Why Would Anthropic Go Public Now? — Jamin Ball, Altimeter

Jamin Ball 0:00 Because the reality is, is OpenAI and Anthropic, they have far more powerful models that they've developed that they just haven't released yet. If you compare the dollar spend in compute a year ago versus today, their compute that went towards serving models has gone up by 2.3X. Their revenue over the same period of time has gone up 14X. So embedded in that is basically a 7X efficiency gain in serving inference, which, I mean, this is a business that I think is almost accidentally profitable, right? Maybe my, my favorite stat of the whole S-1 was $141 billion of capital raised to date, that's inclusive of equity, converts, and SAFEs, and a $134 billion-

swyx 0:47 Cash in bank

Jamin Ball 0:48 ... cash balance today. Like, right now, you wanna go raise multi-billion dollar debt for GPUs? No bid. No bid. They're... Like, debt markets are closed. There are, you know, big debt deals that aren't happening. [upbeat music]

swyx 1:06 Okay, we're here in the studio with Jamin Ball from Altimeter. Welcome.

Jamin Ball 1:09 Thanks for having me. Excited to be here.

swyx 1:11 Uh, Anthropic is IPO-ing. I was thinking of who do I know would be the perfect guest to cover the S-1. You're the guy.

Jamin Ball 1:17 The alpha.

swyx 1:18 You're-

Jamin Ball 1:18 Appreciate it

swyx 1:18 ... you're the guy. Uh, we met before, I think you started your Substack. I maybe was just starting mine. How old is it?

Jamin Ball 1:25 Uh, six and a half years now.

swyx 1:27 Okay. Actually, no, you're older. Uh, so Clouded Judgment is, like, one of the top... Like, what, what are the brag statistics about Clouded Judgements?

Jamin Ball 1:34 Ooh, uh, brag statistics. Um-

swyx 1:36 Like-

Jamin Ball 1:37 C- probably getting close to 100,000 subscribers-

swyx 1:40 Yeah

Jamin Ball 1:40 ... you know, every week. I, I think in six and a half years I've, I've missed one Friday.

swyx 1:44 Yeah, yeah.

Jamin Ball 1:45 Uh, try to be pretty consistent.

swyx 1:46 And the, the, the general... Like, my takeaway is, like, here's, like, public markets comms for the startup investing that you do, but Altimeter is, like, does everything, right?

Jamin Ball 1:54 Yeah. The, the idea was, "Hey, how can I provide some commentary on metrics around software companies?" And, you know, the, the universe of companies that, that typically has public metrics are public companies. Um, as a, as a way to kinda, not guide, but, uh, provide a blueprint to early stage companies and founders of like, "Hey, here are some of the metrics required to be public one day."

swyx 2:14 Yeah.

Jamin Ball 2:14 Um, and so there's lots of tables and charts and graphs on public metrics and valuation multiples. And maybe three, four years ago I, I kinda started writing a little ramble up front, which is more of a, I kinda treat it as a personal journal. Like, what's on my mind, what, what have I been thinking about, what's, what, what, what's top of mind. Um, and, you know, that'll range from anything from, you know, some macro commentary to, uh, market commentary to, you know, trends in AI or what- you know, what- whatever's trending at the moment.

swyx 2:43 Yeah. I think every generation there's kind of a VC blogger who is, like, sort of the guy to cover what's going on. It used to be Fred Wilson-

Jamin Ball 2:51 Yeah

swyx 2:52 ... if you're back in the day. Uh, he used to write AVC, I think.

Jamin Ball 2:54 Yeah.

swyx 2:55 And then, uh, more recent- like Tomas Tunguz.

Jamin Ball 2:57 Yeah.

swyx 2:57 I mean, he's still active, but like-

Jamin Ball 2:59 Yeah

swyx 2:59 ... uh, I feel like you are the king of, like, charts and public stocks right now.

Jamin Ball 3:03 No, I appreciate it. I mean, Tomas was a huge inspiration.

swyx 3:05 Right?

Jamin Ball 3:05 So I, I was at Redpoint for, for quite some time, uh-

swyx 3:07 Yeah

Jamin Ball 3:07 ... with Tomas and, you know, I, I-

swyx 3:09 And now, now you started Theory.

Jamin Ball 3:11 Exactly.

swyx 3:11 And, yeah, so.

Jamin Ball 3:12 Exactly. But I give-

swyx 3:13 Is Redpoint okay? You guys are all leaving. Like

Jamin Ball 3:14 No, Redpoint's great.

swyx 3:15 [laughs]

Jamin Ball 3:15 They're, they're, they're great. Uh, I mean, I owe everything I learned to, to those guys, and I, I have a ton of respect

Jamin Ball 3:20 for them.

swyx 3:20 Yeah, yeah. Amazing. Uh, okay, so we're gonna get right into Anthropic. And, uh, you have a, a, a sort of some spiel, you've had some public dust-ups with people that we won't name.

Jamin Ball 3:30 [laughs]

swyx 3:31 Um, people are just generally very confused 'cause they... A lot of them have never... A lot of our audience are engineers and researchers.

Jamin Ball 3:36 Yeah, yeah.

swyx 3:37 Uh, they've never seen an S-1.

Jamin Ball 3:38 Yeah.

swyx 3:38 They, they've never been through an IPO process, they don't know what a roadshow is-

Jamin Ball 3:42 Mm-hmm

swyx 3:42 ... they don't know how this is done.

Jamin Ball 3:43 Mm-hmm.

swyx 3:44 But more importantly than that, just the me- that's the mechanics.

Jamin Ball 3:46 Yeah.

swyx 3:46 More importantly than that, what does it mean about our general state of AI, right?

Jamin Ball 3:50 Yeah.

swyx 3:51 Like, this is the most visibility we've ever gotten into Anthropic.

Jamin Ball 3:53 Yes, yes.

swyx 3:54 And like, well, now we know a lot more things than we used to.

Jamin Ball 3:57 Yeah, totally. Totally. And just, just to say, uh, say it upfront, um, 'cause I've been asked to say this, so these are... I work for Altimeter, we have a big position in Anthropic, everything I say is, you know, these are all my personal views versus, versus, uh, versus my firm views. Uh, but yeah, it's a big deal. I think, you know, the big labs, OpenAI and Anthropic, seems like every quarter their numbers leak. Uh, but it's a subset of the numbers, it's not, it's not everything. And now we kinda get a full open kimono view of the financial profile of, you know, one of the most important and iconic companies maybe of our lifetime. And just

Jamin Ball 4:28 kind of a, a quick process, uh, process refresher, they'll file an S-1, uh, probably wait about 15 days, there's usually a minimum, until they'll start a roadshow. The roadshow's when they actually start and they pitch investors, IPO investors that, you know, travel across the country, across the world to pitch the IPO. That'll usually last a week and a half, and then it will start trading publicly. So we're probably looking at-

swyx 4:50 End November

Jamin Ball 4:51 ... yeah, like mid-end of November. Um, you know, kind of 15th to 20th, 10th to 20th of November of, of when it actually starts trading publicly.

swyx 5:00 Yeah. Do people IPO past Thanksgiving? That's like a weird dead zone to-

Jamin Ball 5:05 Yeah, it's... You know, between Thanksgiving and Christmas is, is kind of a dead zone.

swyx 5:09 Yeah.

