Calling Anthropic a monopoly at this stage is premature and dangerous — these are still fledgling startups facing fierce competition from Google, Amazon, and Meta, and preemptive regulatory talk risks derailing America's AI lead.

(00:00:00) How do I sound? (00:00:01) You sound perfect. (00:00:01) You sound great. (00:00:02) How do I look? (00:00:03) Yeah, you sound great. (00:00:04) Better than you look. (00:00:05) Got that face made for radio. (00:00:07) You don't look as tired as you have in recent weeks. (00:00:11) That's true. (00:00:11) Oh, really? (00:00:12) Yeah. (00:00:12) Oh, yeah. (00:00:12) Somebody was slagging me for the bags and for my yikes. (00:00:15) I mean, this audience is brutal. (00:00:17) They're brutal. (00:00:18) They're brutal. (00:00:19) It's a good thing I'm rich. (00:00:24) We'll let your winners ride. (00:00:27) Rain Man David Simon. (00:00:33) All right, everybody, welcome back. (00:00:41) It's the number one podcast in the world. (00:00:43) It's the All In Podcast. (00:00:45) With us today, Chamath Palihapitiya, David Sacks, and our fifth bestie, Mr. (00:00:51) Brad Gerstner, is here. (00:00:53) I think David Freberg is suffering from some (00:00:57) socialist related flu. (00:00:59) He's very sick of reading about socialists, but he'll be back next week with two incredible, incredible interviews. (00:01:06) You guys see those Spencer Pratt ads? (00:01:08) Wow. (00:01:08) It's one of the best political ads I've ever, ever seen. (00:01:11) Oh, there's like three or four of them. (00:01:12) There are multiples. (00:01:14) Whoever that social media team is, on fire. (00:01:16) If you get a good social media team and you get a good ad production team, I think it's next Gen. (00:01:20) because these things go crazy. (00:01:22) And Spencer Pratt, if he wins this election, which I think he's going to in Los Angeles, (00:01:26) The reason is what Brad said. (00:01:28) Those ads are incredible. (00:01:30) Well, he's also quite a good debater. (00:01:32) Did you see clips from this debate? (00:01:34) Incredible. (00:01:35) He's so funny. (00:01:37) He's so chill. (00:01:38) Yeah, well, he's up against Karen Bass, who's the mayor, who is basically extremely left-wing. (00:01:43) And then there's someone who's a city councilwoman who's even further to the left of Karen Bass. (00:01:48) I mean, she's often like Fidel Castro territory. (00:01:51) She's an Indian Fidel Castro. (00:01:53) Ramen. (00:01:54) So she was basically, (00:01:56) I guess, criticizing the mayor for the homelessness problem. (00:02:00) And then Pratt pointed out that this councilwoman is actually in charge of all these homeless programs already. (00:02:08) He eviscerated her. (00:02:09) He eviscerated her. (00:02:10) And he basically made the key point, which is, look, the problem here is not lack of housing. (00:02:14) It's an addiction issue and it's a mental illness issue. (00:02:17) And he said, look, if you... (00:02:19) He said if she went to the street, she'd get stabbed in the dead. (00:02:22) Yeah. (00:02:24) Which is pretty accurate if you've been to Skid Row. (00:02:26) I mean, you would not want to walk through there. (00:02:29) It was like the Spider-Man photo. (00:02:30) Wait, Nick's got the clip. (00:02:31) Plays a clip. (00:02:32) Oh God, this clip is brutal. (00:02:34) This is a different clip, but this went super viral. (00:02:36) It reminded me of Trump a little bit. (00:02:37) Let's see. (00:02:38) I'm not sure how to respond to that vision of Los Angeles. (00:02:41) This is a MAGA Republican's idea of what Los Angeles looks like. (00:02:45) This is really not. (00:02:50) Unbelievable. (00:02:50) For those of you listening, he put his hands up and wiggled his hand like, oh my God. (00:02:56) ASACs, you know, Stephen Pratt wins mayor. (00:02:59) Spencer Pratt, Spencer Pratt. (00:03:00) And then the, you know, the ballot initiative, the Retirement Protection and Savings Act, right, it's going to pass, going to pass with big numbers. (00:03:07) This is the, you know, referendum that effectively is going to knock out the wealth tax. (00:03:11) Can you imagine if California effectively passes a constitutional amendment protecting retirement savings and personal assets and banning (00:03:19) the wealth tax and prac gets elected, the message that would send to the country. (00:03:24) That's a very non-consensus view that I'm becoming increasingly optimistic about. (00:03:29) Well, from your lips to God's ears. (00:03:32) But until that message actually is sent, I think I'm going to be... (00:03:36) In Texas. (00:03:37) In Texas. (00:03:39) Yee-haw. (00:03:40) Well, I mean, this is also in the face of, I don't know, just a follow-up story here, but Mandami did like an attack video on Ken Griffin's house. (00:03:47) We talked about it on the pod a couple of weeks ago. (00:03:50) And like literally stood in front of his house, pointed at it. (00:03:52) And this is in the face of like a CEO getting shot. (00:03:56) for ideological reasons, Sam Altman's house being targeted. (00:03:59) This is like a really dangerous thing for Mondami to do. (00:04:02) And Ken Griffin came out today or yesterday in an interview and said, hey, listen, I'm out. (00:04:07) We're going to be putting our efforts into Florida. (00:04:11) And this is the same thing that happened to Chicago. (00:04:13) And he basically said, like, I really felt offended. (00:04:16) And I was, you know, nervous about this (00:04:19) because of my personal safety. (00:04:20) And he called him out. (00:04:21) Mandami came out with like a mealy-mouthed response that didn't even apologize for what he did. (00:04:28) Just doubled down on it, essentially. (00:04:30) All right, let's get to the dot. (00:04:31) I don't know if you guys saw it or not, but. (00:04:32) Who