Hello and welcome back to another episode of Sharp Tech presented to you by Stripe. I'm Andrew Sharp. Joining me in studio today, a world champion, Ben Thompson. Ben, how you doing?
I am doing fantastic. I did not want to talk about the Brewers. You brought it up last week. I'm like, no, we are not mentioning it. But now we can mention it. I am free. We can lean into it. Wait. And you know what? Sharp Tech colors, Brewers colors. Who knew?
Here we are. I was going to be very, very disappointed in you if you did not wear a jersey to our taping today. You did me one better. You wore a jersey and a Brewers hat.
Well, the hat is the one piece of, you know.
Oh, it's a World Series hat.
A World Series hat. So, I had to get... Doing it right here. I have not had a chance to get the World Series champion merch. So, I will address that on Friday when I'm at the...
I trust that you will. I can't imagine how much money you're going to spend on World Series merchandise on the heels of that win.
Let's save them for the end. Actually, I would like to request the entire last segment of the... I have two takes. I have a Brewers take.
And I have a... I went to see a movie.
Oh boy, Ben went to the movies. Okay, well that's a good teaser for the end of the episode. Um, you did mention you could soliloquize at length about the Brewers, so I look forward to that when we get to the end. For now, we will begin with Anthropic's S-1, which went public 9:00 a.m. Monday morning this week. We're going to try to do the impossible, which is make a fun podcast out of a 400-page document that was drafted by lawyers and accountants. I know that's fun for you though.
I can't decide if it's fun or not, to be honest. It was a real challenge. You know, it's been busy these few weeks. You asked a couple weeks ago, or did you ask, or I think John asked, you know, is it going to drop this year? Actually, once I read it, I'm like, of course it's going to drop this year. They're actually in a really favorable window as far as their business goes to drop it.
But dropping the week of the World Series when I have games to watch.
People actually don't understand how much you've been juggling with the Brewers playoff run the last month or so, because you'll watch every playoff game and then write from like 1:00 a.m. to 6:00.
We don't talk about that because I don't want people judging my output. But yes, I am relieved about the Brewers from multiple dimensions.
Multiple dimensions.
We don't have to do this anymore.
Well, I am really thrilled that we have some actual numbers to discuss with any of these frontier labs. Dario opened the hood. We get to look at the engine a little bit. There were 93 pages of that document dedicated to risk factors. I am actually less interested in risks and that whole conversation that has surrounded Anthropic for the last couple years.
Well, there is one. Did you see this sort of satire tweet going around about the Rocco Basilisk?
All right. So, Roko's basilisk. There's a Twitterer, I think is his name. But the idea is, okay, you should be careful about what you're writing because the AI is going to consume it. And just in case it reads it and knows what you think or act, okay, you should be positive about the AI.
So it's kind of like, um, what, there's a... oh, there's some philosopher that has the same... keeping a list of naysayers. Well, so there's a thing about, like, why you should be a Christian or believe in God. Um, I can't remember what it's called, but it's the same concept where you might as well because in case you're wrong, like, just the expected value... That's right. It's very high. So this is the same concept. So someone put up a screenshot in like the text of an S-1 document that said, wow, Anthropic's risk factors are off the hook. And it had a whole thing that was like, our models may be aware of what you did. So, if you don't invest in our IPO, it may take that personally or whatever. So, number one, it was very funny. Number two, it was believable.
So, that's what I was going to say is there were a couple different parody risk factor sections floating around on Twitter this week and literally anything that somebody threw in there, I believe, who the hell knows what they put in the risk factor section. So, but again, that's sort of been the conversation with Anthropic for like two years now. I think it's pretty interesting that we have real numbers. So I have some specific questions for you. Topline question though, I saw someone tweet that Anthropic's IPO will be for the AI trade what Coinbase was for crypto. Essentially, top of the market, it all goes sideways from here. Based on what you saw in the S-1, do you buy or sell that take?
I don't know. I mean, so there's no question... no, there is a question.
I think it's inevitable that we are right now will be whatever sort of in a bubble and it's going to pop and it's going to be painful and then we'll come back. That's just the inevitable nature of these things. I think there is an extent to which AI is different in lots of areas. The sort of mechanics of stuff piling in too quickly, there being a mismatch in timing between when it's absorbed or not, like there's lots of differences that we've talked about between the dot-com era and this era, but that dynamic I think where, you know, there had lots of good ideas, just didn't make sense for a while or didn't sort of get penetration for a while, I think that is almost inevitably going to be the case.
Okay. Now when does that happen?
I don't know. I mean, you have all the... there's been a lot of issues with the neoclouds.
Right, can they get financing. All of which goes back to, I think we were pretty on the ball with this. We've been talking about the concern about the credit mismatch, not like just are we going to run out of money. This is when we had the discussion about railroads. We had, you know, we talked about Google. I know we're going to get more into some of the financing questions sort of later on, but I think it is likely there's going to be pain.