Jamin Ball 5:09 Um, I would say, you know, post-Labor Day to Thanksgiving is generally a, kind of like a hotbed-

swyx 5:15 Yeah

Jamin Ball 5:15 ... uh, time for companies to go public. People are around, people are engaged, um, not on-

swyx 5:19 And, and it's Anthropic

Jamin Ball 5:20 ... not on vacation yet.

swyx 5:21 Like people-

Jamin Ball 5:21 And it's Anthropic. The reality is, is I think-

swyx 5:23 This is- [laughs]

Jamin Ball 5:23 ... people would, people would make time for-

swyx 5:24 I'll go to work for Anthropic. Yeah.

Jamin Ball 5:25 Yeah, exactly. Exactly.

swyx 5:27 Uh, yeah, so okay, that, that's the normal mechanics. I, I think, um, and, and that's it, right? So, so off the top of your head, uh, you know, people should understand that an S-1 is just a, a filing that you have to do. Um, OpenAI has done theirs as well, confidentially-

Jamin Ball 5:41 Yeah

swyx 5:41 ... and then, and then they go public.

Jamin Ball 5:43 Yep.

swyx 5:43 Um, people have filed S-1s and not IPO'd.

Jamin Ball 5:46 That's right.

swyx 5:46 Cerebras famously did that.

Jamin Ball 5:47 That's right.

swyx 5:48 Uh, and then they did it again. Um, and, uh, it, it's just part of the routine process, and you have investor day and you have whisper rumors-

Jamin Ball 5:54 Mm

swyx 5:54 ... and all these things, right?

Jamin Ball 5:54 Right.

swyx 5:54 So- Uh, as someone who is-- unless you're literally a banker or, or fund manager, that's all you need to know basically.

Jamin Ball 6:01 Yeah.

swyx 6:01 Right?

Jamin Ball 6:01 Yeah.

swyx 6:01 There's no, there's no-- This is part of the process, um, and there's a lot of indication to try to make it a successful IPO.

Jamin Ball 6:07 Yeah.

swyx 6:07 Successful is a mixed or loaded word this, these days because, um, typically you'll hear, hear things like, "IPO was so successful, the stock popped 50%-"

Jamin Ball 6:16 Mm-hmm.

swyx 6:16 "... on the first day."

Jamin Ball 6:17 Mm-hmm.

swyx 6:17 That's actually potentially a failure if you're, um, Bill Gurley.

Jamin Ball 6:21 Yeah, exactly.

swyx 6:22 [laughs]

Jamin Ball 6:22 Exactly.

swyx 6:23 Who is like, "Well, you could have got 50% more."

Jamin Ball 6:25 Yeah, exactly.

swyx 6:26 [laughs]

Jamin Ball 6:26 And you know, and just, just for the audience, what, what that mechanism is, is in an IPO, it's like a fundraise on steroids. You're gonna go out and you're gonna raise a bunch of money at a certain price.

swyx 6:34 How much are they raising?

Jamin Ball 6:35 Uh, I don't think it's-

swyx 6:35 Like 20 billion?

Jamin Ball 6:36 I don't think that's been-

swyx 6:37 Okay.

Jamin Ball 6:37 I don't think they've decided yet.

swyx 6:38 I hear a rumor that-

Jamin Ball 6:39 But I think the rumor's like 75 to 100 billion.

swyx 6:41 Fuck.

Jamin Ball 6:41 Um, it'll be a big raise. I mean, typically, uh, what you used to see-

swyx 6:45 Jesus Christ.

Jamin Ball 6:45 [laughs] I know. What you used to see in normal IPOs was companies would float about 10 to 15% of their company, which means, like, what you would sell into the IPO would be 10 to 15%.

swyx 6:55 Yes, and there's an employee lockup and blah, blah, blah.

Jamin Ball 6:56 Ex- ex- exactly. Uh, now it's, it's maybe as, as the companies get bigger and bigger and bigger, uh, they might float less of the company because, hey, you can take less dilution. Uh, you know, it's no different than a company raising a series A, B, or C. It's you're gonna raise money at a certain dilution. Um, and yeah, the Bill Gurley thing is, well, if the stock pops 50% on day one, why wouldn't you have just raised money-

swyx 7:17 Priced it higher

Jamin Ball 7:17 ... at a higher price?

swyx 7:18 Higher. Yeah, yeah.

Jamin Ball 7:18 Or, or, you know, taken on, taken on less dilution.

swyx 7:20 Yeah.

Jamin Ball 7:20 And that's maybe-

swyx 7:22 And we don't have to get into that

Jamin Ball 7:22 ... a whole separate topic.

swyx 7:23 Yeah, yeah. Yeah.

Jamin Ball 7:23 You got lots of nuance-

swyx 7:25 There's-

Jamin Ball 7:25 ... to public markets and IPO allocations-

swyx 7:26 There's tradition

Jamin Ball 7:27 ... and all that.

swyx 7:27 There's conflicting incentives-

Jamin Ball 7:29 Yes

swyx 7:29 ... and then there's the, there's the chance that you actually, you don't pop.

Jamin Ball 7:31 Yes.

swyx 7:31 Um, so all that, let's put it at a high level. Why now?

Jamin Ball 7:35 Yeah.

swyx 7:35 Why IPO now, right?

Jamin Ball 7:37 Mm-hmm.

swyx 7:37 They c- uh, they could do a Stripe, they could do a Databricks.

Jamin Ball 7:39 Yeah.

swyx 7:40 Right?

Jamin Ball 7:40 Yeah.

swyx 7:40 They're not. They're IPO-ing.

Jamin Ball 7:42 Yeah.

swyx 7:42 Why?

Jamin Ball 7:42 Yeah. I think, I mean, I think there's a couple reasons. Well, one, it's all speculation. It's hard to know. But I think there's a number of reasons why companies have historically choosing to go public. Um, there's kind of the obvious, let's provide liquidity to our employees and our investors in, like, an unrestricted way. Um, 'cause of course, you know, now big companies, Databricks, Stripe, OpenAI, Anthropic, kind of like tenders are, are pretty regular, uh, thing-

swyx 8:05 Like once a year-

Jamin Ball 8:06 ... that they all do

swyx 8:06 ... you'll, you'll get like... Well, if you wanna cash out, here's, here's the-

Jamin Ball 8:08 Yeah

swyx 8:08 ... you know, price.

Jamin Ball 8:08 But you know, you'll have a-- You'll be limited in how much you can sell. Um, you know, now once a company's gone public, once they've gone through the lockup, you know, which is anywhere from, you know... So it used to be his- just 90 day, uh, 180 days, six months, and now it's kinda like every company has a bespoke lockup that's, like, dribbled out. Some are shorter, some are longer. Uh, but it's kinda like unrestricted access, and you can generate liquidity for your employees, um, help morale, whatever. Or, um, you need a liquid currency for acquisitions.

swyx 8:36 Yes.

Jamin Ball 8:36 And hey, a company might be more willing to take an acquisition from a public company because the currency you're getting back is, is liquid, versus in a private company, maybe it isn't.

swyx 8:46 It's made up.

Jamin Ball 8:46 You don't really know the value.

swyx 8:47 Yeah.

Jamin Ball 8:47 Yeah, it's Monopoly money.

swyx 8:48 Yeah.

