We're going to start with the news of the day, which is Washington lifts the 19-day Fable 5 ban. It's a quagmire β you've now been entrapped in some kind of preapproval process, and that hasn't been finalized yet. But the zoom out comment is: six months ago you could ship software like a free man, and now you have to get permission from Washington before you do it. It's a big change. I think part of the reason the US is such a dynamic economy is because we don't have a ton of that. Europe does, and now we do. We sneered at GDPR and here we are.
It's like rewriting Atlas Shrugged where John Galt goes to Washington and says, "Why don't you regulate me more? Why don't you take us, Mr. Mooch? Grab some of my stuff." What the hell are these people thinking, volunteering for this stuff?
The world's changed to me. Sam Altman offering 5% of his company to the US government was much more interesting in some ways than whether some suboptimal but inevitable oversight is coming to the LLMs. What problem is he trying to solve? By definition, it's not any of the security issues β it's some kind of macro "AI is going to destroy everyone's job, so we've got to give back." OpenAI produced a nine-point plan about restructuring the taxation system of America to tax more on cap gains and less on income. Remember for context, 5% of Anthropic is $50 billion. Congress raises plus or minus $5 trillion a year, so that donation gets rid of 1% of a raise for one year β and they should restructure their entire taxation system? Sure, we'll get right on that. The House Ways and Means Committee will call a hearing, and it's kicking off a process you won't be able to control.
I completely agreed with you at first, right, 100%. But then I step back β Sam Altman, beyond being CEO of OpenAI, is one of the most successful investors of all time. This is not Intel dying. It's like giving 5% of your company to Shopify like Klaviyo did, so they don't destroy you. Giving 5% to plate the federal government so that you're the good guy now β I think as an investor I'd take the dilution.
Open AI, Microsoft owns 30% of OpenAI. If ownership stake resulted in besties, they'd be besties. They're in a stale marriage looking for a divorce but can't quite pay the tax. Now apply that to the US government. The idea that because they own some of you they'll align with you is just not the way politics works. Go back and look at TARP.
If you really are destroying labor in a $30 trillion economy, do you think the political monster is going to say "I'll settle for five, that's grand"? Bernie's already said he wants 50. Because they believe rightly or wrongly that the impact of this technology is so important that all these things need to be on the table. That belief is what gave them the confidence to raise billions of dollars. But if you believe this thing is dangerous from a cyber and jobs perspective, all these things become next-level logical. And if on the other hand you believe it's not going to put 50% of the US labor market unemployed, then these preemptive changes are a wild overreaction.
Is this purely a marketing exercise? Who are you marketing to? Congress, senators β you're willing to align yourself, you're not this wolf stealing jobs. If I'm going to ask you to regulate or tax Chinese or open source models, maybe it'd be helpful if we had alignment beforehand. Sam is anchoring this idea that hey, 5% will align us with the American people, with the federal government, rather than 50. Maybe it really doesn't matter what we think because it's already happened β the decision essentially has already been made that the federal government will be acquiring a stake in OpenAI, and Sam is anchoring it as the smallest possible stake.
To be clear, the proposal from OpenAI wasn't "we give OpenAI 5%," it was "companies should" β implying everyone should, including Anthropic. So to some extent he's volunteering other people's capital. It's funny β when the internet took off, the whole emphasis was "cut us free." The telecom deregulation act broke up AT&T, Section 230 protected websites from liability, no sales tax for a long time β all of it was "leave us alone" and Silicon Valley did great. Now we're going the exact opposite way: "don't miss us, regulate us, pick us." Oil and gas must be looking at this going, wow, these people are crazy.
I think in the age of AI, massive delusion has been institutionalized. Even two years ago, most founders were fairly dilution-sensitive. Now I find founders β you look at really hot startups and if you peel back the layers, they've done 16, 17, 20 venture rounds, often. Even if each one is 5% dilution, 20 rounds at 5% dilution is a lot of dilution. Look at Anthropic β you've got Dario at 1 point something percent equity, Sam's at nominally zero. Anthropic is the most successful startup of our lifetimes, but the founder owns 1 point something percent.
No one's worried about making their last round high-priced investors money anymore. Literally, no one is β because I believe investors have learned to accept 1x when it doesn't work out, without drama, without blocking, without threats.
It's really odd where one of the two CEOs owns 1.7% and the other owns zero. It totally takes the edge off the dilution conversation because it's someone else's money. As a seed investor I've watched myself be diluted to levels I never thought would happen. I used to think my real entry price was twice what it looked because of dilution, now I'm thinking it's four times.
The only fact-based comment I'll make: the data from Carta says dilution per round is going down. So maybe founders are willing to raise more because the dilution per dollar is lower β you can do more rounds and end up with the same dilution. I'm seeing both β smaller rounds but so many rounds. If I'm doing that 5% to hold off any regulatory issues, man, just do it.