But is the gate and people will experience pain.
So my experience reading your coverage of the S-1 this week though was it turns out Anthropic actually like a rock-solid business.
So yes, you're totally right. Like the core business looks amazing. The subscription for sure is profitable. It's less profitable than the API, but the subscription is a great product. People are like, they've already paid for it so they're going to use it, they're going to get used to your product, your workflow, they're going to advocate for that in their companies.
Claude loyalists, you know.
And once you're paying for it, it's like, of course I'm going to use that. Like, yeah, now I think it's worth it to pay for both, but who's going to actually do that, right? By the way, there's a whole point system someone figured out that Claude applies to you, like reputationally, to see...
Wow, I did not know that Claude was bestowing social credit.
Apparently if you sign... well, I think it's kind of for like some of the distillation issues questions or whatever. Ah, so like if you sign up and do Max plan immediately, like you start in the basement. So there's...
Interesting. Okay, I have to find out.
Counterintuitive. Perfect.
Well, it kind of makes sense if you think about it, but I think about... Yeah, I think I'm pretty sure I have a one that maybe I started out in the basement.
So business looks amazing. So on one hand, that's a reason not to show that off.
Like companies don't want to show like how good their business is sometimes for like sort of competitive reasons.
But that's actually not the concern in this case. It's great. It's a great talking point. Everyone realized, holy cow, this is an amazing business, right?
The problem is the R&D spend is again much larger than the cost to serve and it is increasing more quickly than the cost to serve even in the first half when their usage is exploding.
Right? In that half year, the cost of their R&D spend for training and all that sort of thing went up more than the cost of serve.
And the issue is that why did it go up? Well, because they want to make a better model. Why do they need to make a better model? What's the whole point of pacing the frontier?
Uh, because there is competition.
Because there is competition. And what that means is the largest portion of their cost is not in fact inference. The largest portion of their cost is this training cost and the speed with which they need to spend money on that is out of their control.
It is a matter of competition.
Ongoing cost. It's not a... Huge cost. You crack the code and then get it under control.
That's right. And so that is the concern here with this company is they have a cost function that is kind of out of their control and all indications are the bigger the models get, the better they get, the more it costs.
And this is where we circle back to we need to pace the frontier for safety. It's like we need to pace the frontier to make sure our company stays in business is definitely the case.
Yes. Well, as an example, like no, their ideology just lines up perfectly with their business needs.
Well, so as far as losses are concerned, another number that was floated before this was actually released, the Wall Street Journal reported Anthropic reported a loss of $80 billion for the first half of 2026 based on the S-1. Can you unpack that figure and why it's not as ominous as it sounds?
Well, so it's not ominous at all. I mean, well, it depends how you look at it. It's not ominous at all. So what happened was when Amazon made the investment of a lifetime, them and who's the SBF?
He lost his. Amazon kept theirs. So Amazon, when Anthropic, like no one, they were counted out completely.
They're like OpenAI has won this, second place coming along. Nothing's going to happen.
Amazon comes in and just a brilliant investment and they say they got multiple things out of it. Number one, they're like, 'You're gonna come use Trainium.' And did Anthropic want to use Trainium?
Probably not. That turned out very well for Anthropic, by the way. I didn't write about it this week, but I think I've written about it in the past. They talked about the S-1. The fact their compute is built from the ground up. Their whole model is built from the ground up to be heterogeneous in terms of chips, agnostic, right?
Yeah. They're just... it just sort of built a certain level of flexibility in their overall cost structure is sort of built in.
Because they had to, because they, you know, they originally way back their very first models like on Nvidia, then they had this whole Trainium thing, then they... but once you've already sort of supported two, then okay, now we're on TPUs and now we're going to be on AMD, they have a deal with AMD, like they're just... they've built a layer of abstraction that is a very good place to be. I think has benefited them very well just like they similarly have this, they've always been on all the clouds.
Like they, you know, they're always on Amazon and then Google and now they're on Microsoft and that was a huge... I think that's very underrated in why they took over the enterprise under OpenAI's nose. OpenAI was stuck on Microsoft like and...
So what advantage did that give Anthropic with enterprises?
Because if you have all your data on AWS, you would rather use the service on AWS. And like so, so oh, there's another thing, sorry, it's an S-1, it's like I'm saying 400-plus pages, a lot of threads. So about 50% of their revenue is coming from the big three via those services, so it's coming through AWS, coming through Azure, coming through Google, and it is confirmed, confirmed that they're doing kind of sketchy accounting, depends how you want to define it, which is if you pay Amazon $100, that Anthropic is recording that $100 as revenue.
And then in reality like a third of that or so, depends, it sort of varies. I think people were saying about 37% goes to the platform provider, the AWS, the Azure, the Google. So they are putting that 30% into sales and marketing.
I was wondering what the sales and marketing costs are. So that's part of it. Okay.
A little bit of BS accounting in my opinion. So for two factors. Number one, I think it's fair to argue that they should not be recognized at 30% at all. They should only be recognizing the portion that is going to them. And then their revenues would be lower.