Jamin Ball 8:48 Um, I would say these quasi-public companies, you know, OpenAI, Anthropic, Databricks, I think their currency has almost started to be viewed as cash, not Monopoly money, because they're very high quality, they're very late stage. Um, that effect probably means less. I would say a, a bigger reason to go public, um, because it- and, and it used to be companies went public because now you could access much larger quantums of capital. You know, you- if you wanted to raise $100 million, you know, it used to be the only way to raise $100 million is to, to, to go public. You know, now you, you squeeze-

swyx 9:21 Huge

Jamin Ball 9:21 ... and raise a-

swyx 9:21 Very huge

Jamin Ball 9:21 ... series A, or a seed. A seed of, of more than that. Um, and so it's not necessarily the quantum of capital. I think whether it's US institutions getting bigger, um, maybe folks who used to act as LPs, large pension funds and endowments, acting as direct investors in a company themselves, or kind of the emergence of big sovereign international funds investing direct in companies. There's a huge capital base available to private companies. H- And, and so, like, I don't know if the argument necessarily holds, hey, you, if you wanna raise a lot of money, um, go public. I think OpenAI raised a round of, like, 120, 130 billion, you know, some ridiculous amount.

swyx 9:58 Biggest of all time.

Jamin Ball 9:59 Yeah, ex- exactly.

swyx 10:00 Yeah.

Jamin Ball 10:00 Um, I think maybe a more important

Jamin Ball 10:02 angle is, is more of, like, the debt piece. Um, and so once you go-

swyx 10:06 Let's talk about it

Jamin Ball 10:07 ... once you go public... Um, actually, let me take a step back. If you want to raise debt, um, or if you want someone else to raise debt, not on your behalf, but, like, as a neo Cloud, a neo Cloud might wanna raise debt to build a data center for Anthropic to, to use. Debt providers are gonna look at Anthropic, and they're gonna underwrite Anthropic's credit worthiness. And how it works is they'll go, they'll say, "Hey, we're gonna syndicate however many billions of dollars of debt for you, Anthropic, or via a neo Cloud for you. But we're gonna need, we're gonna need, like, full open kimono on your financials. We're gonna need to know exactly how this business is doing, uh, so that we can underwrite you. And then everyone we syndicate the debt to on the back end, they also need to see it." And I'd say historically, companies have just been hesitant to go full open kimono with their financials to the debt and credit markets because they don't want them to be fully public. Um, but if you want to be rated by the, by the rating agencies, who basically, they'll give you a credit rating. Um, investment grade is the highest. Your credit rating will kind of impact the rate with which you can raise debt at. And the higher your credit rating, the lower the rate that you'll receive. But you, you know, have to be open kimono with your financials. So going public, um, is a great way for companies to basically, you're gonna get rated. I think SpaceX very famously got rated very quickly after going public. Now you can-

swyx 11:29 Too quickly?

Jamin Ball 11:29 ... tap into... Yeah, arguably too-

swyx 11:31 [laughs]

Jamin Ball 11:31 I mean, there, there was a lot of people who said it was, "Hey, it was too fast. They need to season more as a public company." Um-

swyx 11:36 And, and added to the S&P 500 too. Something like that?

Jamin Ball 11:39 Yeah.

swyx 11:39 Right, yeah.

Jamin Ball 11:39 Yeah, new rules. Exactly.

swyx 11:41 Yeah.

Jamin Ball 11:41 Uh, and so I, I think, hey, a lot of these companies, they will be capital intensive, whether it's for training or building out compute clusters, and you wanna tap into the credit markets. And I think if you wanna tap into the credit and debt markets, you can do that much more efficiently- As a public company.

swyx 11:57 Sure.

Jamin Ball 11:57 And, and so again, it, it's probably a confluence of all those things adding up to, hey, why, why go public? I think there's also just if you ask, if you ask these executive teams and management teams, hey, they want the everyday American to be able to participate in the equity upside of these companies.

swyx 12:12 I was gonna bring that up.

Jamin Ball 12:13 Yeah.

swyx 12:13 Yeah, up to including giving shares to everyone.

Jamin Ball 12:15 Totally.

swyx 12:16 Every American.

Jamin Ball 12:16 Totally. And hey, the reality is, is private markets are more of like a cabal. It's a small subset of the population, um, either qualified purchasers or accredited investors or, or, you know, big funds and pensions and endowments get to participate in the equity appreciation of these businesses. But, you know, your doctors and dentists and your everyday American doesn't, and there is just something that doesn't, doesn't feel right about that, right? Like, everyone's-

swyx 12:39 Oh, I mean, straight up you'll get social unrest, right?

Jamin Ball 12:41 Yeah, exactly.

swyx 12:41 And straight up you'll get, you get voted regulation because, like, well, those guys have their AI, and I'm gonna vote against it. I have my vote.

Jamin Ball 12:48 That's right. That's right. That's right.

swyx 12:49 Um, and so, like, th- that is like a, a really an interesting governing mechanism of democracy versus capitalism-

Jamin Ball 12:55 Yeah

swyx 12:55 ... where, like, capitalism encourages centralization-

Jamin Ball 12:57 Yeah

swyx 12:57 ... and democracy encourages, um, decentralization somewhat.

Jamin Ball 13:00 Mm.

swyx 13:01 Um, anyway, so th- there's all that. I think it's very, very valid. Uh, I'll point out for people who don't know about capital structure that, uh, equity is, is junior to debt.

Jamin Ball 13:08 Yeah.

swyx 13:09 And so-

Jamin Ball 13:09 Yeah

swyx 13:09 ... it's nice to have $2 trillion in, in junior [laughs] -

Jamin Ball 13:11 Yeah, exactly

swyx 13:12 ... cash sitting

Jamin Ball 13:12 Exactly

swyx 13:13 ... in your enterprise value, right?

Jamin Ball 13:14 Exactly. Exactly.

swyx 13:15 Uh, and that's like, well, that, that's, that's, uh, that's public. Um, and I think part of that also, I'll, I'll throw in the why now, which is a little bit more advanced than-

Jamin Ball 13:22 Yeah

swyx 13:22 ... than people, uh, know, is, um, interest rates are spiking-

Jamin Ball 13:26 Mm-hmm

swyx 13:26 ... for, for, um, uh, for data center debts w- uh, globally, right?

Jamin Ball 13:31 Mm.

swyx 13:31 Like, is there a source to look that kind of stuff up? Is it just Bloomberg or is...

Jamin Ball 13:35 Hmm, yeah, I, I mean, I think most of these-

swyx 13:36 I mean, there's just charts that people tweet around, but...

Jamin Ball 13:39 Most of these big debt deals will price, um, and there'll be, you know, like an interest rate, an OID. You know, there, there'll be a bunch of things associated, like, with the price, and like oftentimes those will be public once they've priced. Um, but yeah, rates are going up. Um, risk-free rates are going up. The Fed paused, yeah, decided not to hike yesterday or earlier this week. Um, I don't know if that's necessarily a reason to go public, um, but-

swyx 14:07 I- it's part of the why now.

Jamin Ball 14:08 Yeah.

swyx 14:09 If, if it does help with that, that if my debt went from like, I don't know, uh, 3% to 5%, that is billions of dollars.

Jamin Ball 14:17 Yeah, yeah, yeah.

swyx 14:17 Um, and if I can keep a lid on that, that's very important to me.

Jamin Ball 14:21 Yeah. I mean, a lot, a lot of the data center debt deals now are, are getting done over 10%. Um-

swyx 14:26 Yeah

Jamin Ball 14:26 ... and so it's, it's getting, it's getting expensive. [laughs]

swyx 14:28 Okay.

Jamin Ball 14:28 Yeah, it's getting expensive.

swyx 14:29 Yeah, and so, so and then the other, that's the other part of this discussion, which is, um, uh, the ROI on a data center.

Jamin Ball 14:36 Mm-hmm. Mm-hmm.

swyx 14:37 Right? Because the- there, there are a lot of AI skeptics-

Jamin Ball 14:40 Yeah

swyx 14:40 ... who are like, "This is the, the biggest build-out in human history-

Jamin Ball 14:43 Yeah, yeah

swyx 14:44 ... and we are throwing money after this illusory dream of AI."