cares? (00:04:33) New York is becoming a flyover city. (00:04:35) It's an interesting way to put it. (00:04:37) Don't disagree. (00:04:38) All right, first story, Elon just leased all of Colossus 1, his data center. (00:04:43) Wait, what? (00:04:43) He did? (00:04:45) What? (00:04:47) Yes, shocking to Dario and Anthropic, Chamath, on last week's pod. (00:04:51) Go ahead and give yourself a pat on the back. (00:04:53) You said Elon and Dario should do a deal tomorrow. (00:04:56) It didn't happen the next day. (00:04:57) It happened five days later. (00:04:58) So you came close, Chamath. (00:05:00) But no cigar because of Anthropic's obvious compute constraints. (00:05:04) Anthropic just added over 220,000 Nvidia GPUs, over 300 megawatts of energy. (00:05:11) The DL is already having an impact, as we've discussed here. (00:05:15) Claude users have been experiencing rate limits. (00:05:17) Well, Claude has now doubled the Claude code rate limits, removed peak usage caps for paid users, and increased API volumes for Opus models. (00:05:25) XAI is now training their models at Colossus 2. (00:05:29) So they have more than enough compute. (00:05:31) Elon made a great bet (00:05:33) on compute and built up those data centers really fast. (00:05:36) And that is now paying off. (00:05:38) We had the cursor deal we talked about last week. (00:05:40) Let's talk about the emergence of Elon Web Services, EWS, Chamath. (00:05:46) He's now in the hyperscaler, competing against Google Cloud, Amazon Web Services, and Azure. (00:05:52) And I don't know if you had inside information or just a brilliant epiphany, but take us behind the call and what do you think about the deal itself? (00:06:01) I think the deal is fantastic. (00:06:03) I'll say maybe 3 quick things. (00:06:06) The first is, as I mentioned a couple of weeks ago, Anthropic and OpenAI's revenue performance has nothing to do with demand. (00:06:17) Zero. (00:06:18) It is entirely to do with the supply constraints that exist in data centers and specifically in power. (00:06:27) If they had infinite power, I think that their revenues would probably be even more parabolic. (00:06:33) And so all the breathlessness about either exceeding or underperforming a forecast, in my opinion, mean nothing. (00:06:41) I think the five-year view for those two companies is quite robust. (00:06:47) The thing that they really need is more compute and more power. (00:06:52) That's the first thing. (00:06:53) The second thing is, while they need that, we have a very big problem, which is we unfortunately have very poor leadership at the head of most of these AI firms. (00:07:08) I think they are coming off as untrustworthy or too self-interested. (00:07:13) The political reaction now is starting to turn negative. (00:07:18) The community reaction is negative. (00:07:22) You have about 9 gigawatts that are supposed to come online this year. (00:07:27) Almost 50% of it now is being protested. (00:07:30) More than likely, if history holds, most of that will get turned off. (00:07:35) So they will get even more supply constraint. (00:07:38) So that's the setup. (00:07:40) So what's the opportunity? (00:07:41) I think for Elon, if you look inside of how people try to nitpick the SpaceX valuation case, or let's not even, let's give them, sorry, let's be more generous. (00:07:50) When people try to paint the bear case or they try to red team the valuation, the biggest element is the on-the-come value around the orbital data centers. (00:08:01) And by actually landing a bunch of terrestrial capacity, I think you start to blunt that because you can now start to say that even if the orbital data centers get delayed by a few months or a few quarters, even if the technological de-risking of it takes longer, he now has a structural core business (00:08:20) that will effectively subsidize his ability to train Grok, which I think is a really important and under-reported theme. (00:08:28) So you have all this infrastructure. (00:08:30) He somehow saw the tea leaves before most people. (00:08:34) He built to a level of scale and secure power before most people. (00:08:38) It has now become the critical asset. (00:08:40) And now he's kind of king-making. (00:08:43) And I think that that's a really interesting valuation (00:08:47) reinforcement as SpaceX goes through testing the waters and the roadshow. (00:08:52) Brad, your take. (00:08:54) Yeah, no, I think it's well said. (00:08:55) I mean, first, we know that there's nobody better on planet Earth than Elon at converting electrons to tokens. (00:09:01) It's a critically important evolution to the story. (00:09:05) I think our friend Sean McGuire, he sent out a tweet that summed it up well. (00:09:09) And he said, SpaceX has this five-layer cake, launch, connectivity, compute hyperscaler space data centers, and then Apple (00:09:17) applications and models and then other bets, right? (00:09:21) The question on the roadshow has been, but X.AI isn't on the revenue trajectory of OpenAI and Anthropic, and yet there are huge commitments. (00:09:32) And now we see the ace card that Elon's playing. (00:09:35) He said he was building AWS all along or EWS all along. (00:09:40) And so I estimate that this is going to generate in this year an incremental $4 to $5 billion of revenue on top (00:09:47) of what I've seen analysts estimates in the mid-20s. (00:09:50) That's a material amount of incremental revenue to offset the cost of the investments that he's made here. (00:09:56) And that will subsidize, to Chamas' point, all that he's investing to build the next generation of Grok. (00:10:02) Remember, too, that he has three facilities, Colossus, Macro Hard, and Macro Harder, 1.2 gigawatts in Macro Hard and Macro Harder in Blackwell. (00:10:13) So he's given the one that's kind of less