The nature of cyber is evolving from humans exploiting humans to computers exploiting humans, and finally to machines versus machines — which will drive a total rewrite of all operational software over the next five to six years.

(00:00:00) Jason, do you want to tell us about your new favorite podcast? (00:00:04) Oh, it's so good. (00:00:06) My feed is now, because, you know, since cancel culture ended Sachs, everybody uses the R word and the F word right now. (00:00:14) My entire feed on Instagram is either gay or Down syndrome or Bulldogs. (00:00:20) It's one of those three. (00:00:22) And then I stumbled upon the Miss Thing pod. (00:00:25) Miss Thing, and they do a bit called gay name, straight name. (00:00:30) Here's gay name or straight name for David. (00:00:32) This is good news and bad news, Freeberg. (00:00:35) Here we go. (00:00:35) Gay name or straight name. (00:00:39) David. (00:00:40) David to me is straight. (00:00:43) Okay. (00:00:43) Okay. (00:00:43) But he has my perfect body. (00:00:46) It can be confusing because I'm kind of like... (00:00:49) Are you gay? (00:00:50) And it's like, no, I just want to be you, David. (00:00:52) Totally. (00:00:52) Well, it's so like the Michelangelo's David, the male ideal. (00:00:56) It's like incredible body, kind of small. (00:00:58) Sorry. (00:00:59) Yeah. (00:01:00) Oh, it's a little rough. (00:01:02) What are you watching there, J. (00:01:04) Kel? (00:01:05) They basically nailed these two, but okay, keep going. (00:01:07) I don't think Chamoth is on their short list, but I know Jason will come up at some point. (00:01:11) Gay name or straight name? (00:01:14) Maybe this is it. (00:01:14) Chamoth. (00:01:16) One went on the count of three. (00:01:18) Yeah. (00:01:18) Three, two, one, gang. (00:01:21) I think like Italian sweater, like really kind of like loud. (00:01:27) vibrant sweater. (00:01:28) He like wears it to like poker night with his like, his boys. (00:01:32) And like not, I'm not talking like straight poker. (00:01:35) I'm talking like gay poker nights, like at the bar. (00:01:38) Yeah. (00:01:39) Always talking about wine. (00:01:41) Talks about wine. (00:01:42) Always sort of like swishes. (00:01:44) Yeah, exactly. (00:01:46) Yep. (00:01:46) Also, it's so like the guy at the gym taking off his shirt, taking selfies. (00:01:53) Yeah, and everyone else is kind of like, Excuse me, Chamath, I'd like to use the mirror. (00:01:57) Yeah. (00:01:57) I'd like to see myself. (00:01:59) See you at the next gay poker night. (00:02:00) Totally. (00:02:01) You bring the wine. (00:02:03) We'll bring the sweater. (00:02:04) Yeah. (00:02:05) There it is. (00:02:05) Wow, they did do a chamothe. (00:02:07) Wow. (00:02:07) That is fantastic. (00:02:08) That is fantastic. (00:02:09) A shout out to my guys at the Miss Thing Podcast. (00:02:13) Wow. (00:02:13) That was awesome. (00:02:14) I think I'm gay. (00:02:17) I never knew. (00:02:19) We'll let your winner slide. (00:02:22) Rain Man David Cyrus I'm going home (00:02:27) We open sourced it to the fans and they've just gone crazy with it. (00:02:36) What did you do? (00:02:36) Like Cameo? (00:02:37) Did you pay them to do that? (00:02:38) Did it for me as a favor. (00:02:39) So I did it for me. (00:02:40) That's awesome. (00:02:41) Well, thanks to those guys. (00:02:42) Thanks to the Miss Ning Pod. (00:02:43) I've seen those guys before in clips. (00:02:45) I find them very funny. (00:02:47) It's so great. (00:02:48) Shout out to my guys. (00:02:49) That was awesome. (00:02:50) All right, everybody. (00:02:51) Seriously, welcome back to the number one podcast in the world. (00:02:55) It's the All-In Podcast. (00:02:57) With me again, David Freiburg, from our Pioneer, David Sachs, and of course. (00:03:00) Of course, I'm Jason Calacanis. (00:03:01) You can call me Jay Cal if you're here for the first time. (00:03:04) Topic one, OpenAI. (00:03:07) They missed their targets for ChatGPT, Friedberg, both on users and revenue. (00:03:13) Let's talk about it. (00:03:14) The Wall Street Journal says in a breaking investigative report on Tuesday that OpenAI expected to hit 1 billion WOWs weekly active users before the end of 2025. (00:03:28) They missed that and they still haven't hit the milestone 4 months into (00:03:30) 2026. (00:03:31) Also, Chamath, they missed their 2025 revenue target for ChatGPT. (00:03:37) Exact number wasn't specified, but as we've talked about here, they're at a 20, 30 billion dollar run rate. (00:03:42) There's a little bit of accounting nuance that is yet to be worked out in the industry. (00:03:48) Two reasons why this matters, Sacks, OpenAI has $600 billion in spending commitments for compute. (00:03:56) Just to put that in perspective, that's about what they're trading for on secondary markets. (00:04:00) In other words, the entire value of the OpenAI enterprise equals their spend commitments in the coming year. (00:04:05) CFO Sarah Fryer, who is coming to liquidity, is reportedly worried, hey, that revenue isn't growing fast enough to keep up with expense, and OpenAI wants to IPO later this year. (00:04:18) This has put Fryer and Altman (00:04:21) in conflict or maybe there's some natural tension there. (00:04:24) Friar doesn't think OpenAI is ready for public reporting standards according to the Wall Street Journal. (00:04:30) Altman obviously wants to move faster. (00:04:32) So they released a joint statement, this is ridiculous, yada, yada, yada. (00:04:37) Let's go to you, Sachs. (00:04:38) What do you think's going on here? (00:04:39) Are these major headwinds or is this just managing expectations as the leader of the pack in the most important race of our lifetimes, the race towards super intelligence? (00:04:50) I actually have a little bit of a contrarian take on this. (00:04:53) I know that OpenAI had a really bad week. (00:04:56) Like you said, they had that