Jamin Ball 14:47 Yes.

swyx 14:47 And then on the other side, the AI pro people are saying, "Well, your payback period is one year."

Jamin Ball 14:52 Yeah, yeah. Totally.

swyx 14:53 Why is there such a huge difference?

swyx 14:55 [laughs]

Jamin Ball 14:55 It's... Thi- this was kind of the best thing about the Anthropic S-1-

swyx 14:58 [laughs]

Jamin Ball 14:58 ... which is a lot of the metrics that were leaking were your, your ARR, your this or your that, but you didn't really get to see a lot of below-the-line metrics.

swyx 15:04 That's not ROI, yeah.

Jamin Ball 15:05 Um, you didn't get to see the cash.

swyx 15:06 Yeah.

Jamin Ball 15:06 How much are they burning?

swyx 15:07 It's not profit.

Jamin Ball 15:08 Um-

swyx 15:08 Yeah

Jamin Ball 15:08 ... what's, are they profitable? Are they not? And, like, that I think was one of the biggest surprises of the S-1, right? Here's a business that showed 85% gross margins, over 20% free cash flow margins, um, 8% GAAP operating margins. And so even on a GAAP basis-

swyx 15:24 That's new, yes

Jamin Ball 15:24 ... um, highly profitable. And maybe my, my favorite stat of the whole S-1 was $141 billion of capital raised to date. That's inclusive of equity, converts, and SAFEs, and a $134 billion cash balance today.

swyx 15:42 Cash in bank.

Jamin Ball 15:42 And so-

swyx 15:43 Cash and cash equivalents, but yes.

Jamin Ball 15:44 E- exactly. And so you think, okay, well, you've, you've raised this much. You have this much in the bank. That means you've burned $7 billion to get to, you know, the Q3 numbers that aren't in the S-1 but that were kind of, like, broadly reported a few weeks ago, $16.8 billion of quarterly revenue off of $7 billion of lifetime burn.

swyx 16:05 Hmm.

Jamin Ball 16:06 Which, and the burn we should, we should get into 'cause, like, that was a, a huge debate on, like, well, have, well, it looks like they did raise debt, or are they really profitable? What is this net loss on the P&L? There's, I'd say lots of-

swyx 16:17 Yeah

Jamin Ball 16:17 ... uh-

swyx 16:18 So yeah

Jamin Ball 16:18 ... lots of things out in the world that aren't fully true.

swyx 16:21 That is important. I do wanna cover it. Let's, that, that, there's a little bit more in the weeds than-

Jamin Ball 16:24 Yes

swyx 16:24 ... than, than at the start of this conversation.

Jamin Ball 16:26 Yes.

swyx 16:26 But, like, that overall efficiency is by far, uh, if you've been following the relative fates of, let's call it SpaceX, uh, Anthropic, OpenAI-

Jamin Ball 16:36 Mm-hmm

swyx 16:37 ... uh, the, the ones that are going public, um, they are the most efficient lab in the world-

Jamin Ball 16:41 Yeah

swyx 16:42 ... by, by, like, order of magnitude.

Jamin Ball 16:43 Yeah, totally.

swyx 16:44 Um, and, uh, you know, so I, I, when, and when I wrote up the S-1, I was like, okay, well, this is the, the, the, the mystery.

Jamin Ball 16:51 Yeah.

swyx 16:51 Like, why?

Jamin Ball 16:52 Yeah.

swyx 16:52 What's going on? Like-

Jamin Ball 16:53 Yeah

swyx 16:53 ... how are they so much more efficient?

Jamin Ball 16:55 Yeah.

swyx 16:55 And, uh, I a- they, they, in their S-1, they attributed it to the loops.

Jamin Ball 17:00 Yep. Yeah.

swyx 17:00 Um, the improvement loop, the investment loop, something like that.

Jamin Ball 17:02 Yeah, exactly.

swyx 17:03 Um, so people can go read that up. I, I already wrote it up. Um, but, like, uh, you know, maybe they're just smarter.

Jamin Ball 17:08 Yeah.

swyx 17:08 [laughs]

Jamin Ball 17:09 Well, I th- I think here, like, one, one of the best numbers to look at is, like, what is their growth

Jamin Ball 17:14 relative to their, what is their, the, the change in growth relative to the change in serving compute spend.

swyx 17:19 Hmm.

Jamin Ball 17:20 And they do, they break out their compute spend into more, like, training and, um, internal use versus-

swyx 17:25 And by the way, if you have charts to-

Jamin Ball 17:27 Yeah, let's pull it. We do have, I do have a chart on that. Um, so here is, here is a chart that shows the-

swyx 17:34 Let's, let's do full screen on this one.

Jamin Ball 17:35 Oh, yeah, yeah, yeah.

swyx 17:37 Just, just so... Yeah, it's-

Jamin Ball 17:38 Perfect

swyx 17:38 ... it's a little small.

Jamin Ball 17:39 Yeah, exactly. All right, here's a chart, and this is broken down in their S-1, that shows the percent of their compute that goes towards basically serving models, so inference revenue generating, or internal R&D kind of training or maybe safety and alignment or, you know, non-serving of, uh, of, of tokens. And what's interesting is you can see now that, I mean, they're at- 27%, only 27% of their compute today is going towards serving models. Um, here's why this is so interesting. If you compare the dollar spend in compute a year ago versus today, their compute that went towards serving models has gone up by 2.3X. Their revenue over the same period of time has gone up 14X.

swyx 18:21 [laughs]

Jamin Ball 18:22 So embedded in that is basically a 7X efficiency gain in serving inference, which, I mean, this is a business that I think is almost accidentally profitable, right? Where, I mean, a 7X increase in efficiency of serving tokens, if you were, if you were monetizing, you know, a megawatt at 10, $15 million, now you're monetizing it at 70 to $100 million. And here's... This is a little bit of like a, a, a random rant. Um, but software companies have historically told the world and told investors, "Hey, we're an annuity of cash flows. Once we acquire a customer, our net retention's high, and they, they'll, they'll stick with us."

swyx 19:00 Customer lifetime value.

Jamin Ball 19:01 Cu- ex- exactly. "And we're gonna be burning a lot of money early on, um, but we're just in growth mode. We're in customer acquisition mode. Once we c- exit growth mode, we'll be more in harvest mode, and that's when we'll start to become, we'll flip profitable, um, 'cause we're not spending a lot of money on, on, um, on customer acquisition costs." But, you know, the, the general model of a SaaS company was unprofitable over time, then profitable. What's happened with Anthropic, and let me pull up another chart here. Um, I believe I have their... Ah, here are their free cash flow margins. Um-

swyx 19:33 Oh my God, it's so negative [laughs] .

Jamin Ball 19:34 Which I had to cut the chart, right? [laughs] Like, like negative 4000%, negative 1500%, you know, super, super negative. If, if I actually drew the axes to scale, it, it would, it wouldn't look real. Um, and so yes, the, the axis is broken. Um, but this is the free cash flow margin of Anthropic, right? 21% free cash flow margin in their most recent quarter. And so what you're seeing is a business that all of a sudden, because they were able to drive such efficiency in their compute, that's kind of when- why I went on that, like, compute ramble, um, they've been able to monetize at much higher rates and actually deliver positive free cash flow margins pretty early on. And so this is a business that says, "Hey, we've proven that we can operate highly profitably. What we've probably done is underinvested in compute, in compute build-out, and we need to be more aggressive on it." And so my guess is that you're actually gonna see this level off and maybe go negative again as they say, "Hey, we're gonna reinvest, um, and we're gonna invest in a lot of compute. We're gonna build out a lot of compute."

swyx 20:31 Yeah.