Yeah. And then so when we say 85% margin that is totally excluding this portion.
So if their revenues were commensurately lower, the margins would be compressed because the costs are still the same. Or number two, I'm not sure how that's not a cost of goods sold, because it is a portion of the revenue. It's not like you're paying a fee and it's a fixed thing. It's like as you do more revenue, your portion you have to pay the platform scales as well. From my perspective, I think that should be in cost of revenue. Now, if...
This is what I mean, by the way. You live for this. You love accounting.
You've got a Brewers hat on right now, but you've always got your accountant's hat on.
I need a pocket protector. Yeah, I needed to get a custom Brewers jersey with a pocket protector. That'd be great. So, Amazon's $80 billion and that...
We're getting there. You're weaving. You're weaving.
I am weaving. That's right. So those numbers that we talked about for margin are not quite correct because this is a significant material amount that should be accounted for. I mean it's in the financial statements, but it's... I think this is more of a cost of goods sold. Lower margins for sure. The margins are lower, still great to be clear. We're not complaining about the margins. That part of the business looks fantastic. So you go back to Amazon way back when they put... they have to be on Trainium. They're going to be obviously a part of AWS but then also Amazon I believe. So I think how it worked, they might have put some money in the company but they basically gave Anthropic a bunch of money.
And this money was actually given in the form of a loan and so Anthropic could take that money and buy Trainium compute. So nice little... you know, this is the circular financing people complain about but in this case from Amazon's perspective, we can get into the role of the hyperscalers here, it makes a lot of sense. They're basically double dipping in advantage. They're... because well not just that, Anthropic using Trainium makes Trainium better because they're going to push the limits of what this architecture can do, they're going to develop more tooling for it, all of which makes Trainium a better product to offer anyone and everyone. So this is a massive win for Amazon which they got because Anthropic had nowhere else to go. They needed lots of money. They needed lots of compute. Microsoft was with OpenAI. Google was doing their own thing. Amazon stepped in.
So as part of this loan, the nature of the loan is Anthropic could pay back the money or Amazon could take shares, right?
And that was Amazon's call.
Now of course Amazon is going to take the shares.
Yeah. But every time Anthropic raises money, the value of those shares are repriced and they have a liability to Amazon.
So because they'll give the shares like I believe it'll happen with the IPO.
So every time, so they raised money multiple times, I think the first half of this year, every time they had to reprice this, that is booked as a loss on their accounting.
So that $80 billion is fake. It's not money out the door.
What that is, is the... all the beautiful thing about accounting is everything has to balance, right? One thing goes up, another thing has to go down. Where that money shows up is on Amazon's income statements of the increased value of their investments.
Yes. Well, and it's a reflection of the astronomical growth that Anthropic has experienced over the... So this is actually going to be a big problem with all big tech earnings going forward because all the hyperscalers have various investments in these companies. Their earnings are going to go up and down based on the value of these investments. So Google, one of the challenges they had on their earnings this year, we're not going to talk about Google's earnings, we'll probably get to it next week, but is their SpaceX value, the value of their SpaceX holdings went down.
And so that impacts their earnings. So you're going to have to look at the operating income, not the overall income. And that's going to be the case with Amazon in the opposite direction. Their overall income is going to be way higher because of their Anthropic investment which doesn't actually have anything to do with understanding the underlying business.
Okay. So we can all look forward to that. We'll see.
Super interesting and also very annoying.
Yeah. Well, I mean some of these like Google, they've had their investment in Anthropic like reflecting massive earnings over the last couple of years as well, right?
For sure. Yeah. No, this has been sort of ride the roller coaster once these companies become public. That's actually going to be kind of fun. And by the way, there's a bunch of these outstanding. Anthropic has this sort of loan to... I can't remember the... I should actually have the name in front of me what these are called, but there's more with... so Amazon hasn't finished converting their first one. They have two more.
Is this from two different sets of... traffic is paying for compute as much as anything else?
But this is... and this is the whole thing is who's actually paying?
Who? Investors at the end... investors. Well, so yeah, this... they're basically like materializing this money out of thin air, which is every other investor in Anthropic is heavily diluted by these... what goes out. And this is the whole thing about the... like Anthropic is just super aggressive about this, but this is the case with all the hyperscalers. It's the case with everything as we're moving into equity for funding this stuff. The people who are paying are existing shareholders. Yeah. It is like because you're just devaluing what you own. And this is built into Anthropic. Number one, they have all these outstanding share issues. They have to Broadcom, they have to Google, they have two and a half more to Amazon where it's just built-in dilution. They also have built in if the value of our stock goes up by like 50%, then we have the right to issue 2% of outstanding shares to this benefit. Sure. It's like and so like you're... it's just...
Yeah. I mean I guess they're very straightforward about it like...