connected, H100, it's great for inference. (00:10:18) to Anthropic. (00:10:19) He's monetizing it in a big way. (00:10:21) It's terrific for Anthropic. (00:10:23) And it solves what I think was the biggest question in the valuation story, which is, what if he spends ahead of X.AI's revenue? (00:10:33) It takes the pressure, Chamath, off X.AI delivering immediate revenue. (00:10:37) Now he becomes an immediate competitor in the hyperscale. (00:10:41) I don't think this is the last announcement. (00:10:43) I think he's going to make a lot more, (00:10:47) moves in this direction. (00:10:48) I think it will be a material part of their story and their revenue projections as they come together. (00:10:54) And I would just say finally, again, everybody has talked about how we don't have enough power, how we don't have enough compute, how the revenues would not show up this year. (00:11:03) But the chaos that is American capitalism somehow finds a way, okay? (00:11:08) And there's tremendous demand for Anthropic, and we find a way. (00:11:13) I was so happy to see kind of the detente. (00:11:16) And the kind exchange between the team of Anthropic and Elon, because we need all of this in order to produce American frontier models to stay at the frontier. (00:11:27) And then finally, I'd just say, you know, Chamath, you referenced these activists that are protesting delaying these data centers. (00:11:33) in these localities. (00:11:35) One thing I want to dispel this myth, this is not like organic, hyper-local protests by people in a community that aren't being spurred on. (00:11:44) This is highly organized activists that are moving across the country to stir up trouble in the exact same way they did to stop all fission reactors being built 30 years ago in America. (00:11:55) Now we have no nuclear reactors being built. (00:11:58) China's got 100 of them. (00:12:00) Who was funding those activists? (00:12:02) I think we need to really (00:12:03) look into who's funding the activists now. (00:12:06) I'm not saying that there aren't any concerns, but the misinformation about water, the misinformation about electricity bills, electricity bills are going up in the places that are not building data centers, New York and California, because they haven't built any supply on the grid. (00:12:22) In Texas, where you're building the most data centers in the country, electricity costs are going down. (00:12:26) So I think that's a boogeyman that we got to take on. (00:12:33) look, the deal is highly complimentary, as Jamatha and Brad pointed out. (00:12:38) SpaceX has a profitable, I think very profitable, space and telecommunications Starlink business, the satellite business. (00:12:47) But the xAI business had huge losses. (00:12:50) The reason's pretty straightforward. (00:12:51) You need these super large training clusters. (00:12:54) They cost a lot of money. (00:12:55) And until you have a model that's capable of competing at the frontier, you're not making any revenue. (00:13:00) And that problem is compounded by the fact that right now all the revenue is in enterprise, which is to say coding. (00:13:05) We know that XAI just did that deal with Cursor to try and catch up, but they don't have a coding product yet. (00:13:11) So they're not participating in the revenue, but they're participating in all of the costs. (00:13:15) So this deal fixes that problem. (00:13:17) Elon's now able to have a frontier model company (00:13:22) but he's able to now not have these massive unpaid for CapEx commitments, right? (00:13:27) Because he's able to kind of lease that capacity. (00:13:29) So I think it solves a major problem for them and their balance sheet. (00:13:34) And then you have to say that for Anthropic, this is a really great thing because they were compute constrained. (00:13:40) And just to build on that point, I mean, I guess let me be the first to congratulate Dario on winning the AI race. (00:13:47) And you've been, let's be honest, Sax, you have been, on this podcast, you've been moderately critical of that company and Dario himself for being, you know, a little P. (00:13:58) Doomer 110. (00:14:00) And on your X account, you've been even a little spicier. (00:14:03) So now that there's peace in the Middle East of the AI business, what's your take here? (00:14:09) My take is, look, let's just, (00:14:11) honestly and accurately assess where the state of this AI market is at right now and Anthropic's placed within it. (00:14:18) So for the last three years, Anthropic has been growing at a rate of 10x a year. (00:14:24) I think going into this year, probably the conventional wisdom was that there'd be no way to sustain (00:14:29) that kind of rate of growth at this level of scale. (00:14:32) And what happened in the first four months of the year? (00:14:34) First, we find out that from January 1st to March 31st, they grew from roughly 10 billion of ARR to 30 billion. (00:14:41) So it tripled. (00:14:42) And then in April, if anything, the rate of increase seemed to accelerate. (00:14:46) They went from 30 to 44 billion of ARR. (00:14:50) Nobody in Silicon Valley has ever seen anything like it. (00:14:53) Forget about the rest of the country. (00:14:54) I mean, (00:14:55) All we do in Silicon Valley is deal with exponentials, and still people have never seen that kind of growth at that level of scale. (00:15:01) The only thing holding them back in the future was compute. (00:15:05) Now they've made this deal. (00:15:06) They've made other deals as well to get that compute. (00:15:09) I think it's pretty much a foregone conclusion that they will hit that forecast of 10x this year, exiting the year, call it roughly 100 billion of ARR. (00:15:17) And now the only question is whether they hit a trillion in 2027. (00:15:23) We can debate whether