Jamin Ball 20:31 Um, because now they've shown themselves and the world-

swyx 20:33 The capability

Jamin Ball 20:34 ..."Hey, if we need to, we can, we can flip the switch, um, and, and be profitable."

swyx 20:39 Yeah.

Jamin Ball 20:39 Um-

swyx 20:40 And for listeners, uh, a couple things on the accounting side, just because I-

Jamin Ball 20:43 Yeah

swyx 20:43 ... we gotta, like, educate people, right? Um, FCF is typically, like, the most honest version of what profit looks like.

Jamin Ball 20:50 Yeah.

swyx 20:50 Um, but, uh, in this case, you can mostly just regard it as what an operating margin would, profit margin would look, look like. 'Cause basically, FCF just focuses on cash-

Jamin Ball 20:59 Mm-hmm

swyx 20:59 ... and so there's less tricks to play with accrued receivables and lia- and, and offsetting your liabilities and what have you. Um, and so, like, that, that is the roughly-

Jamin Ball 21:08 Yeah

swyx 21:08 ... uh, rough thing. And but also this, the re- the reason people have adjusted re- revenue and adjusted profit is because of nuances in the running the business like this.

Jamin Ball 21:18 Right.

swyx 21:19 Right? Like, GAAP, which is the official generally accepted accounting principles that you have to report, um, doesn't rec- doesn't regard how you want to run the business-

Jamin Ball 21:30 Right

swyx 21:30 ... and split out investment. Uh, it doesn't have investment decisions versus, like, sort of profitability of, of core business decisions, but, um, it all gets smushed into this decisions, right? It's like, so basically, what, what I'm trying to sort of guide people towards, if you do see a change in this margin profile, it is a conscious choice.

Jamin Ball 21:47 Yeah. Yeah.

swyx 21:47 And actually, you should probably split it out, uh, further, uh, between what is actually the core business and, and, you know, is that actually infecting-

Jamin Ball 21:55 Mm-hmm

swyx 21:55 ... or is this, is this more investment, therefore FCF went down?

Jamin Ball 21:58 Yeah. And look, I mean, at, at the end of the day, cash is king. Um, and, like, your measure of profitability should be, is your cash balance going up or is your cash balance going down? Now, if you wanted to poke holes at this, um, the, the reality of GAAP accounting is there's two forms of... Well, very broadly, there's two forms of, of expense. There's the cash and the salary that you pay people, and then there's also the equity-

swyx 22:18 Stock-based

Jamin Ball 22:18 ... and then the non-cash that you pay people. So non-cash, uh, stock-based compensation, equity compensation, that doesn't show up in free cash flow. It's actually added back, uh, because it's a non-cash expense. So there are a lot of people who look at the free cash flow metric and say, "Well, it actually overstates your profitability," because the reality is, is if you didn't grant equity to your hires, they wouldn't join you. And so, like, it is a real expense. Treat it as a real expense. I don't have a chart for this, um, but if you were to adjust free cash flow-

swyx 22:47 It's still profitable

Jamin Ball 22:47 ... for stock-based comp-

swyx 22:48 [laughs]

Jamin Ball 22:48 ... you're still at an 8%-

swyx 22:50 Yeah

Jamin Ball 22:50 ... free cash flow, excluding stock-based comp. And so you're, you're even profitable if you include the big equity grants.

swyx 22:55 By normal rules, yes.

Jamin Ball 22:56 Yeah, ex- exactly. Exactly.

swyx 22:57 And so, like, if you've never seen a software IPO, this never happens.

Jamin Ball 23:00 Yeah.

swyx 23:00 [laughs]

Jamin Ball 23:00 It's very rare. I mean, I think there are, there are a lot of software companies that are free cash flow positive today.

swyx 23:06 Yeah.

Jamin Ball 23:06 But if you factor in-

swyx 23:07 Not at IPO. N- never

Jamin Ball 23:08 ... not, not at IPO

swyx 23:09 ... never at IPO.

Jamin Ball 23:09 Um, but, but even today, the mature ones, even if you factor in stock-based comp, most of them are free cash flow neutral or negative if you treat stock-based comp as a cash expense. Uh, yeah.

swyx 23:20 So all those companies are dying, and everyone's just in- buying Anthropic, right, as the conclusion?

Jamin Ball 23:23 I, I think there's a... Yeah, well, you know-

swyx 23:24 [laughs]

Jamin Ball 23:24 ... they're not quite that, um, but, but certainly-

swyx 23:27 Why would you? Like anything else

Jamin Ball 23:28 ... yeah, exactly.

swyx 23:28 Yeah.

Jamin Ball 23:28 I mean, look, here's a crazy stat. Um,

Jamin Ball 23:31 in the S1, it's like a breadth versus depth one. Um, they called out 50% of the Fortune 500 is using Anthropic.

swyx 23:39 Yeah.

Jamin Ball 23:39 However, if you look at what that cohort, the Fortune 500, is spending on Anthropic as a percent of their expenses-

swyx 23:47 Hmm

Jamin Ball 23:47 ... it's 0.1%. And so you ask yourself, where's that headed? Uh, I think John Collison on, uh, he had Sam Altman on his podcast the other day. I think he mentioned 1% of Stripe's expenses are on tokens today, but he expects that to go to 15 to 20%-

swyx 24:02 Jesus

Jamin Ball 24:02 ... over the next few years.

swyx 24:03 Jesus.

Jamin Ball 24:04 And so obviously Stripe is A very forward-leaning tech company. Your average business in the Fortune 500 isn't gonna show that level. Um, but it could s- show a same order of, or a, a, a same magnitude of growth. So if Stripe is saying we're gonna 15 to 20x the spend on tokens relative to other expenses in the org, maybe you see something similar in the Fortune 1- 500 go from .1% to 1%. Um, but I do think it's, you know, ultimately, where do you think that settles out? How deep can they penetrate their TAM? It's kind of like an important question that, you know, folks should think about. Uh, but to me at least, what it demonstrates is they really haven't, despi- I mean, despite the insane ramp in revenue, you know, growing over 1000% year over year in the Q2 quarter, there's probably still a, a lot of room left, um, for them to kinda continue to penetrate their TAM.

swyx 24:57 Yeah. Uh, and you know, I think the, the word is virtual collaborators-

Jamin Ball 25:00 Yeah

swyx 25:00 ... or, you know, I, I say agents, but-

Jamin Ball 25:01 Right

swyx 25:01 ... you know, you gotta, you gotta make it a friendly term for people. Um, and, uh, you know, I really start to tell people, uh, you should just start thinking about it in terms of, like, what your employee spend versus your-

Jamin Ball 25:11 Mm-hmm

swyx 25:11 ... token spend is.

Jamin Ball 25:12 Mm-hmm. Mm-hmm.

swyx 25:12 And most people are at, let's call it 0.1 or-

Jamin Ball 25:15 Yeah

swyx 25:15 ... or 0 point, or, or, like, 1 to 5%.

Jamin Ball 25:18 Yep.

swyx 25:18 Uh, the median AI company is at 50%.

Jamin Ball 25:21 Mm-hmm.

swyx 25:22 And the top AI companies, uh, you know, let's say 90th percentile Anthropic person is, like, 10x, uh, token-

Jamin Ball 25:28 Yeah

swyx 25:28 ... token to salaries.