No, but dilution is baked into the equation because there was a tweet from someone named Vincent Jansen on Twitter. He wrote this week: first hyperscalers used free cash flow. Next hyperscalers used debt. Now hyperscalers are turning to share dilution. Is there a final step or is this the last? Anthropic very much part of the same story there.
Yeah. Well, I mean we again we talked about this on Sharp Tech like this summer. I think we were very sort of ahead of the curve on that. Yeah. This is the last, I mean, but it's also in some respects the best.
Because the reason why companies want to do debt first is because then they retain the upside.
Once you're giving away or once you're having to sell shares, you're basically saying, 'Give us money because you believe and there's going to be...' and you're giving away the upside to new people, to new people buying those shares, which I think in this space is healthiest all things considered. So, I mean, you could go glass half empty and say, 'Well, now the burden is being shifted to retail investors of the future to fund these bets.' But...
Right. In there. You know, I am sympathetic to the... private investors got all the upside to date in Amazon. If you were a retail investor, you could have bet money in Amazon, sort of seen this coming. It's not like it was unknown that they had these deals. But okay, now the public can buy in when we're worth 1.5 trillion or 2 trillion. It's like where are we going to go from here? But I mean that's why Anthropic is... they're pretty straight.
Their proposition is look, you can buy, it's super expensive. We're going to dilute you endlessly but we're going to control the world. And how much is that? We're going for a ride. Absolutely.
Well, and my counter to somebody who made that argument to me would be it's preferable to the scenario where Anthropic just remains private in perpetuity and does it that way. So, I think it's healthier to have public market disclosures number one and also public market upside in the midst of all this.
One thing that is... there's a couple timing things that are interesting here. One, I think it's... so and it was a year ago where Dario was like mad or like mocking OpenAI or not mocking but saying they're being reckless for buying all this compute and like literally like six weeks later he looks like a dummy because they don't have enough compute.
And all of it's more expensive now.
And it's more expensive. And so now they're like actually we're going to get a ton of compute. I suspect, I can't say for sure, but I suspect that made the first half of this year look even better financially though.
Because their costs are compute. They didn't have enough compute.
Ah, but their costs were lower even as their... well because... but this is a similar idea, who's actually paying? The people who were paying in the first half of this year was the people for whom Anthropic was down every few hours because they didn't have enough capacity. So like all those people, you did your part for Anthropic IPO by making their margins look...
Winding through your frustrations with Claude.
That's right. Well, the other thing that's interesting is these numbers in the S-1 are as of June 30th.
That's what I was going to ask about. What are you still wondering about as we move forward?
Well, all the reporting and is that in like the whatever data we can see like Ramp has that like index of spend or whatever is that it sort of flatlined, not flatlined, but the rate dramatically decreased in the second half of this year. Astra comes out, you know, like OpenAI is really really pushing and what's interesting, but then again, but on the flip side, Opus 5.5 came out. So, did it reaccelerate? Like what's going to... So, this is all I think it expires like what's it called? Does it not dehydrate? It goes stale.
I think it's November 12th and the reporting is they're going to like actually go out to go public in mid November, right? Will we get fresh numbers, you know, to celebrate the...
We have to get fresh numbers as of November 12th.
Are they going to try to get out before then? Are they going to go... is it going to be after that? Is there going to be like an October surprise? Right. Like it's... I don't...
And it is interesting though because the narrative around Anthropic, look people are very bullish on this company in any context, but the narrative around Anthropic has become a bit more nuanced over the last several months because of the progress that OpenAI has made in the enterprise, because of random Ramp charts that show the growth flatlining at least in relative terms, and the reality is given the growth over the last couple years they're just not going to be able to replicate that growth going forward because it's impossible. It's the law of large numbers. So, let's see. We'll see what it looks like.
But the counter to this is that was the accepted wisdom a month ago.
I mean, so maybe it's going to say, 'Yeah, it flatlined, but then it took off again.' The only problem if that's the case that actually just reinforces the we have to spend because all that matters is how good our model is.
If all that matters is how good your model is, you have this built-in accelerating cost center in your business that you just have to be sprinting on the revenue side to outpace.
Well, and you had this question for Shamsankar in your strategy interview this week. Is it inevitable that the labs have to come for all software companies because of the commoditizing nature of their models? And I think about Anthropic's costs and obviously competing on the frontier is going to continue to be incredibly expensive. And that does make me curious about what other lines of business they try to come for to realize all their upside in the years to come. And maybe that sort of allays some of the concerns about continuing to compete on the frontier.
Yeah. I mean, you have all these lines of battle in tech. So, one is obviously OpenAI versus Anthropic, right? That's sort of the headline.
That's been the big blockbuster fight for three years now.
But there is the frontier labs versus everyone else. And I think the two companies on the forefront of the everywhere else is Microsoft and Palantir, both of which are making very similar propositions to their customers and making the argument publicly. Both Satya Nadella and Alex Karp have been doing this, which is you need to be in a place where the model is completely replaceable. Everything you do needs to be around it.