You want to have between 5 and 10% of an investment to matter, but you're seeing now Spark's going to do amazing and own 1% plus or minus. SpaceX β the investor who did the original check when the rockets were still blowing up β is sub 5% or 3-4% of SpaceX. For these huge outcomes, the mental math gets turned on its head. If you have three orders of magnitude larger exit, you can get away with just about anything on the dilution side.
One video that was going incredibly viral was Alex Karp on CNBC, where he said two things I think were standout comments. One is there's never been more skepticism from large enterprises towards frontier model providers, specifically Anthropic and OpenAI. Second, there is real questionability from those enterprises on the ROI of AI within their organizations.
Corporate America is saying "I'm spending all this money, am I getting anything?" β which is the ROI comment. And corporate America is saying "am I giving them all this information, are they training on it, are they learning my business and going to sell my business to everyone else? What's my IP?"
Obviously it was a self-serving comment because then Palantir will solve these problems for you, Mr. Corporate America β and the stock went up 9% on the day. Stylistically it's a crazy style, but oh my god, the points were spot on. Anybody on the application side is going to be sensitive to token model costs, that's real. Whether OpenAI and Anthropic are really training and slurping up all of our data seems slightly exaggerated based on their terms of use today. But this was the same week that HubSpot had to walk back that it was going to share all your prospecting data with other customers β pull all of Harry's, Rory's and Jason's verified contacts. Their customers erupted and they rolled it back within a week.
I think vendors overall are going to push the limits on training on your data. OpenAI and Anthropic kind of lied about the books, and they definitely lied about training on YouTube. Every vendor seeing massive competition or slowing growth is going to be tempted more and more to cut corners on training privacy. If you're Palantir selling to the government and highly regulated industries, that's a great play β you can't really trust these guys not to share your data.
The other thing Karp mentioned correctly was that Anthropic in particular had "opinions" about how their AI should be used by the DoD. When people are giving you millions of dollars, they don't want your freaking opinions, they want your technology. He did a very good job of positioning himself on that side of the table.
We saw this week Meta launches a cloud business to sell excess AI compute and compete with the neoclouds β hosted or raw GPU rented by the hour like CoreWeave or Nebius. Market reacted well, 10% jump, single day gain, biggest in 5 months. Why not? If you've got the capacity, why not lease it? Didn't bother SpaceX, didn't bother Amazon opening up AWS back when it had excess e-commerce capacity.
Two companies have done the same thing: buy a load of compute to build proprietary assets, fail to build those assets, then decide instead to sell that compute to others. Both had a very positive market reception. You ask yourself: is the market thinking there's a goldilocks scenario where short term Meta has excess compute and long term they have a wonderful use for it? Or is the market simply saying both of you have failed at your long-term goal but being a cloud provider is a great business, go team?
At the margin, the entrance of SpaceX and Meta into the neocloud business was worth exactly the 10 to 15% decline for Nebius and CoreWeave. The bad scenario is if a whole load more companies go through the same journey β we think we need all this compute but we can't build something useful enough for it, and then you're only left with a few buyers of compute (OpenAI and Anthropic) and a whole lot of sellers. Everything is true up until the moment that compute demand isn't there at the margin. That's not happening now β it's never been tighter. But if that changes, all these assumptions go out the window.
Nvidia starts financing its own demand with compute-now-pay-later β essentially letting providers access GPUs through revenue sharing and credit support instead of paying upfront. They're basically selling you the chips up front, recognizing that hardware revenue up front, and giving the buyer put-back rights if they can't use the compute.
It's basically hedging the risk. It's accounting legit (ASC 606) but it is pretty aggressive. Their top three customers in the data center business have gone from the 80s to the 50s or something like that as they diversify away from hyperscalers β but there's a lot of contingent liability they're taking on. As long as the raw demand for compute keeps going up and to the right, these deals will look wildly smart. If that slows down and there's excess capacity, these deals will look horrible β you'll be debooking prior revenue because your customer went bust. I don't think anyone's managing for downside right now. The time to manage for the downside is when no one is managing for the downside.
Anthropic opens talks with Samsung to build its own AI chip. And today, DeepSeek announced they are starting to build their own chips. Is this the natural progression of an ever maturing industry? Will everyone build their own chips?
I last week said I thought it was mad, but I saw comments β you got to own the compute, if you don't own the compute you're screwed, a little like crypto: if you don't own the keys you don't own the asset. The second argument is more technical: if you build your own silicon, you can optimize it for your model and get significantly more efficient than buying a general purpose computing platform from Nvidia. But I still find myself going β if you're at the app layer and that's where your value is, and then you have the model, then the hosting provider, then the chip β needing that amount of vertical integration just feels weird.
The only thing that makes zero sense to me is the argument that we need to build our own chips because we have very specialized needs that Nvidia can't meet. If you're driving that much volume to them and you need a special version of a chip, they'll build it for you. This is just responding to believing that the margins are so high, Nvidia needs to recapture that margin. Everyone's kind of dancing around it.