Jamin Ball 25:29 Yep. Yep.

swyx 25:30 And so, uh, that is probably the right comp of, like-

Jamin Ball 25:33 Mm

swyx 25:33 ... well, what are you spending on humans?

Jamin Ball 25:34 Yeah. Right.

swyx 25:35 And, like, probably 30%, right?

Jamin Ball 25:37 Yeah.

swyx 25:37 Like-

Jamin Ball 25:37 Yeah

swyx 25:37 ... and, well, okay, so-

Jamin Ball 25:38 Could be more, even. I mean-

swyx 25:39 Yeah

Jamin Ball 25:39 ... humans could be, like, four- you know, total headcount could even be, like, 40, 50%-

swyx 25:42 Yeah, depends what kind of business you're running, right?

Jamin Ball 25:44 Yeah, exactly. Yeah, exactly.

swyx 25:44 And so, like, if you're a physical business, fine. That's, like, until we have good robots, then that's not, that's not gonna happen. But what are you spending on knowledge worker humans?

Jamin Ball 25:51 Mm-hmm. Mm-hmm.

swyx 25:51 What is... And a knowledge worker humans, that includes coders, but also just, um, admin and ops and what have you, and, and-

Jamin Ball 25:58 Yeah

swyx 25:58 ... and legal and whatever you want.

Jamin Ball 25:59 Yeah.

swyx 25:59 And all of that. And some part, like, that percentage, the, the to- the ratio of tokens spent to human spend was zero to whatever-

Jamin Ball 26:07 Yeah

swyx 26:07 ... and now it's gonna rise.

Jamin Ball 26:08 Yeah.

swyx 26:09 And-

Jamin Ball 26:09 Well-

swyx 26:09 How far can it go?

Jamin Ball 26:10 Yeah. [laughs]

swyx 26:10 Like-

Jamin Ball 26:11 Totally. Totally. It can go a lot. I mean, look, I, I think there's, there's a lot of, like, hyperbolic statements being made, both positive and negative, about OpenAI, Anthropic, the big labs. Um, I think it really boils down to three main questions. Can Anthropic stay on the frontier? Is there a business, and a durable business to be built at the frontier? And if so, how much is it gonna cost? And, and, and, and then in reality, two and three kind of, like, relate to each other. Um, can it make enough money to cover all the costs to be at the frontier? And that's, that's the one where... I think I actually have a couple... I think I have one chart-

swyx 26:48 Yeah, please

Jamin Ball 26:48 ... on this.

swyx 26:48 Please.

Jamin Ball 26:49 Let me, let me see which one. What did I want to bring up? This one. Yeah, let me pull up, um... No, let me pull up this one. This is a good one. Um, so

Jamin Ball 27:05 this is kinda the second question of, like, how much will it cost to be at the frontier? Or, or, or even maybe let's go to, like, what, what can you build a business at the frontier? The other interesting stat that Anthropic had in their S-1 was for the three months ending September, 80% of their revenue was driven by models that were six months old or younger. And so what does that tell you? The vast majority of their revenue is coming from their most recent, most powerful models, and that's something that I kind of expect to continue. I think, you know, when it comes to frontier tokens down to open weight tokens, I think there'll be a massive market for both. Uh, I think the trend will continue to be on a, as a percent of tokens, open weight kinda continues to eat share, but as a percent of wallet share in dollars, frontier continues to take the lion's share. So I expect frontier tokens to be the most economically valuable, to capture the most market share. I expect open weight and cheap tokens to capture the most market share in terms of total tokens.

swyx 28:06 Yeah.

Jamin Ball 28:06 And I think kinda this stat on 80%, which is much bigger than I would've expected, just in models six month old or less, kind of demonstrates the value of being on the frontier. Okay, so then let's go, next go to the second question, which is, well, what is it gonna... Or the third. Like, what is it gonna cost? Um, and here's a chart, I think... Oh, and there, there's... I'll, I'll have to caveat this chart. Um, but a lot of people look at the Anthropic S-1 and say, "Oh my gosh, there's $856 billion of, of-

swyx 28:34 Off the books

Jamin Ball 28:35 ... of, like, not even off, but, like, committed deals that they've signed.

swyx 28:39 Yes.

Jamin Ball 28:39 How are they gonna pay for it? You know, they're at whatever, whatever their ARR is, something in, like, I don't know, 70, 75, 80. You know, s- whatever. I think the last number was 65. Um, the, like, Q3 number was 16 point... But whatever, 70 plus or minus. Um, how are you gonna pay for 855 billion?

swyx 28:55 You know, like-

Jamin Ball 28:55 Like-

swyx 28:56 Have, have you read the rest of the S-1? No. [laughs]

Jamin Ball 28:57 Well, what I, what I... So what I tried to do, what I tried to do in this chart is I broke down the, the... 'Cause they, they broke down their commitments. Um, they didn't do it by year, but they s-

swyx 29:07 Is this normal, by the way, that you would have year by year?

Jamin Ball 29:09 No. Well-

swyx 29:09 'Cause I feel like this is unnecessary

Jamin Ball 29:11 ... so this, this, year, year by year, this is a little bit of the jam in, um, best interpretation. They'll give a dollar-

swyx 29:17 Oh, no, I've seen variations of this.

Jamin Ball 29:19 Okay. Yeah, yeah, yeah.

swyx 29:19 Yeah, yeah.

Jamin Ball 29:20 It's... And so I think what you have to ask yourself is, in the commitments that they've made, and I've tried to annualize the commitments, if they said this is a five-year deal-

swyx 29:27 Yes

Jamin Ball 29:27 ... I've just made it, you know-

swyx 29:28 It's, this is-

Jamin Ball 29:28 ... a five-year period

swyx 29:28 ... by, by the way, every banker in the world has done this already.

Jamin Ball 29:31 Yeah.

swyx 29:32 Like, they did this on, on the night of the, uh, the S-1.

Jamin Ball 29:34 Mm-hmm.

swyx 29:34 Right?

Jamin Ball 29:34 So you have to ask yourselves, can you pay for it? Okay, well, at a fixed basis, and then the, the gray bars is m- maybe kind of, like, less hard commits, but commits they wanna make.

swyx 29:44 So the-

Jamin Ball 29:44 Can you-

swyx 29:45 ... the ex- the SpaceX one is in there, right?

Jamin Ball 29:47 Yeah, that, exactly.

swyx 29:47 That's the one they can walk away from nine- in 90 days.

Jamin Ball 29:49 Ex- exactly.

swyx 29:49 Yeah.

Jamin Ball 29:50 Um, will they be able to fund 100 billion a year? And so it isn't 850 billion a year, it's can they fund 100 billion a year? Now, there's a couple ways they could do that. Well, one, do you expect their growth to continue? Um, which is a question of, like, will there continue to be a business at the frontier? Um, or you go back to the, well, only 25% of their compute today Is allocated towards, um, revenue-generating or kind of inference-serving use cases. If they were to just double that and add zero compute, they would, they could, in theory, double their revenue. And so, like, you look at the headline number of 855 billion, you're like, "Oh my gosh, that's a huge number. How are you gonna do it?" Some of that is very back-weighted. Um, but I, I, I think when I think about the trajectory of the business, um, the TAM that's been penetrated today, I don't think it's far-fetched at all to say, hey, like the business trajectory will more than cover, the revenue will more than cover the, um, the expense. And I think I have another chart in here. This is another great one. This is the compute margin. So this looks at if you were to take their revenue and subtract all compute reve- uh, uh, in- internal models training-

swyx 31:06 Same quarter matched, right?