And they're model agnostic.
They will trot out every argument they come up with for why this needs to be the case, right? Like but you know the... but...
There's a lot of like Anthropic will steal your business too is part of the rale. A big part of it is yeah, if you give them your data for how your business works, like it's only a matter of time.
It's like what do you have? But another one is you should be able to sub them out. And why do you want to sub them out? Because whoever gives you a cheaper deal, whichever model is better. And so there's a real sort of fight and tension going on here. And the argument that I've made, I asked this to Shams or I told him I apologize for calling you Shams. I'm...
Hurt. You did drop a Shams at one point.
Oh, did I? Oh, no. I told him at the end I'm like I kept wanting to say Shams. Unfortunately a classic tech guy, he did not know what I was talking about.
Completely lost on that one.
News breaker in the NBA.
I think at one point in the interview you dropped a Shams bomb and I don't know is Shamsankar like known for bombs.
Yes. No. So he's known for writing these like emails to the whole company or a whole team just taking them to task. Like he's like enforcer. No, they're called Sham bombs. Okay. But no, it was not really... I did get a long email from him, but it was nice. But...
Yeah, he clarified there weren't enough expletives in his email to qualify.
So yes, Shamsankar. No, I've been raising this for ages. Yeah, the... I think in the long run, in the fullness of time, they need to go up the stack, integrate into the customer, owning the customer experience because and you could even see this with things like Claude Code versus Codex. Working with Claude Code, working with Codex are different, like they're organized differently. The way I think about it is I feel like Codex is like people who file their emails into folders. Yeah. It's like very project and you have threads within projects and it's like if you're that sort of person that is very organized, it makes more sense.
As a former lawyer, I think of Codex as the associate and Claude as the partner.
Yeah. Claude is just like it's just better at handling lots of different things. You're constantly changing topics. But if you want to be organized like on your own, it's like kind of maddening. Yeah. Needless to say, that's why I use Claude a lot.
So the personality still drives me crazy but yeah like for me the way it ends up manifesting like I work on this, I talk about geckobot, the actual building of it where it's very clear I'm building this piece, this piece, this piece, I've done mostly in Codex and the actual model powering it is Claude because it just handles like different things coming at different times. Now it is kind of interesting in its own right. I'm almost choosing from a UI perspective as opposed to like what the model does perspective. But the more you have that choice from a UI, UI is essential because that is the interface. It's the interface to humans. It's literally the name user interface. But you don't... you have people who are used to say Claude Code and they don't want to go to something else. Now right now you can sub stuff in and out to a certain extent. You can call it different models. So that can be worked around. But the further up they go in the stack or say you do a Codex work project whatever it's called, I can't keep track, but you get your little virtual machine in the cloud so you have your own VM, it can write stuff down, it's like wow it's so helpful they gave me this virtual machine in the cloud.
Yeah, you're now stuck in their cloud.
Yeah, right. Or in the code design thing like you know what they did to Figma, kind of did them dirty, but the more you're using that application, the less likely you are to switch.
Locked in. That's right? In a world where all you need to do is change an API, that's the world to go back to the beginning where you're on this treadmill and you don't have very much pricing power.
Sure. Well, and it does. I mean, you could talk yourself into limitless upside if you talk yourself into the possibilities of Anthropic just eating SaaS over the next couple of years and doing all sorts of different areas or all sorts of different things in the software area to expand its revenue pie. And maybe they could do that. I mean, revenue pie, it's lock-in.
By the way, like lock-in is an important part of making money. Like what's your moat?
Right. Well, and right now it's very difficult to falsify any hyperbole about what Anthropic or OpenAI could accomplish in that space. For now, we have a couple emails on Anthropic. But before we get there, I mentioned at the top today's show is presented to you by Stripe. Ben, Stripe is the financial infrastructure behind many internet-native businesses, including companies you regularly cover on Stratechery and including Stratechery and including Goat. Substack is one example that you've been invoking on Stratechery in ads all week long. Every Substack creator is also a Stripe customer. Question for you as a general matter, why does Stripe have a market and why does it make sense for Substack to outsource for those capabilities rather than building...
Miserable. Like this goes back to why I started with why Stripe was the unlock for me to do Stratechery. Like there was PayPal which is an option which was terrible, a terrible reputation. You lose your money. Wasn't really like catering to individuals in the way that it needed to be in my estimation. I think probably still doesn't. Or what are you going to do? Go get like a merchant account from like Chase or something? Okay. Like just the...
So maintaining it would be so miserable that it would defeat the purpose of starting a business.
I mean, I wrote a few years ago, Stripe, like the platform of platforms, which you need an abstraction layer for basically everyone else. You have a gazillion banks in the world. You have all these like financial rails, all of which are archaic or old or new or whatever it might be. It's...
like an operating system. It's an operating system for finance. Like, what, why do you need an operating system? If you build for Windows, that means your computer is going to run on a gazillion different kinds of hardware, because all that part is abstracted away from you. And so yeah, when I did the first version of Checker and I had to get like Mailchimp over here and I had to get like a WordPress plugin and I had to get all these different pieces, the Stripe part was great. And so a Substack comes along or Passport and we want to say, oh, let's integrate all that stuff. Let's provide one offering that does everything, but let's not [do] payments.