Wall Street Journal article which said that they've missed their numbers. (00:05:00) They've missed their 1 billion user growth target. (00:05:03) They've missed their revenue numbers. (00:05:04) That's called into question whether they can afford the data center commitments that they've made. (00:05:09) And then in addition to that, they've also had the lawsuit with Elon happening this week. (00:05:14) So in the press, it ended up being, I think, a pretty bad week for them. (00:05:16) But I have a contrarian take on this, which I think (00:05:19) that over the past week or two, if you look at kind of what's happening at the product level, it's been a pretty good couple of weeks for them. (00:05:25) They released ChatGPT 5.5 and the reviews from people I talk to in Silicon Valley have been really strong. (00:05:33) You talk to developers, coders, they're very happy with it. (00:05:36) At the same time, (00:05:38) Opus 4.7, which is the latest Anthropic release, appears to be a bust. (00:05:41) People are complaining about it. (00:05:43) They're, in a lot of cases, they're rolling back to 4.6. (00:05:46) They're saying that Opus 4.7 is rationing compute. (00:05:50) It's reducing thinking time. (00:05:51) Not as good. (00:05:52) There were some bugs in Claude. (00:05:54) So if you just compare ChatGPT 5.5 to Opus 4.7, it does appear that OpenAI has had a better couple of weeks at a product level. (00:06:05) And I think there's reason to believe that (00:06:07) The product improvements will continue. (00:06:09) GPT 5.5 is based on a new base model called SPUD, which is the first base model upgrade they've done in, I don't know, over a year. (00:06:18) And having a new base model will pave the way for future improvements as well. (00:06:22) So I think OpenAI is feeling pretty optimistic about their product right now. (00:06:26) And I think you're starting to see on X, some of the developer mojo is shifting. (00:06:31) I'm seeing a lot of people saying that they are shifting their coding usage from (00:06:37) Opus to GPT-5.5. (00:06:40) So I think that Sam may end up being right, but for the wrong reason. (00:06:46) And what I mean by that is that when he made these big compute commitments, it was based on those estimates of hitting the billion users on the consumer side and hitting those revenue targets. (00:06:58) The consumer business ended up being weak. (00:07:01) So they missed those targets. (00:07:02) But in the meantime, coding has become the all-important sector of AI. (00:07:08) And because they made all these compute commitments and they built out these data centers, they have more compute than Anthropic right now. (00:07:16) Anthropic is token constrained. (00:07:18) It's reducing their ability to serve Mythos, for example. (00:07:21) It's causing them to engage in compute gating with Opus 4.7. (00:07:26) And I understand why Dario made that decision. (00:07:27) I'm not saying, I mean, it was a prudent business decision. (00:07:30) I'm not criticizing him for it. (00:07:31) But I think, again, I think Sam may end up being right here for the wrong reason, which is he missed on consumer, but enterprise is going gangbusters and is giving him the ability now, I think, to catch up on cursor. (00:07:44) Here's your poly market. (00:07:45) Which is the all-important market right now. (00:07:46) Of course, and we talked about Grok and Cursor teaming up last week, Elon and the team over there. (00:07:51) Poly market showing now a 32% chance that OpenAI goes public. (00:07:55) By the end (00:07:56) end of 2026. (00:07:57) This is down from 60% in December. (00:08:00) And Shamath, you gave a bit of a warning. (00:08:02) Hey, there's only so many dollars to go around. (00:08:05) SpaceX IPO is obviously getting out first. (00:08:08) And now if OpenAI doesn't go out this year, and Anthropic does, this sets up an interesting dynamic. (00:08:15) What are your thoughts here, generally speaking, about the massive commitment (00:08:20) that OpenAI has made? (00:08:21) Are they going to run off the cliff or will it wind up being brilliant even if it wasn't strategically for the exact reasons? (00:08:29) I think they're going to be fine. (00:08:30) I think this is a multi-trillion dollar company. (00:08:33) I think Anthropic is a multi-trillion dollar company. (00:08:36) I think the thing that's happening right now is a complete misunderstanding of what's actually happening inside of the world of AI. (00:08:46) And there is one very specific choke point (00:08:49) that is constraining everything, which is access to the power that's necessary to drive these tokens. (00:08:55) To the extent that OpenAI missed, I think what that is, an insight to not enough compute capacity today. (00:09:02) And that problem is only getting worse. (00:09:04) You've already seen that with Anthropic as well, where they just found a way to economically induce Amazon to give them enough capacity so that you don't have to route through bedrock to get to the Anthropic models. (00:09:19) You're also seeing them do differentiated deals now with economic participation on top of what they already had from folks like Google to give them more capacity. (00:09:28) What is my point? (00:09:30) Everything in this market is power constrained. (00:09:33) The reason that these folks may miss a number or a forecast have nothing to do with demand. (00:09:39) It is entirely 100% due to the supply of the power necessary to generate the output token. (00:09:46) There was a really interesting thing that was just announced today that will make this problem even worse, which is what you're starting to see now is backlogs build up of not just the access to the power, but then the componentry that's actually necessary, not just recips and not just nat gas turbines, but now you're talking about transformers and all the actual tactical grid infrastructure. (00:10:11) Why