Jamin Ball 31:07 Exactly.

swyx 31:07 Okay.

Jamin Ball 31:07 Exactly. So even as- what you're seeing is even as they've shifted more of their compute to non-revenue-generating use cases, they've still scaled their compute margin to well over 50%.

swyx 31:20 Mm-hmm.

Jamin Ball 31:21 Which what you're-- So what you're seeing is they're getting hyper-efficient at the compute they bring online. The compute they decide to use for revenue-generating use cases is becoming, it's become 7X more efficient over the last year. I think you have to ask yourself, will they continue to ramp compute and capacity? I think the answer is yes. I think if you ask anyone at Anthropic, again, this is all speculation, they'd probably say, "We wish we started doing it sooner. We wish we were more aggressive."

swyx 31:46 They're trying to commit to more compute. That's-

Jamin Ball 31:48 That's right

swyx 31:49 ... something if you don't talk to any of these guys, you come away with like, oh my God, they're, you know, they, um, these guys are absolutely crazy and being fiscally irresponsible.

Jamin Ball 31:57 Yeah.

swyx 31:57 Uh, but I think if you have any confidence or if you see any, um, uh, if you see up close the, the demand for intelligence that people are snapping up and immediately consuming whatever you give them-

Jamin Ball 32:08 Mm-hmm

swyx 32:08 ... then like, yes, actually, you're trying to get more compute, and-

Jamin Ball 32:11 Yeah. Yeah

swyx 32:11 ... that, that's, the, the game is not to reduce the commitments, it is actually to get more commitments.

Jamin Ball 32:15 Yeah. Yeah.

swyx 32:15 They, they would have a trillion if they could.

Jamin Ball 32:17 Yeah, I don't think I've spoken to anyone in the AI industry, certainly not anyone at the big labs who's, who has said, "Yeah, we're, you know, I feel, I feel-

swyx 32:23 We're good. Yeah

Jamin Ball 32:23 ... good about the compute we have."

swyx 32:24 [laughs]

Jamin Ball 32:25 It's, it's almost always... And, and I think a underrated factor on compute demands in the future will be compute that's used for safety and alignment and verification. Like, I think more and more compute is going to go towards those functions, whether that's on research-

swyx 32:42 Yeah

Jamin Ball 32:43 ... or just on verification and alignment that happens at inference time. That will be-

swyx 32:48 Yeah

Jamin Ball 32:49 ... an increased compute demand for all of these companies.

swyx 32:50 Yeah, but we're probably not gonna break those out, you know? But like it'll-

Jamin Ball 32:53 Yeah, of course

swyx 32:53 ... it'll be like maybe 5% now and-

Jamin Ball 32:55 Yeah, of course

swyx 32:55 ... 20% in the future.

Jamin Ball 32:56 Of course. Of course.

swyx 32:57 Um, yeah, I think, you know, I, I would, I would just tell people, like, if you think this is bad now, wait till you see the OpenAI S-1.

Jamin Ball 33:02 Yeah.

swyx 33:03 Uh, and then like, it's not bad. It's like that's the whole point is like they're actually really trying to get compute from every source they can get-

Jamin Ball 33:09 Mm-hmm

swyx 33:09 ... and like, uh, OpenAI's buying TPUs, like- [laughs]

Jamin Ball 33:13 Yeah. Yeah

swyx 33:14 ... like there was a cold day in hell-

Jamin Ball 33:15 Every-

swyx 33:15 ... that that happened

Jamin Ball 33:15 ... they're buying TPUs, right? They have their own chip, uh, initiative, Jalapeno-

swyx 33:20 Yeah. Yeah

Jamin Ball 33:20 ... which has pretty rave reviews.

swyx 33:21 Uh-huh.

Jamin Ball 33:21 Anthropic staffed up a compute team to build their own chips. Obviously, the hyperscalers are building their own chips. It's a, I mean, every-

swyx 33:28 Mad dash and-

Jamin Ball 33:29 It's a, it's a mad dash

swyx 33:30 ... only, you know, only winners are TSMC, NVID- uh, NVIDIA, and ASML-

Jamin Ball 33:35 Yeah

swyx 33:35 ... uh, whatever, in the supply chain stack.

Jamin Ball 33:37 Totally. And then I think, look, it's, it's getting harder and harder, especially in the United States. You have a lot of pushback. Um, I mean, you have all kinds of pushback. Data centers are too noisy. Don't want them in my neighborhood. Uh, they add a lot of air pollution.

swyx 33:47 The water.

Jamin Ball 33:47 Don't want it. It's gonna... The water.

swyx 33:49 Oh my God, the water.

Jamin Ball 33:50 The water. It's, which is it uses too much water, which it doesn't. Um, it's the electricity. It uses a lot of electricity. It's gonna increase my electricity bill if I live in close proximity, or it's just the simple, AI is gonna kill us, and I don't want to enable that. Which again, you have all kinds of pushback. It's become hyper, hyper-political, but I think the unfortunate thing is it has become a lot harder. You do see a lot more pushback. Um, midterms are coming up, right? Who, who kind of like takes control in like the House and the Senate? Like w- could have a big impact on regulation or lack thereof around data center and intelligence center build-outs. And the good thing is that it's getting harder and harder to build these data centers. Uh, behind the meter is harder and harder, and so you do have a natural governor on the build-out, and you're, we're, I think we're a lot less likely to overbuild too quickly. That leads to a supply glut, right? Kind of the, the internet bubble analogy of the early 2000s, where we built way too much fiber. There wasn't as much demand for the internet. There was a lot of dark fiber, right? Fiber that had no bits and bytes traveling over it. We have no dark GPUs today. Um, and so it's, it's probably a good thing that there's some rate limiter. It will stop us from entering a huge bubble, but it's also bad [chuckles] because we need, you know, the industry needs more compute. Um, especially as these models get more and more powerful, you know, the security concerns go through the roof. You also need a lot more, like the defenders need more compute. The defenders need more capabilities.

swyx 35:15 Oh, yeah.

Jamin Ball 35:15 Which, you know, I don't, I don't know, personal belief, I, I think we're gonna be compute-constrained for, for far longer than, um, than the world thinks.

swyx 35:24 Yeah. Um, I, I mean, I'd be curious if what the, what the market's priced the world as thinking-

Jamin Ball 35:30 Mm-hmm

swyx 35:30 ... and then doing the work on like what do we actually think and have. The delta is probably, you know, could be three years, could be 30 years.

Jamin Ball 35:37 Yeah. Yeah, yeah, yeah.

swyx 35:38 And, uh, and that's a very material thing to figure out.

Jamin Ball 35:40 Yeah, of course.

swyx 35:40 Um-

Jamin Ball 35:40 Of course

swyx 35:41 ... I, I think one thing to guide people towards is, well, there's all this political stuff, and I, uh, honestly, I would say, um, Meta does a lot of work, and I think increasingly OpenAI does a lot of work in like, well, here's how we, we make sure we're a positive, um, presence in the communities-

Jamin Ball 35:56 Yeah

swyx 35:56 ... that we're in, right?

Jamin Ball 35:57 Mm-hmm.

swyx 35:57 Yes. You, you do need to take care of that. Um, and, uh, but I think like, uh, coming back to, again, research relevance, ROI, and, um, what it means about the industry, um, this profit, um, move, uh, here, I, I am, I, I do wonder sometimes, and this is a thing that has come up in our other podcasts, specifically tying in the research side-

Jamin Ball 36:19 Mm

swyx 36:19 ... is how much left is there to improve, right?