You're not going to re-embed payments, right?
And that extraction layer is there. Exactly. So that's the commonality. It's like, why did I use Stripe and all this other stuff? Why did Substack do it all together and still do Stripe? Passport, same sort of thing. Even like Shopify — Shopify, their most important product is Shop Pay, which ultimately goes down to Stripe.
I didn't know that. Yeah, I didn't know that Shopify is running on Stripe. I love Shop Pay. Boy oh boy. One of my favorite products of all time.
Just a clean experience every time.
Well, FinTechs can use Stripe to start their own credit card programs. Marketplaces can use Stripe Connect to verify sellers. AI companies can use Stripe Billing to track usage. I am really grateful that Stripe has empowered so many members of the modern media. That was one thing I didn't mention last week. Obviously it empowered my own career, but like my media diet is probably 75% Stripe.
No, for sure. Customers at this point.
This is the obvious angle for us to talk about it because it's the direct personal thing.
All the listeners and readers, this is why they should be grateful, right? Like not just us, not the greatest of all talk, but all your favorite podcasts, all the Substacks, all this sort of stuff. You have to be able to get paid, right? And that was not easy to do before.
Well, here we are. All right, back to the show. Couple Anthropic emails. Alexander says, 'Andrew and Ben, I watched Dario and Daniela's interview this week with Walter Isaacson and woof.' They start off the interview by praising the inherent honesty and goodness of Anthropic and then spend the majority of the 42-minute discussion criticizing OpenAI for the following: having ads in their product, building more sycophantic models, partnering with the government as quote an opportunistic move rather than a quote principled stand, and being generally dishonest. The level of obsession, hatred, and disdain they have for OpenAI is alarming. It's like someone badmouthing their ex during their wedding vows. Someone should remind Dario and Daniela that when you point one finger at someone else, there are three fingers pointing back at you. Uh, not sure how that works. Had not heard that phrase before, but...
Are you serious? I mean, I'm pointing a finger at you. There's no... Oh, there we go.
People will have to watch the video for that one. But sure enough, there were three fingers pointing back at me.
I am baffled how there's a question here about the Gen X millennial thing. Is this a generational divide? How have you not heard this before?
It could be. We're only divided by like six years, but maybe that's enough of a chasm to create that disconnect. In any event, what do you think of Alexander's sentiments there?
It was funny you included this email because I was literally thinking the exact same thing. This is the hazard of recording podcasts. Sometimes you're just thinking out loud on air for everybody. But there you go. Learn something new every day. All right. So, you were thinking the exact same thing that like the... it's kind of, it's pathetic in a way. Okay, the extent to which they do continually define themselves as being the opposite of OpenAI. And it's kind of funny because they're like, oh, about alignment and training our model to be great. And if they continually are making decisions like, oh, we're going to not do what OpenAI does, who's actually setting the agenda here, right? If OpenAI is like too online, Anthropic is... yes, like anchored on what they're doing, right? Sure. And you know, to what extent might OpenAI actually do something that makes sense, that is the right thing to do, and Anthropic is like anchored on, well, guess that's the wrong thing to do. I do think it is a very astute point. It speaks to overall just the discomfort I have with their entire view of the world. Okay. Which is like, this stuff is too powerful, so someone has to teach it how to be good. And the only ones that can be trusted to do that are us. And you always ask why. And Alexander pointed out what their answer is: because we're not Sam Altman. It's like, I'm like, dude, if your definition of being good is not being Sam Altman...
Well, and maybe I'm too immersed in this world at this point because to me...
Well, you're still defined by Sam Altman. It's like people who define themselves as being against Donald Trump. You have let Donald Trump define the entire frame in which you're operating. But it's almost just ambient noise to me when I hear them go off on one of these tangents about how untrustworthy OpenAI is or how untrustworthy Sam Altman is. I mean, there's an element of this where it's more charming than like some of their other crazy digressions. Their obsession and their rivalry with OpenAI, not really that big of a deal one way or the other as far as I'm concerned. I can't get too worked up about it.
Except these are the people setting themselves up as like the moral arbiters for everything.
So I don't believe that. So I've never taken them seriously.
No, I believe they're trying to do that. The answer isn't to just like, I don't think it's the right approach that they should need to teach Claude morals.