that's true or not, but look, if they do that, I think they'll easily be the most valuable tech company in history. (00:15:34) In fact, they might even be more valuable than the rest of the Mag 7 put together. (00:15:38) Just to give people some basis for comparison here, you know, the biggest tech companies, Apple, Nvidia, Google, Nvidia, Google, I think they kind of do around 4 to 500 billion a year right now of revenue. (00:15:49) I guess Nvidia is a little bit of a different category, but you look at (00:15:54) The hyperscalers. (00:15:55) Yeah. (00:15:55) I mean, Google is doing what, like 120 billion 1/4, something like that, 100 billion 1/4? (00:16:01) Correct. (00:16:01) But growing at what, 20% year over year? (00:16:03) Not 100%. (00:16:04) It's certainly not 1000%. (00:16:07) So the fact that Anthropic could be on track, in fact, let me correct that. (00:16:12) You see them going to the Maggie. (00:16:13) It'll be a Maggie. (00:16:14) I'm saying something else, which is that unless something about their current trajectory changes, (00:16:23) Anthropic will be the most powerful monopoly ever created in human history. (00:16:29) Again, it will be, you know, a trillion dollars of ARR growing at some exponential. (00:16:34) Interesting. (00:16:35) Dario calls it AGI. (00:16:36) I call it the biggest monopoly in human history. (00:16:38) Interesting to hear that word, monopoly, Sachs. (00:16:41) Very interesting placement. (00:16:43) Chamath, go ahead, and then we'll... (00:16:46) In 2025 was 420 billion. (00:16:49) Microsoft was 300 billion. (00:16:51) Alphabet was 390 billion. (00:16:53) Amazon, 700 billion. (00:16:55) Nvidia, 190 billion. (00:16:57) Meta, 185 billion. (00:16:58) Tesla, 110 billion total, about 2.3 to 2.35 trillion. (00:17:03) So if Sachs is right and Anthropic, you know, can tack on a trillion, it won't be the MAG 7, it'll be the MAG 1. (00:17:12) Just to put it in perspective, though, Dario and Dworkish said he thought the combined AI revenue (00:17:18) of the market leaders would be about a trillion and 29. (00:17:21) I love what you're saying, Sachs. (00:17:23) I think there is unlimited TAM. (00:17:25) We may be over our skis a little bit in terms of the forecast. (00:17:28) If you back your way from compute, right, they expect to have 5 gigs by the end of this year, 10 gigs by the end of next year. (00:17:35) It's kind of hard to get to those numbers for a single company, but I do believe that the trajectory that they're on, I totally agree with you, is on an exponential that not many people believed in four months ago. (00:17:46) Right, so then the question is, okay, I think (00:17:48) we all agree they're on an exponential curve and that the TAM is big enough to support that. (00:17:53) Just one data point on TAM, my understanding of the total market size just on coding is 1 trillion, meaning that a trillion dollars a year roughly is spent on software developers and all things related to the creation of software. (00:18:08) Now, I'm not saying that they eat that entire market, but I can easily see the market for software. (00:18:13) I think it's more than that. (00:18:15) hold on, doubling from a trillion to 2 trillion, given that coding tokens, basically 10 Xs or 100 Xs, the value of that market and the ability to generate code. (00:18:27) So I think we all will agree that the TAM here is large enough to support a trillion dollars of revenue. (00:18:32) Brad, I think you bring up a couple of really important constraints. (00:18:35) First, there may not be enough compute and there's not enough energy. (00:18:38) I'd say the second big one is what's the competitive reaction going to be? (00:18:41) Totally. (00:18:42) Because I would say at the beginning of this year, (00:18:45) All these Frontier Labs were playing around with a lot of different things. (00:18:47) I mean, Anthropic was the porcupine. (00:18:50) They believed in one thing. (00:18:52) All these other companies were kind of acting like the fox who thinks they're good at a lot of different things. (00:18:56) They were doing Nano Banana, they were doing Sora, they were doing image generation, they were doing fantasy character chatbots. (00:19:04) In hindsight, they were doing a lot of things that appear to be kind of a waste of time. (00:19:09) The whole market appears now to be coding and the things that we built on coding tokens, like co-work, like agents. (00:19:15) And so there is going to be a competitive response here where all the other guys realize, oh, wait a second, we were mis-focused. (00:19:22) They're going to get focused. (00:19:23) I just don't know how much share they're going to be able to take. (00:19:26) It does look like OpenAI has already made the pivot. (00:19:29) We hear very good things about Codex now based on GPT-5.5. (00:19:34) 5.5 is based on a new base model called SPUD. (00:19:37) I think they're very optimistic about continuing improvements. (00:19:40) Their rate of growth appears to be accelerating now because of 5.5 (00:19:45) So look, there's reason to believe that OpenAI can take some share here. (00:19:50) I'm sure that Google won't be asleep at the wheel. (00:19:52) They're very, very good at coding. (00:19:54) They've got a really good team and Elon just tied up with Cursor. (00:19:57) So there is going to be more competition, but still what you have to say (00:20:00) And I think all of us know this from Silicon Valley, is you always want to be the company in the lead that's on that trajectory, where all you have to do is maintain inertia, whereas the other people have to change something in order to put themselves back in the race. (00:20:16) So this is when I say somewhat sort of facetiously, congratulations Dario on winning the AI race. (00:20:22) I don't mean that he's won it, but he is