is this important? (00:10:12) If you look at the actual amount of gigawatts that are under construction, we have a huge mismatch now. (00:10:21) People have announced all these projects, Jason, but less than half of it is actually being built. (00:10:27) Less than half. (00:10:29) Most of it is stuck in red tape. (00:10:32) Most of that is because there are these supply chain delays. (00:10:35) So there's no credible strategy to turn any of this stuff on. (00:10:40) Who will this hurt? (00:10:42) It will hurt Anthropic and OpenAI the most. (00:10:45) Who will this benefit? (00:10:46) It will benefit the hyperscalers, specifically Oracle, Amazon, Meta, Microsoft, and Google. (00:10:53) And now what you're going to see is a negotiation and a trade back and forth. (00:10:58) How much equity do I have to give up? (00:11:00) How much control do I have to give up to get access to the compute versus (00:11:05) How badly will I miss my growth forecasts if I don't? (00:11:07) And now what that means is, and we spoke about this last week, that's a huge lane for Grok to just run through and SpaceX to run through, because they have a ton of excess capacity. (00:11:19) And so I think the cursor deal was the appetizer. (00:11:23) But if I were Elon now, I'd be running all over this market because if the models catch up in quality, I think he could also do something really crazy with Anthropic or OpenAI right now. (00:11:33) Maybe not OpenAI because of the... (00:11:35) We'll get into the lawsuit in a moment. (00:11:37) The baggage. (00:11:37) Yeah. (00:11:39) But man, he and Dario should do a deal tomorrow. (00:11:41) So you're framing, hey, the limited resource here is compute. (00:11:46) The demand is off the charts. (00:11:48) No, the limiting resource is power. (00:11:50) Power, which then powers compute, which then provides tokens, which then services the massive developer and co-work and all these other projects that consumers and enterprises can't get enough of. (00:12:03) Got it. (00:12:04) And Jason, (00:12:05) The other factor that complicates that for Anthropic and OpenAI is all the stuff that's sort of sitting around thumb twiddling, 40% of that is going to get canceled because they've done such a poor job of creating a good positive halo around AI that 40% of all the announced projects get canceled because 40% of all projects in the last four years have been canceled. (00:12:29) Yeah, and there are some bad feelings about data centers, AI, jobs, et cetera, and that's causing some headwind. (00:12:36) People are literally doing violent things in society and blaming data centers and AI for it. (00:12:43) I don't want to give it too much airtime. (00:12:45) Freeberg, what's your take on the chessboard we're looking at here, either through compute energy or through going public on a business level? (00:12:55) the strategic nature of capital, compute, and energy now playing a role in this massive amount of demand. (00:13:03) Still a ball in the air kind of game. (00:13:06) BCG had this theory, I think I talked about this once before, called the rule of three, where they've shown time and again that any stable, mature, competitive market evolves to a four to two to one (00:13:20) ratio of market share for basically 90% of the market. (00:13:23) So there's a market leader that has four times the market share of the second place. (00:13:28) That's two times the market share of the third place. (00:13:30) This is the case in pretty much every mature kind of competitive market. (00:13:33) So you can kind of think about AI probably evolving into a consumer market and an enterprise market. (00:13:38) OpenAI, even if they're not at a billion, they're still at 900 million weekly users, which is well ahead of whatever Claude is at. (00:13:47) I think Claude is probably sub-100 million. (00:13:49) Sacks, you may know. (00:13:50) And then Gemini is probably closer to them. (00:13:51) It's 700 to a billion, somewhere in that range, probably pretty neck and neck with OpenAI. (00:13:58) So the consumer market looks like it's trending towards a ChatGPT slash Google fight for 1st place and 2nd place, and then probably anthropic in 3rd place, and maybe Elon emerges and takes off enabled by his compute capacity. (00:14:13) And then the enterprise market is a little bit of a different story, and that's its own market, which is kind of anthropic, or probably Google in the lead, actually, if you look at all the Vertex use. (00:14:22) Google claims that 75% of GCP customers (00:14:26) are active users of Vertex. (00:14:28) So there's probably a pretty sizable market share that Google's captured on the enterprise side as well. (00:14:34) This is also probably why Google stock has absolutely ripped over the last couple of months, is they're literally in first place or fighting for first place in enterprise and consumer. (00:14:44) But I still think that there's a lot of opportunity, to Chamath's point about the compute and energy capacity constraints, in improving how we actually scale and deploy (00:14:55) models in both the enterprise and the consumer setting, and it is such early days. (00:14:58) And I just want to highlight this paper that came out from MIT from these two scientists. (00:15:04) And these guys published a paper on pruning techniques in neural networks. (00:15:09) This paper showed that you could actually reduce the size of these networks by 90% and (00:15:14) get the same accuracy out by pruning very large models down to smaller models, and then you can make a selection on which model to run for inference. (00:15:23) And by doing this, you can actually reduce inference costs by 10x. (00:15:27) You can get 10x the output per energy unit that goes into the data center with no loss of accuracy. (00:15:34) And so it's a really interesting, call it algorithmic technique that can be applied to the