Speaker 1: I want to start with you, Josh, and I'm going to hear from you and Bryn first, because you're our longest owners of NVIDIA. And I'm wondering what you make of today's sell-off, how you're putting it into context, the kinds of questions you want answered, and what all of this means to you.

Speaker 2: I think there's a component of today's sell-off where people are very reasonably recalibrating their expectations for how big the total addressable market for new GPU sales is. But then I think there's an even bigger component that's just speculators being unwound. This is the most over-owned stock in the market, and for good reason. People have absolutely gone crazy with options trading here, with leverage. We've got vehicles that are 2X NVIDIA. We've got all kinds of zero-data exploration, weekly options. It's an absolute casino that's been built on the back of a legitimately amazing growth story. And so I think if you didn't have that circus in town, you'd probably see the stock down 5% to 10%. But because you have all that leverage, you're seeing much more. I think there are two very important things that people need to know about what's happening with DeepSeek AI and the way it's being interpreted on Wall Street. The first is it doesn't matter if it's a Chinese government PSYOP or not. The technological innovation of having an LLM train itself through reinforcement learning is impressive. The cost efficiency of doing inference with only 7 billion parameters rather than 700 billion parameters is impressive. The possibility of being able to do more model training and inferencing with less usage of power and less chips is impressive. It doesn't mean, though, that chip demand is at risk. What I think it means is you're more likely to see an acceleration of AI everywhere, all over the economy. And I think that that's kind of becoming the default version of what's to happen here. The idea of LLMs becoming commoditized, Scott, has always been on the table. That's why Zuckerberg released LLMA2 the way that they did. We've seen this in previous technological revolutions. In the cloud, for example, Google gave us Docs, Sheets, and Slides, which commoditized Word, Excel, and PowerPoint. But everywhere in the world you go, you will still see people using Word, Excel, and PowerPoint. I don't think there's a line of Fortune 500 CTOs who are dying to take a fistful of Chinese AI technology and shove it right into their cloud data. It's unfathomable. So I think this is a little bit overdone. And then the second thing I would tell you here, maybe the more important thing, most of this conversation about the impact of DeepSeek is happening on Twitter. If you know anything about TwitterX, if you know anything about X, you know it's loaded with hedge fund managers and asset managers who absolutely loathe the Mag 7. They've been waiting for this moment for two years. These stocks, Nvidia in particular, have been making them look bad for 24 straight months, and they're dying for this moment where small caps are up, Europe is flat, value stocks are outperforming growth, and they view this as comeuppance for the people who have been riding these giant tech stocks to huge gains. And I think because there's that desire to see these investors get beaten up a little bit, maybe that's why some of the rhetoric about this being the end of Nvidia is getting so overdone. I think there's some wish-casting happening here from some of the pseudonymous hedge fund managers.

Speaker 1: Bryn, I think one of the most important words that Josh used in that answer was if, because there are a lot of ifs, and there are a lot of unknowns, whether this is all even possible or not. Stacey Raskin at Bernstein, who by the way is going to be on with me today on Closing Bell, says, we believe that DeepSeek did not build open AI for $5 million. They say the panic he does. He says the panic over the weekend seems overblown. Cantor Fitzgerald today says this is actually very bullish for compute and for Nvidia, almost certainly leading to an AI industry wanting more compute, not less. They'd be buyers of shares on weakness like you're seeing today, which is the worst day for this stock since March of 2020. Do you agree with that?

Speaker 3: Do I agree to buy more shares today? I'm not sure. Let's see it flesh out. But where I do agree with Stacey, and I think that as investors, this is what a good old-fashioned unknown unknown looks like. It comes out of nowhere, no one's prepared for it, and this is what you see, Nvidia down 17, Vistria down 30. But what I do know, if you read the technical paper, it says apparently in the technical paper that $6 million does not include costs associated with prior research on architectural algorithms, et cetera, et cetera. So I think that you have this, what I would call a fast follower that took all of the success of ChatGPT, OpenAI, et cetera, really smart folks. The smart folks are not stuck in the U.S. They're all over the world, and they've done something incredibly innovative. And so I think that the overreaction is this like $6 million number and to think that they did this out of nowhere when the paper actually says that cost does not include these other items, which we don't know. And so I think as an investor, you have to like understand your positioning, understand to Josh's point about the large language models. You know, I'm a big fan of Alex Karp. I'm sure he's kind of laughing today at Palantir because he's been saying forever, LLMs are the commodity. It's what you build on top of that is what's gonna create the return. And so if Masa and Larry Ellison and Sam Altman are building all these data centers to go just infer for OpenAI, I think you call that into question. But if firms like Microsoft, Amazon, et cetera are building these to create in Tesla, robotics and actually applications for us to use, then I think the story is very much intact. But for right now, since we are in this unknown, unknown situation that we came into the weekend, I think you have to take a step back, kind of see how it fleshes out. But I absolutely don't believe that this is the beginning of the end of NVIDIA. Yeah. Weiss, you own the stock.

Speaker 1: I mean, we have the luxury of everybody owning the stock today. Mark Andreessen tweets that DeepSeek R1 is AI Sputnik moment. For those who are wondering what all of this means, and as I said, there's a lot of ifs. We don't really know what all of this is really about, what's true, what's not true, the cost of this versus the cost of that. What do you think about your position for those who have large positions in NVIDIA today who loaded up over the course of the last 18 months because they feel like that's the epicenter of the AI revolution story? What do you do?

Speaker 4: Well, unfortunately, and fortunately today, but it's only today, NVIDIA is not a large position for me, as I've been saying. It is for some. It is for some. It is for many. And what I would say is that the reason it wasn't a large position is because necessity is the mother of invention. So I don't recall any time in the history of technology that I'm aware of where a company, and I've said this repeatedly, where a company like NVIDIA has been able to hold their technological edge. So I don't know what to do about NVIDIA now. To Bryn's point, there's so much we don't know. This is a knee-jerk reaction, largely driven by the market that we're in, meaning that ETFs are 50% of it, lots of leverage. But here's what I do know. If this is the future, in other words, lower costs, then I will tell you that Google, that Microsoft, that Meta should be up to date because their costs of building out new capabilities has just come down meaningfully. So it's puzzling to me. Now, Meta's up, maybe because of that, but all those CapEx budgets that analysts had a problem with, well, if this is true, it could take a long time to flesh this out. Well, then that's very positive for them in terms of the NVIDIA, in terms of the Vertivs, in terms of all the others. Which are all down.

Speaker 1: I mean, Vertiv. Vertiv's getting down like 27%. Yeah. A lot of the power providers to the AI story are getting hammered today. Right. It is certainly not limited to either NVIDIA or the hyperscalers themselves.

Speaker 4: Right. So in order for those stocks to be down as much as they are, you have to believe that AI innovations are going to stop. Now, there's some technical nuances like, will the fabs and will the data centers run as hot? Particularly the data centers run as hot. Will they need as much cooling? Well, that remains to be seen. My suspicion is that the cost now building them goes down so that you will still need as much power, because we're so short power, you will still need as many cooling systems. So on a day like today, I'm not doing anything because I don't really understand what's real and what's not.