winning it right now. (00:20:25) But here's the brilliance of what Elon's doing. (00:20:28) If you look at the existing business, (00:20:30) which is Starlink and basically the launch services at SpaceX. (00:20:33) Incredible business, obviously, 20 billion this year, I think, is the estimate. (00:20:38) But if you look at the footprint of Amazon Web Services, Azure, and GCP, you're looking at, you know, $300 billion in revenue and a market cap of combined 5 trillion, 4 trillion, if these were independent companies. (00:20:56) And if you look at what is Elon's core competency at Tesla, it's building factories. (00:21:02) And if you look at the footprint of these factories, they're huge. (00:21:05) What are data centers? (00:21:06) They're basically big giant factories. (00:21:07) And then if you look at energy, what else is Elon extremely good at? (00:21:12) This is the battery deployment, and he's also got solar deployment from the SolarCity, often criticized acquisition he did years ago. (00:21:19) So you put this all together, if this is 5 billion, as I think you referenced, Brad, if it's $5 billion in incremental (00:21:27) Elon web services business, and he's a NeoCloud, what could he build on planet Earth? (00:21:34) What could he build inside of Teslas in terms of extra compute? (00:21:37) What could he build inside the Powerwall? (00:21:39) What if the Powerwalls had his new fabs in them and you built a distributed system from home to home? (00:21:45) The Powerwall has compute in it, the cars have compute in it, and of course, the ultimate manifestation of this where nobody can complain is you go right out into space. (00:21:56) And that's what he's going to do. (00:21:57) And the sneaky small part of this announcement from Elon and from Anthropic and Dario was they're also interested in space. (00:22:06) So look for the race to go from factories and data centers to homes. (00:22:12) The Powerwall with compute in it, it's already online, right? (00:22:15) And Starlink also gives them the ability to do distributed compute to people's homes. (00:22:20) Again, you could be paying people to put power walls with compute in it. (00:22:23) That's going to be the next shoe to drop, I believe. (00:22:25) Did you guys see the deal that was announced yesterday between Pulte Homes, which is a huge builder? (00:22:32) Yes, they're doing it as well. (00:22:34) And Span? (00:22:35) Yes. (00:22:35) Nick, just throw this up here. (00:22:36) It's super cool. (00:22:37) What's happening is that these guys are putting many data centers with NVIDIA GPU clusters (00:22:45) beside every home and then allowing people to actually run those things. (00:22:48) And that's just incredible. (00:22:49) I thought that was so cool. (00:22:50) It's a great pivot. (00:22:51) What this company did originally, Chamath, was they did the power panel. (00:22:54) They made smart power panels. (00:22:56) So you know when you flip your breakers, all those breakers are in an app. (00:22:59) I looked at it for my house, but I guess they pivoted to add this, and I think Bass Power, Brad, you're an investor in it. (00:23:04) They're going to do the same thing. (00:23:06) Zach Dowell is doing that. (00:23:07) One of the things I'd just say in response, Jason, to what you just said about Elon, right? (00:23:11) This is why the SpaceX IPO is going to trade at 40 (00:23:15) to 50 times revenue. (00:23:17) Okay. (00:23:17) So next year, if they do 40 to 50 billion and this thing goes out at 2 trillion, right, that they're going to trade at a really high revenue multiple compared to the mag five that are trading at like 25 times earnings. (00:23:31) And there's only one person on the planet who has a future pipeline of innovation and the largest TAM in the world because he's playing in all these different spaces that can command that multiple and it's Elon and it's deserved and it's great to (00:23:45) Tesla has that same Elon variable in it as well, which is people value his companies at, I would say, two times market, three times market, four times market because of the future pipeline. (00:23:58) And they devalue Apple because they don't have somebody like Elon or Steve Jobs there who is giving them the future. (00:24:04) I don't think it's devalued. (00:24:05) I think we're probably valued if you don't have an Elon and you have somebody like, I think that's exactly what it is. (00:24:10) We talked about this last week, but explain why you think it's different. (00:24:12) I think all of these companies are actually very, fairly valued. (00:24:15) valued, and then Elon World gets a premium. (00:24:19) And that premium is because of what you guys said. (00:24:22) That I agree with. (00:24:23) The big message that I take away from this, which the markets and retail are telling you is, you guys have stopped innovating. (00:24:32) There's a lot of incrementalism. (00:24:35) And we as a society aren't benefiting broadly the way that you told us we would be. (00:24:41) And so maybe this is the best way for them to get this message, which is to whack their valuation. (00:24:46) And by the way, I'll just say it again, when Tesla and SpaceX merge and we have all things Elon and Elon Corp, okay, which will happen probably by the end of the year, maybe it'll happen in the middle of next year. (00:24:59) It's going to then break everybody's brains again, because you'll have this one asset, as you guys said, that will trade at a valuation premium. (00:25:07) And some people will say it's unexplainable. (00:25:10) And I think it's logically explainable, which is everybody else has stopped innovating. (00:25:15) People know how to draw more blood from the stone, how to target better ads. (00:25:21) That does nothing for society anymore. (00:25:23) That's literally. (00:25:25) In