’s just, it’s just incredibly bad news. So please keep your eyes on SoftBank, any announcement it’s got because they are a, a big player. They’re at least a rook or a knight on the chess board of Oh, absolutely. Global AI investing.
[00:02:13] Lee Jackson: Yeah, I’d say more of a night.
[00:02:15] Doug McIntyre: More of a more of a night, not a rook.
[00:02:18] Lee Jackson: All right, let’s stick with that. Let’s, let’s get, I’m looking back on what we did before. Let’s get all the people that sold Nvidia, right?
[00:02:29] Doug McIntyre: so Nvidia ended up not being that popular stock with some institutions, SoftBank, which is the Japanese company that has ended up funding, we’re saying it will fund, a lot of these initiatives, including open ai, including Stargate, which is the Donald Trump $500 billion AI infrastructure.
[00:02:51] Doug McIntyre: They sold all, all of their Nvidia, mostly to fund their investment in open ai. And this is part of the nutty, singular stuff I keep talking about. So.
[00:03:04] Lee Jackson: Well, it, it’s, it’s just incredible because. as, as we’ve mentioned before, when talking about SoftBank, they are the preeminent, AI player, but why did they sell it all?
[00:03:17] Lee Jackson: Now? They’ve owned it for a long time and they’ve sold some in the PA past, but I mean, you know, typically you keep a little bit, don’t you keep a little bit in your hand, but so they sold it all, but who else sold it all? Peter Thiel. One of the most respected guys in Silicon Valley sold every share that he had
[00:03:36] Doug McIntyre: And also the chairman of Palantir (NYSE: PLTR).
[00:03:38] Doug McIntyre: So he’s right. He’s up to his neck in what goes on in ai. It’s not as if some super rich billionaire just decided to sell it. It’s one of the major global players. . And AI decided to sell all his Nvidia.
[00:04:07] Lee Jackson: But, you know, he nailed it all at once and this all came not long after the, Of big short fame, basically saying he owns, you know, he closed down his hedge funds, sent the money back to the investors and took his own, his, his own funds and bought a huge amount of Nvidia puts to go along with his Palantir puts, which I’m sure, Thiel didn’t like as much, and he was extremely negative.
[00:11:09] Lee Jackson: I mean, and there’s plenty of people in the data center game already that, that are way, you know, digital realty (NYSE: DLR) and Equinix (NASDAQ: EQIX) and you know, companies like that that are huge and all around the world. You know, they’ve got a quantum leap headstart on a lot of these AI ones, and, it’ll be interesting to see how, how it affects their overall, business because they’ve been outstanding for the last few years all above and all above. And, and it’ll be interesting.
Top Gaining Stocks
Top Losing Stocks