Ep6. AI Demand / Supply - Models, Agents, the $2T Compute Build Out, Need for More Nuclear & More

**Bill Gurley** (0:00)
If World War II never happened, the atomic bomb never happened, and someone just showed up in 2024 and said, I figured out this thing, it's nuclear fission, people would be like, oh my God, like, probably be more excited than you are about AI, right? Like, because it would solve all of our problems.

**Brad Gerstner** (0:31)
Hey, man, what's going on? Nice hat.

**Bill Gurley** (0:34)
How you doing?

**Brad Gerstner** (0:35)
Welcome, horns.

**Bill Gurley** (0:36)
Definitely.

**Brad Gerstner** (0:37)
I mean, is this a little promotion for the state of Texas we got going on?

**Bill Gurley** (0:41)
Yeah, yeah, I'm proud of the state of Texas, and a little bit, you know, I'm sad that the, I was especially watching the women's tournament quite a bit, and Texas just came this close to the Final Four, but the games last night with Iowa and LSU were just amazing. So anyway, it's been a fun time.
So yeah, reminiscing, perhaps a little bit.

**Brad Gerstner** (1:02)
March is a good month.
Well, speaking of reminiscing, I've been reminiscing, you know, I'm on spring break with my son, and we see these markets around AI getting, there's a lot of commentary about how frothy they're getting, and it reminded me of that question, you know, it's different this time. People say the four most dangerous words in the investing universe, but yet every big breakthrough we've gone through in tech, it actually has been different, right? And so as analysts, as anthropologists, as forecasters, right? We have to try to sort this out, both in the short run and the long run, right? And it's hard to do. I mean, you do this bottoms up, you do this tops down, you study it.
Lots of people in 1998 knew the internet was going to be massive, right? I mean, Henry Blodgett calling Amazon 400, right? People thought it was blasphemous, but it didn't stop us from having a boom and a bust along the way, right? And today Amazon's at 3000 bucks, almost 10X what Henry, you know, got shouted down for saying in 1998 And I think at the time he said it was a 10 year call. But the fact of the matter is trying to marry up the short and the long term, I think is really, really tough.

**Bill Gurley** (2:17)
Well, and I think it gets even tougher if enthusiasm builds because that impacts the entry price on the marginal investment that one might make. And I think this particular moment in time is very, very difficult for investors that are looking at the marginal investment because the prices infer some amount of optimism already.

**Brad Gerstner** (2:46)
For sure. And so that's really what I think we're going to dig into a little bit today. Tap on this and go a little bit deeper.
Maybe starting with this idea is there is increasing evidence that demand for training and inferences is maybe deeper and wider than we thought. We're reading a lot of headlines about the world building ever bigger supercomputers, and then a lot of conversation about what those bottlenecks become. But why don't we just start with this question about why do we need bigger? And I guess first principles, generative AI produces these tokens. These tokens are a proxy for human intelligence. And there's a lot of conversation about there's no limit to how much incremental human intelligence we want to buy.
You and I have talked a fair bit about copilots for engineering. We talked about that with Dara, copilots for call centers. And in a pretty short period of time, these have become really ubiquitous development projects for almost every enterprise in just a few short years wanting these copilots. And now we're seeing a lot of conversation about autonomous agents. I think Benchmark is investors in LangChain. Harrison Chase gave a great talk last week about what's next for AI agents that I encourage folks to watch around planning, UI and memory. And, you know, we hosted an AI dinner last week and there was an interesting conversation that came up just about a search use case within the enterprise. And in this particular company, the CTO took their million, well, 100,000 lines of code, their entire code base, about a million tokens, dropped it all into the prompt and then just asked if it spotted any common bugs. And he said it found six like non-trivial bugs in the code base. So the point is, we've gone from copilot to some of these maybe a little bit more autonomous systems, magic and cognition, all of this in less than two years.
And let me ask you the question. If we compare this to 1998, like looking at just demand, you know, do you think this demand is as real as it appears?