fact, it does the opposite. (00:25:26) There is no good left. (00:25:28) That was literally the exact point I was making when you cut me off. (00:25:30) If you look, Tim Cook's greatest innovation, Tim Cook's greatest innovation before you cut me off was Apple TV. (00:25:38) Not even the hardware product. (00:25:39) It was just spending money and making a Netflix knock off. (00:25:41) There's been no other products in. (00:25:44) Hold on. (00:25:44) Let me finish again before you interrupt me making my choice. (00:25:46) You don't like that? (00:25:47) Oh, you like being interrupted? (00:25:47) Oh, yeah. (00:25:48) Well, OK, go back to the interesting. (00:25:53) If you look at their track record, and I think this is why we had a change there, is they have not done anything innovative. (00:26:00) And in fact, the things they were doing that were innovating in AI or self-driving cars, they shut down. (00:26:05) They won't take any swing through the bat. (00:26:07) So they are getting penalized in their valuation penalized. (00:26:10) They're just not getting a premium. (00:26:11) They're not getting penalized. (00:26:12) I think they're getting penalized. (00:26:13) Like every metric they're trading at incredible valuations. (00:26:16) Just look at them. (00:26:17) Oh, no. (00:26:18) I don't. (00:26:18) I mean, if you compare the two valuations, I think they're (00:26:20) being penalized. (00:26:21) Anyway, let's anybody else want to get in on this before we move on to the next one? (00:26:25) Yeah. (00:26:25) There is no world in which Google and Meta and Apple and Amazon could be viewed as being penalized in valuation. (00:26:36) There is very clearly a world where Elon gets a massive premium because he's innovating. (00:26:41) You're saying the same things. (00:26:42) You're saying the same thing. (00:26:43) We're not the same thing. (00:26:44) It's not the same thing. (00:26:45) Listen, we're debating semantics here. (00:26:48) I'm not letting you off the hook, Saxy Poo. (00:26:51) When Sax is very deliberate in how he speaks, they said he's the captain of the debate club in his 20,000 word article this week, and that he's a master debater. (00:26:59) He's A masturbator. (00:27:00) And (00:27:02) You slipped in. (00:27:03) You slipped it in. (00:27:05) Are you saying that the FTC or whoever should be going in and looking at anthropic? (00:27:11) Oh, Brad's book is getting attacked. (00:27:12) Headwinds. (00:27:13) You said they're a monopoly or they're heading to monopoly tactics. (00:27:17) Are that what you're saying? (00:27:20) look, I mean, we know that tech markets have a history of consolidating down and turning into either monopolies or duopolies. (00:27:28) And if you just look at the revenue right now, there's only two companies making substantial revenue on AI. (00:27:36) It's Anthropic and OpenAI. (00:27:38) We know that OpenAI is growing at 3 to 4X, which is incredible at the level of scale they're at. (00:27:43) Anthropic, though we said, is growing at an exponential 10X a year. (00:27:47) And if they just do that, (00:27:49) For 18 more months, they'll be by far the most valuable company in human history, and they'll have unprecedented control over the most important technology of our time. (00:27:58) So I don't know what you call that, but it is something to think about. (00:28:03) And I guess I do have a thought experiment for you guys, which is, I just want you to think for a second about the case of John D. (00:28:12) Rockefeller, who I think is known as probably (00:28:16) the most successful monopolist in American history. (00:28:21) But he wasn't very good at PR. (00:28:22) He was terrible at PR. (00:28:24) Everyone sort of recognized how ruthless he is. (00:28:26) We've seen movies like There Will Be Blood, which is basically about him. (00:28:29) In any event, imagine if John D. (00:28:31) Rockefeller was way better at public relations. (00:28:35) And instead of calling his company Standard Oil, he called it Safe Oil. (00:28:40) Okay, let's just play this thought experiment. (00:28:42) Clean, beautiful coal. (00:28:43) Yes. (00:28:44) Safe oil. (00:28:45) We call it safe oil because as we know, kerosene is dangerous. (00:28:49) Their first big product was kerosene. (00:28:51) And kerosene can light your house or it can burn it down. (00:28:54) And in the wrong hands, it can torch a city or you can use it to make a bomb. (00:28:59) So John D, let's say, should have called for the creation of a new government agency (00:29:05) to regulate the safety of this product. (00:29:07) And they could have done rigorous testing, licensing, common sense regulation. (00:29:12) There would have been a very intense debate over safety standards. (00:29:15) You know, what should the proper wick thickness be? (00:29:18) And should we allow all those dangerous independent refiners, right? (00:29:23) And I think people would have gotten so wrapped up in this debate over what constituted safe oil or safe kerosene that they would have missed what was really going on. (00:29:33) which is that Rockefeller was building the richest, most powerful monopoly of all time. (00:29:37) In fact, people might even have called Rockefeller an effective altruist, because of course, he was so concerned about the safety (00:29:45) of his product. (00:29:46) I love it. (00:29:47) Shout out to David Sachs's writers. (00:29:49) Great, great, great writers. (00:29:51) Newman? (00:29:51) Newman wrote this? (00:29:52) No, I wrote it. (00:29:53) Well, Emmy Award for best writing in a dramatic monologue goes to Newman. (00:29:58) Wow, Sachs. (00:29:59) No, that's my writing. (00:30:00) You landed it. (00:30:00) Very good, Sachs. (00:30:02) I thought after the Elon Anthropic detente, where Elon said, you know, complimented Anthropic, and David started