00:11:12 · John Coogan, Jordi Hays
They can send it over to, Google, Anthropic, all the other buyers. So as the conversations went on, I gradually began to understand where the pessimism was coming from. In the end, they were all asking the same question. So what comes after AI coding? What will be the next engine of ARR for Anthropic and OpenAI? Several candidates came up, AI for us, for science, AI chip design, and so on, but none of them clearly convinced investors. I couldn't offer a clear answer either. Sure. There are rumors that Anthropic has made major breakthrough in both AI for science and AI chip design, but what we actually want to know is when any of that turns into money. You know, some of these initiatives, if they're if they're Yeah. [00:11:54] S1: I think the main the main thing is, like, what what Joe's referencing is, you know, I was talking about, like, we need a step up in capability in a meaningful way because there are there are a bunch of businesses in the world Yeah. Small and large, where they want to be using more AI Mhmm. But they can't really find a way to spend more than a couple $100 a month [00:12:15] S0: Yeah. [00:12:15] S1: Per person Yeah. Let's say. And so AI spend is still tiny relative to the other costs in the business. Yeah. And so they're they're sitting there saying, I'm willing to spend a lot more. [00:12:25] S0: Yeah. [00:12:25] S1: I just can't find a way yet to apply it Totally. Outside of coding and a couple of these other domains.
00:12:25 · Jordi Hays, John Coogan
[00:12:25] S1: I just can't find a way yet to apply it Totally. Outside of coding and a couple of these other domains. [00:12:31] S0: Yeah. Yeah. I mean, if you just add up the revenues from all the harness companies, I don't wanna necessarily call them wrapper companies, but like the companies that are in AI enablement, AI diffusion, AI for law, AI for finance, these types of companies. You add up all of that, even if the labs were able to just immediately get that revenue, doesn't move the needle that much. And so Well next... Yeah. [00:12:56] S1: They are getting a lot of the revenue. [00:12:57] S0: Oh, yeah. Already they're getting a ton of revenue. But, what what what I'm saying is like like, an engineer can easily justify, a $100,000 token budget a year if they're making $203,104 $100,000 in their comp package. Jensen's floated like a one to one ratio, and that doesn't seem that crazy for certain roles. Much harder to justify You're a lawyer. Right now. [00:13:20] S1: I think you floated more
00:13:20 · Jordi Hays, John Coogan
[00:13:20] S1: I think you floated more [00:13:21] S0: aggressive. Yeah. But, it's it's that, right now, everyone who has a marketer on their team, everyone who has a a... In in house counsel, everyone who has a CFO, they would all love to be able to go to them and say, yeah. Your budget for AI is a $100,000 a year. Go spend it, and let's make sure the ROI is there. But we're generally confident the AI... That the ROI is there. Right now, people are like, you can have $200 a month. You can have maybe $2,000 a month, but you're not getting $20 a month yet. And so that's like the the original, like, white collar work pitch that then has kind of melted away because of the job displacement discussion. So now we're into AI for science and AI for chip design. But a lot of those times, even if you have a breakthrough, like, there's just going to be... In terms of adoption and actual monetization of a new drug, there are just gonna be people who say, I want a phase three trial, and I want, you know, humans to have tried this drug for nine months straight. I want... You you can use AI to design it. You can use AI to do all the different testing, and and and and you can, work for... On AI for the manufacturing. But at the at the at the end of the day, I wanna see that there's an evidence that a human took this pill and survived a year later. And that's what a lot of the FDA trials do. And so that's a little bit slow even if the balance sheet is there from the big, biotech companies.
00:14:48 · Jordi Hays, John Coogan
[00:14:48] S1: Also, yeah, on the on the AI for science side and r and d Yeah. I just think, like, generating a bunch of... You know, we talked about this with Dylan and the semi analysis team. Yeah. Like, there's the there's, like, the the idea generation Mhmm. Part Mhmm. Which, like, already there's a lot of ideas. And then there's the actual, like, r and d process Sure. To figure out if something is like viable. Yeah. You're gonna use... The the second part takes a lot of time. My my concern with like how that revenue will ramp over time is that like you just can't... Even with AI, you're not gonna pick, I think at this point, anything where it still has to tie into the real world [00:15:31] S0: Yeah. [00:15:31] S1: Right through like, you know, human trials, things You're like not gonna like pick [00:15:38] S0: physically You're not gonna say into silicon into a fab. It is You're not gonna It's like high definition slide a thousand. [00:15:44] S1: You can run 5,000 simulations [00:15:46] S0: Yeah. [00:15:46] S1: In software, but then once it's in the real world and and Yeah. You're just not...
00:15:51 · John Coogan, Jordi Hays
[00:15:51] S0: And we've seen this with both the model progress on anything that can't be... Can't... The RL can't happen in a millisecond on a server. It it... If it has anything in the external world, that's a huge that's a huge drag on the speed with which it can diffuse. Yeah. But but we'll we'll we'll we'll see. I mean, at the same time, like, the market is good at pricing breakthroughs. Like, the GLP ones drove a trillion dollars of market cap. Yeah. Way before they were even in the hands of people. Like, they were, like, very limited rollouts to people for... Through doctors. It was pre diffusion effectively, and and the market was able to capitalize that early because it was expected and understood. And so, you could see at least some of the financing be underwritten by breakthroughs in AI for science or chip design or any other harder field that takes more time. You can actually match the capital to that. So, there's still lots of optimism there. Let me tell you about console. We still got our console stickers on here. Console builds AI agents that automate 70 of IT, HR, and finance support, giving employees instant resolution for access requests and password resets. So to summarize the Citrini article, Citrini overhauls thematic model book calls peak in AI infrastructure CapEx. Citrini says AI infrastructure asymmetric risk has shifted to downside. Citrini expects higher costs for US data centers after democratic nature. So I think that we brought the the... Hey. We didn't have to sell the soundboard. Okay? [00:17:19] S1: Yeah. [00:17:20] S0: We might count 5%.
00:17:20 · John Coogan, Jordi Hays
[00:17:20] S0: We might count 5%. [00:17:21] S1: We still got this. We grabbed this on the way out. [00:17:24] S0: Yeah. That was the last thing. So to me, Warren's macro regime twist... Shifting toward late cycle stagflation tilt. So, lots of lots of rough stuff on the timeline. Steven Balaban's extremely bullish on Neocloud. If there's pushback on data centers in some districts, that will artificially suppress supply of the exact product that, Lambda sells compute. The demand for compute will continue to grow exponentially. There's never been a better time to be a cloud at Citrix. Right? And the replies, how much active data center power capacity do you have? Asking questions. Yeah. It's sort it's sort of his Christmas. Speaking of Christmas, how many days till Christmas? [00:18:02] S1: Pull it up.