existing large models to actually make them much lower energy use. (00:15:43) So if you think about it, you're firing up a very large model to answer a very simple question. (00:15:48) You can actually prune away that model. (00:15:50) Now, (00:15:51) This is probably going to be the case in AI applications as it is in traditional Google search. (00:15:57) There's a long tail of searches, but there's a few searches that account for a large percentage of search volume. (00:16:02) It's like, what is the weather? (00:16:03) What are the movies, times? (00:16:06) What's the stock price? (00:16:07) Like there's a certain set of things (00:16:09) that make up the bulk of consumer energy. (00:16:11) And there's probably a certain set of things that probably make up the bulk of coding output as well. (00:16:15) And so if you can get that 80% of searches or chat interfaces or coding requests, (00:16:22) reduce down through pruning techniques to smaller models, and then you have a whole set of smaller models that can be called dynamically, and you reduce inference cost by 90%, you can make much more use, call it 10 times the use on data center and energy capacity than we can today. (00:16:36) So I would argue that we're still in the very early days of getting efficiency in terms of output and tokens, and we're just in the very kind of early stage of that, which also unlocks the opportunity for guys like Elon to reinvent how this is done and potentially compete pretty aggressively. (00:16:50) There are two ways to win. (00:16:51) You could. (00:16:51) throw compute at it, or you can do SLMs, small language models, and VSLMs, verticalized small language model. (00:17:00) So if you had a verticalized small language model for the weather, let's say, that doesn't exist, but. (00:17:05) You can use it as an example. (00:17:06) They will have one for travel as an example. (00:17:08) When you hit Google for flight information, it's obviously going to route you to something lighter and faster that uses Google Flights, and Google Flights has been incorporated into Gemini. (00:17:19) Gemini now is right behind 700, 750 million users (00:17:25) And it's exactly what we discussed, I don't know, 18 months ago on this podcast, Freeburg, that what if they put it at the top and what would that do to their search revenue? (00:17:34) Search revenue is surging and they're also surging. (00:17:38) So they figured out a way to balance those two competing forces, having search results that are AI enabled and still getting people to click on links. (00:17:47) They've done it brilliantly, apparently, and the stock is rewarding them. (00:17:50) I'll just add one statement to what you said, Jacob, which is like, you're using what I would call a human heuristic on (00:17:55) smaller models. (00:17:56) And I think what we're evolving to is humans don't intuitively know what this model is. (00:18:02) It's not just a verticalized model, but there are going to be models that will be discovered through automated pruning techniques that will then be working in concert. (00:18:11) So lots of small models that link together, and we don't define each model by some human heuristic, like this is a search travel model. (00:18:18) This is a maps model. (00:18:20) We don't know why these models work the way they do when they get broken down, but I do think that that's really where the evolution (00:18:25) solution is happening. (00:18:26) So effectively, a model becomes a macro model. (00:18:29) It's got lots of smaller models underneath it that can be dynamically called, and that allows you to have 10X the inference for the same unit of energy. (00:18:36) Sax. (00:18:37) Let me just build on your point about Google, J. (00:18:39) Cal. (00:18:39) I would say that if there's a single reason why... (00:18:43) OpenAI did not hit its user targets and its revenue targets, certainly around consumer. (00:18:51) You'd have to say it's because Google managed to take meaningful share. (00:18:55) They were basically nowhere. (00:18:58) A year or so ago, Sergey came out of retirement, helped focus the company. (00:19:04) And like you said, they did a brilliant job. (00:19:06) improving Gemini and putting it at the top of search, incorporating it. (00:19:09) Now that being said, again, I don't think the news is all bad for OpenAI because I do think that the 5.5 release was great. (00:19:16) We're hearing really good things about Codex. (00:19:18) I do think that Codex... (00:19:20) Codex is taking share in coding tokens right now. (00:19:24) And I just think we're in a really interesting place where these companies are constantly one-upping each other. (00:19:30) I mean, 2 weeks ago, it looked like Anthropic was going to be completely dominant, right? (00:19:34) I mean, Anthropic was growing at 10x, OpenAI was growing at 3x, and it looked like... (00:19:38) And then the servers started going down. (00:19:40) Did you see that this week, Sax? (00:19:41) The servers were going down. (00:19:42) People were in my office were complaining. (00:19:44) We can't get on Claude. (00:19:46) Listen, competition brings out the best in everyone. (00:19:48) Anthropic forced (00:19:50) OpenAI to compete, Google's forced OpenAI to compete in consumer. (00:19:54) I just hope the market stays competitive for as long as possible. (00:19:56) I do think that's what's best for consumers, our economy. (00:20:00) and for our country overall. (00:20:01) Let me just say one other area where I think OpenAI had a good week is in this red hot area of cyber. (00:20:11) Obviously, Anthropic made a huge splash with Mythos. (00:20:14) It hasn't been commercially released. (00:20:16) They're compute constrained, but as a proof of concept or training model, it hit a new level of capabilities with cyber. (00:20:21) But now, OpenAI has released a new model called GPT-5.5 Cyber, which has just been through