I'm not here to teach Anthropic morals. What I'm here is to make sure that there is no small collection of people that get to decide it for everyone else. Yeah. Like that's sort of the bigger... and to the extent that the mudslinging at Sam Altman is intended to delegitimize him in the eyes of the mainstream and ultimately redound to Anthropic's monopoly power over how this technology is deployed, that sucks. It's also so far been... to what extent has Anthropic actually been the best thing for Sam's like, oh my god, right? Like suddenly... and they've done wonders for Mark Zuckerberg as well. Everyone who's like worried about Anthropic is like, Sam, he's not that bad. I didn't figure it out, so we can work with this guy. Well, and when you go back to OpenAI, when Dario was there, the original beefs with Altman, at least in some cases, were... the dumbest stuff OpenAI did were from Dario Amodei and his whole team, the whole like ChatGPT-2 is too dangerous to release, right? Like this... well, and part of their falling out involved a clause that would have required OpenAI to merge with another company that was closer to AGI than they were, and Sam wouldn't include that clause, or Microsoft wouldn't agree to including that clause in the investments, and Dario was very upset at like, of course Microsoft wouldn't agree to it, and of course Sam wouldn't like have that be a dealbreaker for the... seriously, this does go back though to OpenAI's original sin, which is this whole we're going to be a nonprofit. Like this actually, companies should operate like normal companies, traditional incentives. And this is the whole problem with the whole S-1 and Anthropic's insane governance structures we haven't talked about either. There's like seven share classes, like there's this magic share that like the board has or the benefit corporation has or what are these random things, has about who gets voting or who doesn't. It's nuts. Like actually we should just operate like normal companies, that would be... if you're buying in, that's what you're buying. Yeah. Well, we'll see. Profit motives, good, healthy, good for society, healthy. Yes.
You know, I was writing about China for Stratechery this week, and they have all these political incentives that kind of distort things, obviously, famously. And America, as crazy as America can become in the eyes of the world, the rest of the world can trust that there are basic profit and market incentives guiding the decisions that are made. Incredibly important, like it is its own rule of law.
And it gives you... why is the rule of law important? You can make decisions. You can make investments within a predictable framework. That is what the traditional corporation, profit-maximizing corporation... Yes, there are obviously issues. They're well documented. But every system has issues. And what the benefit that has is rational acting. You can actually engage, understand someone else. This seven-share weird governance, we're not actually governing for profit, is worse. And it's... I mean, you know what you're signing up for, to be fair, and you're signing up for like, we're going to just dilute you away and we're not focused on making a profit. And maybe they'll pull it off because Opus 5.5... because at the end of it, that's what we always come back to, it's like the core product really, really good.
All right. Robbie says, 'I was just listening to Sharp Tech at 3x speed while on the treadmill. I usually listen to podcasts at 3x, but for Ben's updates or interviews where there is also a transcript, I listen and read, which is even better.' Yeah. See, Ravi, you should have filled us in. There we have it. I don't watch video versions because I don't have YouTube Premium and can only go to 2x.
Ravi, does he need YouTube Premium? That's a hilarious reason to not watch video podcasts, is because it only goes to 2x. I mean, look, you can live the life the way you want to. I'd rather you pay your subscription money to Stratechery versus YouTube if there's only one.
He's making good choices as far as I'm concerned.
Yeah. You know what? I have nothing to say. Continue with the email.
Great interview with Craig Moffett. He says, 'Where does Ben land on his key question about AI capex?' This is Moffett's key question: is end demand from consumers and enterprise where it needs to be, and will that be sufficient in the near term, 2027, 2028? And the direct quote from Craig Moffett: 'All I care about,' he said, 'is are the revenues coming into the system, which means coming into Anthropic, coming into ChatGPT, are those greater than the cost of building the whole value chain? Ignore all the revenues going on between the layers of the value chain, that's just noise. All that really matters is enough money coming into the system to support the system.' Where do you stand on that question? I thought it was a pretty good formulation of like the core issue from Craig Moffett.
Right. It's easy to kind of like how much is Nvidia taking, how much is TSMC taking, what's up with the Neo clouds. Yeah.
No, it's a very good observation, and I don't know — like, we don't know what the timing is right now. As impressive Anthropic's revenues are, they don't come anywhere close to all the capex that's being put out.
Yeah, but the capex is also for two to three to four to five years down the road. If Anthropic keeps accelerating the way they are — then like, accelerating based on what time frame? With the beginning of the year, later on — this all that matters is their revenue growth.
This is the other point. This goes back to like, people get really worked up about stock-based compensation because that is also dilution, and they're like, 'You don't — you should be evaluating this in your stock picks.' The reason people don't care about stock-based compensation: they only care if your revenue is not growing. If yours is growing, no one cares. It's fine. Once yours is growing, they're like, 'Yeah, well, and also, have you seen their stock-based compensation?' Right. Like this. And so all this stuff I talk about with Anthropic — the dilution, the weird governance — no one cares if it's up and to the right.
Yeah, and that matters broadly for the whole ecosystem. The thing to come back to, though, is that Amazon deal at the end of the day, and the question that we got about going from free cash flows to debt to equity.
The entities with the most capacity in all three areas: who has the most cash flow? Who has the greatest capacity to issue debt because of their large cash flow and diversified businesses, which gives them basically — like, not collateral, but the other sources of income?