off with a bit of a compliment. (00:30:10) I thought we maybe were past this first. (00:30:12) It's ridiculous to think of this as a monopoly. (00:30:14) You know, we're talking about (00:30:16) annual run rate revenues, David, but on a gap basis, they're doing about the same revenue as OpenAI in the month of March. (00:30:22) Okay, so we're way ahead of ourselves. (00:30:24) By the way, five months ago, everybody thought OpenAI was going to run away with this. (00:30:28) Google's revenues are very substantial in AI. (00:30:31) And by the way, Google, Amazon, et cetera, these companies are producing $100 billion of free cash flow to justify their incremental investment. (00:30:40) At the same time, you have these two startups that are still fledgling, that are still fragile in the scheme of things. (00:30:46) You of all people should know we've got the best competition in AI on the planet, which is why we're at the frontier and kicking the tail of everybody else on the planet. (00:30:55) So I just want to see these companies compete. (00:30:57) I want to see DC stay out of the way. (00:30:59) The last thing I want to be doing is, you know, seeing people talk about this and throwing roadblocks into the way of the competition. (00:31:06) All right. (00:31:07) I think anthropic. (00:31:09) Well, let me, hold on, let me translate Brad for you. (00:31:12) Don't with my paper is what he's saying. (00:31:14) He's got bets on me. (00:31:16) Sachs, Washington, D.C., don't with Brad's paper. (00:31:19) Sachs, do you want to get into the regulation stuff right now as a segue? (00:31:24) Let me respond to Brad and also translate what I'm saying satirically. (00:31:28) Okay. (00:31:29) Satirically. (00:31:29) So first of all, nobody wants to see these companies compete vigorously more than me. (00:31:36) That was the whole premise. (00:31:38) of the action plan that we worked on last year is we want to bring out the best in everyone. (00:31:42) This is how America's going to win the AI race. (00:31:44) We have five major labs vigorously competing. (00:31:48) And as long as that competition is taking place, that I think that's a good thing. (00:31:52) Doesn't mean we can't have guardrails and the rest of it, but basically competition should be our North Star. (00:31:57) All of that being said, okay, what I am pointing out, and I think it's historically true, that people in Washington have woken up to monopolies (00:32:07) on the late side, not early, right? (00:32:10) Because I mean, once a company has won 80% on the market, that's when they wake up and say, oh, we have a monopoly here. (00:32:16) And I'm not saying that they have a monopoly yet, but if the trajectory continues for just 18 more months, then I think it will be in this unprecedentedly powerful position. (00:32:28) And hold on, and I don't think people should be distracted from that fact (00:32:34) by this rhetoric around safety, because someone like Rockefeller could have used it too. (00:32:40) And I do think, I mean, just like one last point on this, I do think that if you actually look at what a lot of the safetyist policies are calling for, they're basically calling for a form of regulatory capture. (00:32:53) And they're calling for things that would create a stronger moat around this monopoly or duopoly that's in the process of being created. (00:33:01) And it would get in the way of competition. (00:33:04) So again, I think that people might not have such a charitable view of all this safety rhetoric if they understood that what was being created here is the biggest monopoly in human history. (00:33:16) And I think we should just be a little bit more skeptical about some of these altruistic claims. (00:33:21) I can't believe that David is like, you know, talking monopolies when we haven't even left the starting gate (00:33:30) of AI. (00:33:30) I think this is a, to me. (00:33:35) There's only two companies with revenue. (00:33:36) The last thing I want is DC trying to preemptively, preemptively, which would be like a disastrous consequence, get in the game of picking winners and losers at the starting line of AI. (00:33:48) That would be a disaster. (00:33:50) Brad, did you just put another soapbox on top of the soapbox you were standing on? (00:33:55) Look, Brad, like I said, my North Star is competition. (00:33:58) As long as there's competition going on, I support it. (00:34:01) However, hold on, hold on. (00:34:03) We know that monopolists want to stop competition, and they use regulatory capture to do it. (00:34:09) And furthermore, they do things like ban their competitors from using their product. (00:34:13) What conceivable reason did Anthropic have for banning OpenClaw using its models? (00:34:18) That is anti-competitive, is it not? (00:34:21) I would double click on it. (00:34:23) I would double click on it. (00:34:24) I might not file, but I would double click. (00:34:26) Okay, listen, Chamath, the girls are fighting. (00:34:29) Let's keep moving through the docket. (00:34:30) We're going to be here all day with these two. (00:34:32) And one thing that you're going to need to act on very quickly is the All-In Summit. (00:34:37) It's selling out fast. (00:34:38) Don't miss it. (00:34:38) Speakers are top tier again. (00:34:40) Freeberg, busy working on some amazing speakers. (00:34:43) Sax will be there. (00:34:44) He's flying in and out every day for four hours. (00:34:46) And then we're (00:34:49) We're going to have a lot of networking stuff going down. (00:34:50) We're building some networking software. (00:34:52) So when you come to our events, you get to meet people. (00:34:54) That's what we always say. (00:34:55) My playbook for events. (00:34:57) If you learn something from the speakers every day, one or two things, if you meet somebody new and