a bunch of tests, and they've shown, this was testing done by the AI Security Institute, that GPT-5.5 is the second model to complete (00:20:36) one of their multi-step cyber attack simulations end to end. (00:20:39) So it has the same level of capability as Mythos, and it does appear to be commercially ready. (00:20:47) You know, they've got the compute to serve it. (00:20:49) So I do think that that's a big accomplishment. (00:20:53) I mean, look, we knew that other cyber models were coming. (00:20:56) It wasn't just going to be Mythos. (00:20:58) In fact, within six months or so, all the frontier models are going to have Mythos level cyber capability. (00:21:03) But it's impressive that (00:21:05) OpenAI got this GPT-5.5 cyber out so quickly. (00:21:10) And I think 5.5 might be the first cyber model that cyber defenders actually get to use. (00:21:17) Because again, I don't think they're as compute constrained as Anthropic is. (00:21:21) And this is an incredible opportunity, you know, for the CrowdStrikes and Palo Alto networks of the world, both of which have been on the program. (00:21:29) They come out and they start attacking this space. (00:21:31) Man, you could really (00:21:34) see everything get tightened up. (00:21:37) And this could be an incredible revenue stream for everybody who's got, whether it's Cursor, Claude, or OpenAI or Gemini, this is an amazing opportunity to tighten up as much as it is to get attacked. (00:21:49) Can I make a point about that? (00:21:50) Because look, there is so much fear right now, almost the level of panic about mythos. (00:21:55) People are treating it like a doomsday weapon or something like that. (00:21:58) It's not. (00:21:58) It's simply that the frontier models have reached the point where they're capable of automating cyber activities, just like they're capable of automating coding. (00:22:08) But that means that a model could power up. (00:22:12) a cyber attacker or cyber defender the same way they can power up a coder and allow them to discover a lot more vulnerabilities. (00:22:19) So there is obviously a risk there, but I think it's important to understand that Mythos or GPT-5.5, it doesn't create the vulnerabilities. (00:22:28) It just discovers them. (00:22:29) The bugs were already in the code. (00:22:31) They were sitting there waiting for some hacker to discover. (00:22:34) If we can now use AI to find (00:22:37) these bugs in advance, these vulnerabilities, and patch them, then you actually harden our infrastructure and you harden our security. (00:22:46) I also believe that this leap from, let's call it pre-AI cyber to post-AI cyber, it's going to be, I think, a big one-time upgrade cycle, because again, you're going to find all these dormant bugs and vulnerabilities. (00:22:59) But I think that once we get past that upgrade cycle, you're going to reach a new equilibrium between AI-powered cyber offense and AI-powered cyber defense. (00:23:07) It's going to become a lot more normal. (00:23:09) It's not going to feel like this huge disruption, which is to say, I think people are treating this as like some existential threat. (00:23:17) I don't think it is as long as everyone does what they're supposed to do, which is use the new capabilities to harden their code bases and infrastructure and security before the hackers get ahold of these capabilities. (00:23:28) Yeah, and if Chamath, if you were to look at this, to build on Sax's point, there are about 5 million or so security experts in the world. (00:23:36) And we talked about token cost, 40 hours of tokens just pounding it a week, you could create another 5 million for $100 per chief security officer per (00:23:49) security experts. (00:23:50) So it's the volume of security expert agent sacks, to your point. (00:23:53) Yeah. (00:23:54) You could have 50 million of them, 100 million of them. (00:23:57) They're not finding something unique. (00:24:00) They're just, they never sleep. (00:24:02) They're relentless in their pursuit of these problems. (00:24:04) It's a really great point. (00:24:05) Just kind of just refine that. (00:24:06) So yeah, there's probably 5 million people in the cyber industry, but there's probably only a few thousand really elite hackers. (00:24:11) Sure. (00:24:12) those hackers didn't have the time to go after the entire surface area of every possible attack vector out there. (00:24:19) And so if you train a model to do what they do, obviously, like you said, it can operate with a scale and speed that the human hacker can't. (00:24:27) So obviously, what you need to do is get these tools in the hands of the white hats, let them (00:24:34) do the cyber attacks themselves to then find the vulnerabilities and patch them before the black hats get a hold of these capabilities. (00:24:40) But I think it's just one last point on this and I'll stop. (00:24:43) It's just, it's really important to understand that the Chinese models are going to have these capabilities within approximately 6 months. (00:24:49) Oh, they have them now in Deep Sea 4 for sure. (00:24:51) They've got some level. (00:24:53) Well, Deep Sea 4, I mean, Deep Sea 4 is impressive in a lot of ways, but its capability is not at the frontier. (00:24:59) It's maybe 80, 80, 85% of, let's call it the American frontier. (00:25:03) in on this. (00:25:04) Let's get Chamath in. (00:25:05) Two things. (00:25:06) The reason that this is even possible is because humans are error prone. (00:25:11) And when humans code, they create holes. (00:25:14) And so humans exploiting humans is where we've been for a long time. (00:25:19) Now we have computers exploiting humans because the computers go and seek out all these bugs that humans wrote. (00:25:25) In the next phase, it'll be machines versus machines. (00:25:29) And so I think the nature of cyber is going to completely change. (00:25:32) Probably in the next