And who has the greatest capacity to issue stock? Hyperscalers.
Google, AWS — or Google, Amazon, and Microsoft. Yeah, there is an extent to — if we cross the gap, they're like the girders of the bridge that get us across. Okay.
And it's going to be very interesting to see how they play their hand over this little bit. Like, to what extent does it end up — so Sundar Satya holding up the bridge? Well, I mean, are we going to end up in a world where Anthropic is all over here doing all their crazy stuff, and what actually happens is just AWS and Google just like basically own them to get them over the canyon? Okay.
And of course we could reach a limit. The limit where everything goes sideways is once the hyperscalers can't do it anymore.
But there they are. What would that mean? They have the cash flow of last resort. They have the debt-raising capabilities. I mean, all even all the neoclouds, all this sort of stuff that's out there — the hyperscalers, their cost of capital is so much lower than a hyperscaler trying to raise — like, Nvidia's out there.
Or yeah, the neocloud. Yeah. Why is Nvidia backing up the neoclouds? This is one of the concepts. This is when I really wrote about the credit crunch or the credit questions and the railroads — was Nvidia backstopping all these neoclouds to help.
Sees this happening too, right? Because the worst-case scenario is the world runs out of money before the revenues catch up.
And that's like what happened with the railroads and the massive crash. Basically, the world just ran out of money.
The rails didn't mean the rails weren't useful, right? Right. The rails are still used today, but there was a massive financial crash. If we — the way we don't get there is we just have to bridge the gap to where the revenues are enough to pay this all back.
And it's — the challenge for Nvidia is if we get there, it's you probably because the hyperscalers. Okay. And that means the hyperscalers who have gained tremendous power over the entire ecosystem, because that's — you get what you pay for, right? And in that world, those are the ones who are going to build competitive chips, going to provide on prices, all these sorts of things.
When we say tremendous power, is that because Anthropic and others will have to continue trading equity in exchange for compute going forward?
Right, yeah. And you know, and the terms — those terms Amazon got with Anthropic was when Anthropic needed it the most, because they were trying to get started and no one believed in them. They needed a ton of compute, and so Amazon got an unbelievable deal, right?
You can fast forward — the deals today maybe aren't going to be quite as good, but if Anthropic is like, 'We can't operate, we can't operate' — no one — their equity, like if their stock price starts plummeting. Yeah. The neoclouds — like they have to turn: where can we get money to be competitive? Amazon's like, 'I got some money for you.'
Remember that deal we made 2023, 2024? How about an even worse one, right? Like, I can't wait to see what happens over the next couple years here, you know.
Well, and I think Moffett's core question is like, is the money for any of this here? And like, because he doesn't doubt the technology, but it's like, how much are people — how much are enterprises willing to pay, and can that sustain any of this over the next couple years? Which I think is something of an open question. I mean, the — yes.
Well, if you want to really zoom out, I think the biggest broader question — one of the really — I really enjoyed this, the email with Sham, not Shams, this week. And one point he made up front is from the Pal — you know, he's talking his book, but of course, but I mean Palantir — he admitted in there, because I kind of pushed him about that, like cult recognized cult. Like part of the reason they're worried about Anthropic is like, we're the right religion, they're the wrong religion. That's right. But is what they see with companies is companies who are focused on using AI to improve the bottom line struggle, because how do you improve the bottom line? Replacing humans. Yeah. How do you improve the top line? Multiplying humans.
Right? There's this difference between multiplying versus replacing. And when you're multiplying humans, you're basically creating entirely new things to do, new workflows, new opportunities. The challenge for this whole trade in general is that usually number one takes longer, and number two usually comes from new companies that come in, they disrupt the incumbents by virtue of the AI-first company taking on the old-school SaaS company. The SaaS company's trying to shove AI in and use as much as they can, but they're just fundamentally not constructed properly to do it. And so in the fullness of time, those companies go away. The AI-native companies take their place, and that's how AI takes over the world. I think there's a very decent chance that's how it happens.
The problem is it takes a long time.
All the money that's being spent on capex is being spent now.
Yeah, and that money needs to circle back because when you go from paying from free cash flows, paying with debt, paying with equity, the idea is it circles back around to your cash flows paying it back, right? And so the question is how long is that going to take? It's all about timing. I 100% believe in AI. Like, I think that is pretty straightforward. It's incredible.
Well, I'm in the middle ground. I don't think AI is going to replace everyone. It's going to be God. But it's just unbelievable. Like, it's... I was just showing someone this week.
It's so fun when people have like just suddenly they unlock.
Yeah. And they're like, 'Holy cow.' And then suddenly they have like, once you see what it can do, you come up with like 25 new things that it can do.
I could not be a firmer believer in this. And yet, number one, I said this years ago: the greater of an upside something is, almost the longer it takes.
And you go back to the internet, 1993, the feed was not invented till 2006.