you eat some great food and have some fun, you get two or three of those things. (00:35:06) Man, even if you get one, you're going to come back to the event. (00:35:08) You're going to get all three all day long, all in.com slash events, Los Angeles, September 13th, 14th and 15th. (00:35:15) Apologies to everybody asking, but liquidity is sold out and we've shut down (00:35:19) the wait list. (00:35:19) There's just no more room. (00:35:21) All right. (00:35:21) The White House allegedly, possibly is considering, according to reports, an FDA for AI that would vet. (00:35:30) You heard that correct, folks. (00:35:32) That would vet new models for safety. (00:35:35) The thing we've been talking about not doing here, the thing David Sachs has spent the last year on, the White House is considering. (00:35:41) New York Times reported Trump is considering an executive order to create an, quote, AI working group. (00:35:47) This group would include tech execs and government officials who would, quote, examine potential oversight procedures. (00:35:54) including, quote, a review process for new AI models. (00:35:59) Oy, according to the report, the catalyst was, wait for it, Anthropic's mythos model, which reportedly scared, spooked, made people really nervous at the White House. (00:36:10) Quote, the White House wants to avoid any political repercussions if a devastating AI-enabled cyber attack were to occur. (00:36:18) They want a CYA, according to the New York Times. (00:36:21) Kevin Hassett, that guy, the director of the (00:36:24) The National Economic Council confirmed the report on Fox Business. (00:36:28) Here's your 15-second clip. (00:36:29) We're studying possibly an executive order to give a clear roadmap to everybody about how this is going to go and how future AIs that also potentially create vulnerabilities should go through a process so that, you know, they're released in the wild after they've been proven safe, just like an FDA drug. (00:36:46) Additionally, friend of the pod, (00:36:50) Scott Besson had something to say. (00:36:52) What we've had in the past month was a step change in the power of one large language model, but we're going to see it from the other AI companies. (00:37:01) What we are determined to do is work with our AI companies to allow them to continue innovate. (00:37:08) But our charge at the US government is maintaining safety. (00:37:11) And there is a very important calculus here between innovation and safety. (00:37:17) And at the US government, we're going to make sure that things stay. (00:37:20) There you go, Kevin Hassett and Bessant. (00:37:25) Slightly different positions here, Brad. (00:37:26) What do you think? (00:37:27) Actually, I don't think they're slightly different positions, but I would agree that Kevin bringing up the FDA kind of muddied the waters. (00:37:34) I talked to Kevin last night after that clip ran. (00:37:37) and I asked him, I just said, do you think FDA is the right analog here? (00:37:42) And he said, I was only bringing it up to say that we want them to show us the models so that we can coordinate them. (00:37:48) Obviously, our job is to make sure that the government is prepared, that we harden our systems, that our intelligence agencies are up to speed. (00:37:56) But he does not think, and I can't find anybody, (00:37:59) on the right, that believes that we're going to move to an approval regime, right? (00:38:05) The approval regime, this idea that you're going to have to share every model with an FDA in Washington and they're going to have to pre-approve the model is a disaster. (00:38:14) Sachs has been effectively fighting against this correctly over the course of the last year. (00:38:19) It would just, it would lead to three bad things. (00:38:21) Number one, we do not want to put the Washington in the position of picking winners and losers when it comes to these models. (00:38:27) We're winning. (00:38:28) We're on the winning (00:38:29) horse in America. (00:38:30) We're out in front of the rest of the world. (00:38:32) There's no reason to change horses and regimes at this point. (00:38:35) And we don't want to burden this with more democracy. (00:38:38) But at the same time, obviously, I call these pre-AGI or AGI models, Mistral, SPUD, et cetera. (00:38:44) I see a lot of coordination going on between the industry and government. (00:38:48) I think we can do an even better job of evolving that framework so that everybody in government is on the same page. (00:38:54) We need to build more capacity in government to quickly be able to do the cyber review on the (00:38:59) these models. (00:39:00) Right now, it takes too long when the coordination does occur. (00:39:03) So we need to have a finite amount of time to get government feedback, et cetera. (00:39:08) But the last thing that we want is an FDA of models sitting in Washington. (00:39:12) Kevin understands that. (00:39:14) Scott Besson understands that. (00:39:15) So I expect that we will continue down the path that we've been on. (00:39:19) Shamath, obviously, I think we all agree we don't need an FDA for AI, but (00:39:25) There are things that reasonably people would want to have guardrails around AI. (00:39:31) I'm sure you would agree. (00:39:33) It shouldn't be a total free-for-all. (00:39:35) So what's your take on this? (00:39:36) Is it just somebody gave a bad analogy here, or maybe some people were weaseling their way into the White House to try to shift things when Sachs was back at home or something? (00:39:45) What's going on here? (00:39:46) Give us the, because that's what people say. (00:39:49) They say the last person to talk to Trump kind of has his ear and that things can bend a certain way. (00:39:55) I don't think it's a, I think that there's a pretty profound vibe shift.