five or six years, there'll be so much reason to rewrite all of the software that runs the world. (00:25:41) in one part because you're going to be asked to show more operating leverage and revenue growth, but in another part because everything else that was handmade in the past is just fundamentally insecure. (00:25:51) Either way, all roads will lead to all the operational software that runs the world will get rewritten. (00:25:57) More and more of it will be written by machines. (00:25:59) More and more of it will be impregnable as a result. (00:26:02) But then the cyber threat actually will only increase. (00:26:05) Because then you're going to try to figure out how to use a machine to inject something into another machine so that some agentic loop injects some malware or injects a bad token. (00:26:14) And I think that's a very complicated thing. (00:26:16) What I will tell you is, I'm not even sure if I'm allowed to say this, but a very good, probably the best cybersecurity company in the world, run by one of the very best CEOs in the world, who may or may not be speaking at Liquidity, would tell you that they (00:26:35) have penetrated and can essentially manipulate every model. (00:26:42) Let me just say it roughly that way. (00:26:44) Okay, perfect. (00:26:44) Yeah, and at the Breakthrough Prize, which three of the four of us were at, I talked to George Kurtz, the other person you were kind of describing was not that person. (00:26:54) Yeah, I'm talking about Nikash and George are the two guys leading this, Palo Alto Networks, CrowdStrike, and they understand, what George told me was there is just a line out the door of people who want this product (00:27:05) or service. (00:27:06) And if you look at it, Freeberg, like the murder rate, like we're sitting here with the lowest murder rate in the history of humanity. (00:27:14) It has gone down massively in our lifetimes, but massively over the arc of history. (00:27:18) I think that's what's going to happen with cyber. (00:27:20) There is only so many attack vectors and the remaining attack vectors are just going to be human factors, right? (00:27:25) Freeberg, that's always been the case. (00:27:27) And as we make the software more resilient, then the weak link is, you know, the secretary who puts her (00:27:35) Post-it note, with the password there or the accountant who, uses their dog's name plus 123 for their password, right? (00:27:43) That's the historical one. (00:27:45) Okay, let's, anything you want to add, Freeberg, as we wrapped there? (00:27:49) Oh, that's a pleasure. (00:27:49) Why is your bed so messy, by the way? (00:27:50) Why can't you just ask the room service to come in and clean your? (00:27:53) Listen, I can tell you what happened. (00:27:54) Listen, I'm here in Atlanta. (00:27:56) And also, why don't you have a suite like where there's two rooms? (00:27:58) Like, is it just one room? (00:27:59) This hotel only has one room. (00:28:00) It is just one room. (00:28:01) Yes, it is. (00:28:02) Either you're cheap or poor. (00:28:04) Which one is it? (00:28:05) I'm cheap. (00:28:05) I'm cheap. (00:28:06) Here's what I'll tell you. (00:28:07) Here's the situation. (00:28:08) I'm in Atlanta for the Knicks game tonight. (00:28:10) Here's what I do. (00:28:11) I just want to explain to you value for value. (00:28:13) Some people spend their money on private jets and they spend $30,000 flying to Atlanta. (00:28:18) I spend 30,000 on courtside seeds. (00:28:20) I don't want the suite. (00:28:21) I want to put it into the seat and you can do both. (00:28:23) You can do both. (00:28:24) I guess I could do both too. (00:28:26) I'm in the process of becoming. (00:28:28) I don't understand. (00:28:29) Of embracing my richness. (00:28:31) OK. (00:28:31) If you've already convinced yourself that you should spend. (00:28:34) $30,000 for quartside tickets, which I think is outrageous. (00:28:38) But okay, you've already convinced yourself. (00:28:39) They're more like 10K each, but yeah. (00:28:41) A hotel room that has two rooms, okay, probably costs 15% more than what you're paying. (00:28:47) It's 2X, but yes, you're right. (00:28:49) I'll get the I'll get the hotel. (00:28:50) Okay, or 20% more, but it just looks like 200 bucks a night. (00:28:54) So you pay 400 a night, you get another room. (00:28:55) I mean, it's Atlanta. (00:28:57) The most, I'm in the best hotel, the most expensive hotel is 500 a night in Atlanta. (00:29:00) It's no big deal. (00:29:01) But everything's sold out because all the Knicks people are coming here. (00:29:04) Everything is sold out because the Knicks are here. (00:29:09) So we have to look at your dirty bed. (00:29:10) It's gross. (00:29:11) The bed's not that dirty. (00:29:12) Come on. (00:29:13) Oh, just deal with it. (00:29:14) Okay. (00:29:14) Take it out in post. (00:29:15) I have a private jet story about flying to Atlanta. (00:29:18) Oh, tell us something. (00:29:20) There we go. (00:29:20) You're reminding me. (00:29:21) Okay, so yeah, there was some event there. (00:29:23) So I flew my team there. (00:29:26) there's a few people on my plane. (00:29:27) And it's kind of a long flight. (00:29:29) So it's like 4 hours or something. (00:29:30) From the back, yeah. (00:29:32) so I went in the back to sleep. (00:29:34) Well, first, we started the flight and I had a few bottles of Pappy Van Winkle on the plane. (00:29:39) And so we started off with like a drink. (00:29:42) And then I went in the back and fell asleep and I woke up basically when we landed. (00:29:46) So I come out and all three bottles are basically cashed of having a Van Winkle. (00:29:51) Oops, those are like 2 grand a bottle. (00:29:53) No, there are more. (00:29:54) These were like antique bottles. (00:29:55) Like one of them was like. (00:29:56) I have one of those from your plane. (00:29:57) I have one of those from the old