Kling raises 2.8 billion at an $18 billion valuation. Biggest AI video business in the world, doing 500 million in Q1 ARR, clearly going to go public on the Hong Kong stock exchange soon. Interesting in the context of OpenAI shutting down Sora. Why if Kling can pull this off, why the hell couldn't Sora pull it off? Higsfield just announced they're at 500 million in revenue, actually doing 2 million a day now in credit card billings. Kling is just one of the models they use.
Is there a Chinese valuation bubble potentially in AI, like there have been in prior rounds? DeepSeek raising at 50 billion, a gross discount compared to any western alternatives, and Bytedance at... it's a counterargument. It wasn't clear to me a year and a half ago that video demand would be this insane, but now that people are beginning to build films on these platforms, the amount of video you can consume is infinite.
Sora was pretty good when I'd run all four together β Kling, Sora, Vio and the other big Chinese one, you can run them all on Higsfield. It's kind of a bummer they shut it down, they just couldn't make it cost effective, and maybe a slightly inferior product. If you're OpenAI, your highest and best use of that compute is enterprise-centric where you can make real money β more money in coding than consumer video. 500 million to OpenAI/Anthropic today is nothing, it's below the materiality line and a distraction. For Kling it's not a distraction, it's a wonderful business.
The most commercially successful AI video product on earth is Chinese. The top six models as of today on OpenRouter are Chinese. Do you think China's running away with the model there?
The US is clearly running away from it in terms of frontier models, and the Chinese counter strike has been open-source models, distilled in some part from OpenAI and Anthropic, but they are clearly numbers one to six in terms of the non closed source financial model. I just got back from two weeks in China and Hong Kong β now I think Jensen was right about this, because when you're in China, OpenAI and Claude and Anthropic will not serve you. You cannot access it. What do you expect China's going to do? The second largest economy in the world is going to build things that are as competitive or better than we are because you can't even use ours.
Actions have consequences. If you believe there's a national security concern on these models and chips and you've made that decision soberly as a government, you can choose to block access. But you can't expect the other side to say "okay, you caught us, we give up." They said "no, we'll build them all." There's information out from China that the Chinese government is saying maybe we'll deny access to overseas users of Chinese open source models too β kind of hilarious since both sides think letting the other use their models is dangerous.
Microsoft launches $2.5 billion and 6,000 people to embed engineers inside enterprise clients, targeting the MIT finding that 95% of enterprise AI pilots deliver no measurable P&L impact. Amazon made the same move 2 days earlier. Is this a continuation of the shift from a model to a services ecosystem?
I think it's going to fail. We have some of the best FDEs at these leaders. One leader's FDE went on paternity leave for 3 months and the new one told us they couldn't fix our bug for 3 months. This was a public company that had to wait until August to fix the fact that their AI was still talking about an event that already happened in May.
I disagree, I think it will work in a limited but interesting sense. I don't buy the 95% MIT stat as gospel, but I do buy your story β even a smart company needs assistance to make this work. The solution is not "don't get support," the solution is someone has to build a business with people capable of answering your questions. Corporate America is an oil and gas company, a banking company. On the other side of the table you have Anthropic and OpenAI who are product companies to their core. You need something in the middle β services companies to help them adopt.
Every technology company either goes bust or lives long enough to become next generation's IBM. IBM was the enabler to the PC and to some extent the cloud, helping corporate America adopt when you don't have an amazing product yourself but you do have large enterprise trusted relationships.
Now OpenAI and Anthropic are the companies with the new incredible product, and Microsoft is the more mature company with the enterprise relationships who is going to build a large services business β just like HP did, just like IBM did. It won't be nearly as profitable as selling operating systems. The depth even at the hottest companies is not there β there is no depth to the FDE chart. It's going to lead to a lot of tears, but it's still better than trying yourself. We're back to the early days of B2B software where a couple people understood how it all works and no one else could solve the problems.
Every deployment at Harvey has an FDE and a lawyer. If you have a lawyer and a very experienced technical resource deploying Harvey, that might be what you need for a successful deployment. When all you're buying from a vendor is a database, all you need is a database expert β but when you're buying intelligent answers about your own business, you'd better be sure those answers are grounded in your industry's facts. Everyone will probably be some combo of tech expert and domain expert.
Ashton Kutcher, one of the most successful investors of the last few years in terms of SPVs and OpenAI, Anthropic via Sound Ventures, announces he's leaving his own VC firm to start a new one with Morgan Bella, previously at a16z and then NFX. It's just crazy to leave your own firm like this, but I actually don't think there's a deep dark story here. I think this guy is so successful β he doesn't need to worry about the firm's brand, he's already the name people remember, not "Sound Ventures."
The 11 Labs secondary at 22 billion β I don't think it's a high valuation, but the one I think is interesting is: as an employee today, why would you join something that you don't believe will have secondary options? If I was a hyper talented employee, I would not want to go somewhere without liquidity. There's only so many 11 Labs that can pull off a tender offer of 22 billion. The whole trick for employees, just like it is for VCs, is to join something that isn't doing a tender offer today, get a healthy grant, and join a company that within a year or two when you've vested 50-60%, starts doing tender offers. Tender offers are the proxy for public now that the IPO window takes 12 years in some cases.