Falcon. (00:29:59) Yeah, they were like these vintage. (00:30:01) They were $4,000. (00:30:02) I remember you. (00:30:03) Anyway, you can't even find this shit anymore. (00:30:06) So these guys, they asked me like when Lance like, hey Sax, how much did it cost? (00:30:11) for you to fly us to this event? (00:30:13) And I said, about $8,000 in jet fuel and about $12,000 of Pappy Van Winkle. (00:30:21) You got to you got to fuel the vibes as well as the plane. (00:30:25) It's Is Atlanta nice? (00:30:27) I've never really spent time. (00:30:27) Do those people still work for you or are they? (00:30:29) Are they? (00:30:30) Yeah. (00:30:31) They called in Atlanta. (00:30:33) Is Atlanta nice? (00:30:34) Listen, last year I went to the Detroit games and that city was on the rebound. (00:30:37) Atlanta has an incredible opportunity to rebound. (00:30:39) I'll say it that way. (00:30:41) There's a great (00:30:41) It's a great opportunity for them to upgrade the city. (00:30:44) I went to Waffle House at midnight last night. (00:30:46) There was no shootings. (00:30:47) Okay, let's keep moving. (00:30:48) By the way, do you get royalty points at the Best Western Atlanta or no? (00:30:53) I get double points because I use my Best Western Visa card. (00:30:56) Yeah, it's everywhere you want it to be. (00:30:59) All right, use the promo code JCAL and get 1,000 extra points. (00:31:03) In other OpenAI news, (00:31:05) Musk versus Altman, the trial of the century, maybe the decade, has started. (00:31:12) Elon is, of course, accusing OpenAI of breach of charitable trust, unjust enrichment. (00:31:17) He's accusing OpenAI of essentially flipping A non-profit into a for-profit. (00:31:22) He's seeking $150 billion in damages that they revert back to a non-profit, that Altman and Brockman be removed. (00:31:29) And there were some fireworks between Elon and the OpenAI lawyers. (00:31:33) Elon kind of leveled up the discussion. (00:31:35) He said, (00:31:35) Quote, if we make it okay to loot a charity, the entire foundation of charitable giving in America will be destroyed. (00:31:42) That's my concern. (00:31:45) Obviously, there's a ton of interesting nuances here. (00:31:48) Specifically, Greg Brockman keeping a diary where he was journal maxing his plans, like a Bond villain here. (00:31:57) And the excerpts from his diary include, conclusion, we truly want the B Corp. (00:32:03) The true answer is that we want (00:32:06) Elon out. (00:32:06) If 3 months later we're doing B Corp, then it was a lie. (00:32:09) Can't see us turning this into a for-profit without a nasty fight. (00:32:13) I'm just thinking about the office and we're in the office and this story will correctly be that we weren't honest with him. (00:32:20) In the end, it's still about wanting a for-profit just without him, yada, yada, yada. (00:32:24) Freeburg, your thoughts on this case. (00:32:26) Is Elon going to win? (00:32:28) I just don't know why Greg Brockman's got a frigging diary where he's like literally documenting. (00:32:33) I mean, I love the guy, but what the fuck is he thinking? (00:32:36) Like, you're just sitting here at home and like, let me write about the crime I'm committing or let me write it like, and let me record it. (00:32:42) And by the way, let me never delete it. (00:32:44) I don't understand this. (00:32:46) It's not just journal maxing, it's discovery maxing. (00:32:50) It's smoking gun maxing. (00:32:52) I don't get it. (00:32:53) I don't get it, man. (00:32:55) I mean, did you guys remember from The Wire in that scene where the guy's like, is you taking notes on a criminal conspiracy? (00:33:04) This has got everybody in the room. (00:33:06) Can we play that clip? (00:33:07) It's like, what are you doing, Greg? (00:33:10) Nick, is you taking notes on a criminal conspiracy? (00:33:14) What the is you thinking, man? (00:33:17) If you're going to commit a crime, you do not write down the date and time of the crime in your journal. (00:33:23) Well, look, we don't know it's a crime. (00:33:25) Let's not. (00:33:26) Okay, sure. (00:33:26) Text, do you write a diary? (00:33:27) Don't care if it's a crime, but... (00:33:28) Well, yes. (00:33:29) No, you keep even shenanigans. (00:33:31) Tamar, do you keep a diary? (00:33:33) What do you think? (00:33:34) Jake, do you keep a diary? (00:33:37) I believe rumination. (00:33:38) No, I'll tell you right now. (00:33:39) Rumination is the path to unhappiness. (00:33:43) Nobody gives a shit about your feelings. (00:33:44) Writing your feelings down is only going to make you miserable. (00:33:47) Talking to your spouse about your feelings. (00:33:50) Just go to a beautiful dinner, sit courtside at the Knicks, and do what I've been doing for 30 years. (00:33:57) Recharge Maxim. (00:33:57) Recharge Maxim. (00:33:58) And as the register goes up, all you have to do is work. (00:34:02) Start new projects, 9 out of 10 foul, place 9 out of 10 bets, one wins, and you're golden. (00:34:08) Go sit courtside at the Knicks game. (00:34:09) Keep going. (00:34:10) Life's too short. (00:34:10) And just keep moving forward. (00:34:12) Don't write anything down. (00:34:14) Period, full stop. (00:34:15) That's good advice. (00:34:16) Yeah, I just, the biggest surprise to me was this guy's got a diarrhea. (00:34:19) I just, I don't know anyone that (00:34:20) I've never heard of this. (00:34:22) So anyway, that was shocking. (00:34:24) Besides that, I have no view on what's going to happen with the case or what the judge will do. (00:34:28) I have no comment on the case either. (00:34:30) I think it's weird that Polymarket hasn't budged, even as all of this discovery has been published. (00:34:36) It's effectively at 42 or 43% that Elon wins. (00:34:40) So one of the friends in our group chat said, what may just happen is that Elon technically wins and he's just credited back the $40 million. (00:34:50) And so maybe that's what this poll is front-running. (00:34:54) But on a totally separate note, I think, Jason, I know you say it as a joke, but this idea of just keep moving forward, don't ruminate, I think is very good general life advice for everybody to follow. (00:35:07) The modern day therapy industrial complex and the medication industrial complex, I believe is- It pivots around rumination. (00:35:15) Well, it does pivot around rumination. (00:35:16) Yes. (00:35:17) That is the gateway drug to all these things. (00:35:20) Think about your problems. (00:35:21) when these people go to therapy, you ever hear these people? (00:35:23) Howard Stern's like, I've been in therapy with the same person two or three days a week for 40 years. (00:35:27) I'm like, okay, what's the incentive for the therapist to stop charging you $1,200 an hour? (00:35:31) There is none. (00:35:32) Then they lose a revenue stream. (00:35:34) They lose a customer. (00:35:35) It's all a giant fucking fraud. (00:35:37) Sacks, in terms of this case. (00:35:39) I wouldn't go that far. (00:35:40) I do think that there's a lot of value in kind of (00:35:43) untying some of these Gordian knots that people have because of how they grew up. (00:35:48) But there's a difference between that and being specific and just randomly ruminating because I don't think there's a lot of productive. (00:35:54) You've got an acute issue like in trauma in your life. (00:35:58) Yeah, sure. (00:35:58) Unpack it, figure it out. (00:36:00) I'm just talking about this never ending self-improvement, you know, ruminating thing. (00:36:05) But getting back on topic here, Sachs, (00:36:09) And what's the, and we're talking about a jury, I believe in Oakland. (00:36:13) No, but it's a bench trial. (00:36:14) This is important. (00:36:15) It's a bench trial where the jury is advisory in capacity. (00:36:19) But ultimately that judge, she will make the final call. (00:36:22) And she'll do the damages. (00:36:24) And so is this a case acts of like, we've got a Bay Area jury judge and we've got... (00:36:33) Elon, who's considered a bit right-wing and people don't all agree in that area in terms of his politics. (00:36:39) And then you have this Sam Altman, New Yorker story and people finding out that so many different people feel they got screwed by him. (00:36:48) You put these two things together, it's impossible to handicap where this turns out. (00:36:52) Sacks, your thoughts? (00:36:54) Well, yeah, I don't think this is about politics. (00:36:57) I mean, I guess you could argue that (00:37:00) what Elon is seeking, which is to protect the charity, is if anything, a left-coded sort of principle. (00:37:06) Although I don't really think it's left versus right. (00:37:08) Look, I don't want to take sides on this trial. (00:37:11) I'm just watching like everyone else. (00:37:12) The last time I weighed in on some Elon litigation, I got deposed for six hours. (00:37:18) Remember that? (00:37:19) Because they just assumed that somehow I know something. (00:37:22) I've never talked to Elon about the case. (00:37:23) I don't know anything about it. (00:37:25) I'm going to see what happens like everyone else. (00:37:27) Now, one thing I will (00:37:29) say, having just read some of the coverage, is that apparently the company at some point did offer Elon shares in the company, but he thought that there was something kind of icky about it. (00:37:43) Do you remember this? (00:37:44) Yes. (00:37:45) Because at one point I said on our show, when this dispute started happening before it became a court case, I said, look, if OpenAI at a certain point decided they had the wrong structure, they should have just gone and done a make right with Elon and he should have been a shareholder on the cap table. (00:38:01) What I didn't know is that apparently they did try to do something like that, but Elon turned it down because he did want the entity to remain A charitable (00:38:13) He had a principled view of it according to the reports and was like, no, we're trying to save humanity. (00:38:19) And then you're giving this keys to the kingdom to Microsoft. (00:38:22) That's all come out. (00:38:24) And I also have not talking to Elon about any of this. (00:38:28) But my guess is like most of these things, there'll be some sort of settlement or something here. (00:38:34) But maybe he takes it to the mat. (00:38:36) Who knows? (00:38:37) Judge Rogers, who's doing this 61 year old Obama appointee, (00:38:42) Politics has played a role, Sachs. (00:38:44) They have had to tell the jury, like, however you feel about these individuals, politically, whatever, please put that aside. (00:38:49) But of note is that she oversaw the Epic Games versus Apple trial over apps for exclusively ruled in favor of Apple with some caveats that they don't have a monopoly, et cetera, et cetera. (00:39:01) So this is going to be a (00:39:03) A really interesting one, I think the worst case scenario is OpenAI, for OpenAI is they have to unravel this somehow. (00:39:11) And that would delay the IPO, that would cause chaos in shareholders. (00:39:15) And I guess the best case is some sort of settlement. (00:39:18) And if Elon put the first 40 or $50 million in, he's due 10, 20, 30% of the company after dilution. (00:39:26) All right, let's keep moving through the docket. (00:39:28) Lots more to discuss and good luck to everybody in their lawsuit. (00:39:31) And those of you betting on the market, All In Summit, selling out fast, our 5th edition, Los Angeles, September 13th to 15th, go to allin.com slash events and speakers are going to be top tier. (00:39:44) Apparently Freeberg is having this as his major creative outlet. (00:39:49) I heard some back channel Chamath today that he's going to be doing Broadway musical (00:39:56) illusion is. (00:39:57) I got a tap dancing situation. (00:39:59) He's lit.