Jamin Ball 36:22 Yeah.

swyx 36:23 So was this-

Jamin Ball 36:23 Yeah

swyx 36:23 ... kind of like a one-time upgrade of our inference stack-

Jamin Ball 36:26 Yeah

swyx 36:26 ... wh- which, you know, there was like something like a 7X improvement in, in our inference efficiency.

Jamin Ball 36:31 Yeah.

swyx 36:31 Because like you can't, like- Probably, like, there's no improvements forever until the end of time-

Jamin Ball 36:37 Yeah

swyx 36:37 ... in our inference stack.

Jamin Ball 36:38 Yeah.

swyx 36:39 Like, unless we fundamentally change the chip, um, we probably-- like, we probably started out pretty unoptimized-

Jamin Ball 36:45 Mm-hmm

swyx 36:45 ... and we are now a lot more optimized than we were two years ago.

Jamin Ball 36:47 Yeah.

swyx 36:47 Okay. Well, there's a limit to how optimized you can get-

Jamin Ball 36:50 Mm-hmm

swyx 36:50 ... just like physical, like, joules moving around and-

Jamin Ball 36:52 Mm-hmm

swyx 36:53 ... bits and bytes moving around. So, uh, that, that is my question in terms of, like, the, the-- how much efficiency there's left-

Jamin Ball 36:59 Yeah

swyx 36:59 ... to squeeze.

Jamin Ball 37:00 Yeah.

swyx 37:00 And maybe there-- it might cap out-

Jamin Ball 37:03 Yeah

swyx 37:03 ... you know, in terms of, like, that, that year-on-year growth.

Jamin Ball 37:05 Yeah. Well, look, I th- I think, um, I, I have a chart I'll get to in a second, but I think this is a debate that's existed for quite some time.

swyx 37:12 Yes.

Jamin Ball 37:13 Right? You had, you know, the-- I don't remember when this was, two years ago? Uh, you know, like, the Ilya moment of, like, "Hey, the data and the internet is a fossil fuel. We've, we've exhausted it. There's no more data left to train on, and, like, we're hitting a plateau." Yeah, that was years ago. Then we hit inference time reasoning, average time compute, and reasoning models, you hit a new paradigm. Um, but I think maybe embedded in your question is, was the gap between the frontier and the leading open-weight model, is that shrinking? Like, is, is the gap that the frontier-

swyx 37:44 I don't-- Yeah, I don't mind the, the gap.

Jamin Ball 37:46 It-

swyx 37:46 Uh, it's not something I talk about.

Jamin Ball 37:47 It's-- But I think it's interesting-

swyx 37:49 Okay

Jamin Ball 37:49 ... because, like, what-- the real way to think about or to answer that question is not the, the gap between the best leading open-weight models today and the frontier models today, because the reality is, is OpenAI and Anthropic, they have far more powerful models that they've developed that they just haven't released yet. Like, this is a chart that Mark Chen published, uh, from OpenAI, actually, switching gears from Anthropic to OpenAI.

swyx 38:10 Yeah, he's been on the pod, Cooking Pod.

Jamin Ball 38:12 Oh, nice. Nice.

swyx 38:12 Uh, we, we, we made some soup for researchers.

Jamin Ball 38:15 Amazing. Amazing. That's-- I, I love that pod. Um, but you know, when they solved Navier-Stokes, that big math, uh, problem, you know, Mark came out and said, "Hey, here's a model." Uh, uh, it may be dark blue, light blue is harder to see, but, like, the top bar is an internal model we used to solve that problem. Uh, the blue bar is Astra. "Hey, look at how much more powerful the internal model is to what's publicly out there," which on some ways that, like, could scare you a little bit, right? Like, hey, these labs have even more powerful models. So if you think about what's the gap between open-weight and frontier today, I think it's actually far wider than people expect, because there are just far better models at the frontier that haven't been publicly released yet, either whether that's safety, testing, alignment, um, some combination of everything, you know, unclear. But I do think you're still seeing the frontier accelerate, um, even if it may appear that, um, what's released is plateau- I'm not even sure if it is. It does look like what's released has been plateauing.

swyx 39:16 Uh, no, no, no, no one's saying that it is-

Jamin Ball 39:18 Yeah

swyx 39:18 ... plateauing today. It's, it's, it's, it's we're all discussing could it plateau in the future?

Jamin Ball 39:22 Could it, yeah.

swyx 39:22 Right? Because that-

Jamin Ball 39:23 Certainly

swyx 39:23 ... that is a pretty important thing to figure out-

Jamin Ball 39:25 Mm-hmm

swyx 39:25 ... for your investing-

Jamin Ball 39:26 Yeah

swyx 39:27 ... uh, which is, uh, part of the, the deal here.

Jamin Ball 39:30 Totally.

swyx 39:31 Um, yeah, so, so that is a very well-phrased point. I thought you were going a different area-

Jamin Ball 39:36 Yeah

swyx 39:36 ... where you were gonna argue about bitter lesson, w- and you didn't, so kudos on that. Um, and, like, I, I think that's a very, a very fair point that I, I think, uh, we're, uh-- we've got still every generation, model generation still shows quantum leaps.

Jamin Ball 39:49 Mm-hmm.

swyx 39:49 This is only in math-

Jamin Ball 39:50 Mm-hmm

swyx 39:50 ... but OpenAI has a pretty good track record of generalizing beyond just math-

Jamin Ball 39:53 Yeah

swyx 39:53 ... or, or coding.

Jamin Ball 39:54 Yeah, totally.

swyx 39:55 Right? And so I, I do think that is. My point was a little bit more as on the financial analyst hat on-

Jamin Ball 40:01 Yeah

swyx 40:01 ... which is you had this one time 7x improvement in revenue efficiency-

Jamin Ball 40:05 Mm-hmm, mm-hmm

swyx 40:05 ... or whatever. Um, a-and, like, it's not like they raised prices.

Jamin Ball 40:09 Yeah. Yeah.

swyx 40:10 Right? So, like, where is it coming from?

Jamin Ball 40:12 Oh, here's another good one-

swyx 40:12 Do you know what I mean? They didn't raise prices.

Jamin Ball 40:14 I have for you. Here-- Oh, I think I have a chart.

swyx 40:15 So, so, so y-you know what I'm saying, right?

Jamin Ball 40:17 Yeah.

swyx 40:17 They have 100-- They, they have one-

Jamin Ball 40:19 Yeah, it's actually opposite. This is an awesome chart.

swyx 40:21 Yes.

Jamin Ball 40:21 This shows the cost to complete-

swyx 40:24 They're charging less?

Jamin Ball 40:25 It-- The ch-- Yes, exactly. The, the cost-

swyx 40:26 They're charging less and making more money?

Jamin Ball 40:28 The cost to complete a task with each model family. And so what you see is with whether it's Fable or Opus, and then you have Fable 51, um, and Opus 55, you see about a 25%, 25% to 40% reduction in cost to complete the same task with each successive model family.

swyx 40:48 Yeah.

Jamin Ball 40:48 And so it's kind of like, yes, the costs are going, the costs are going way down. Their compute efficiency is going way up. And I think a lot of that just demonstrates-- I mean, again, it's gonna sound a little, like, VC cliché, um, but, like, Jevons paradox-

swyx 41:01 Oh my God

Jamin Ball 41:01 ... like, as things get-- I, I know, I know, I know.

swyx 41:03 I knew it. I knew it was gonna come. I was like-

Jamin Ball 41:04 You can't have a podcast on AI without that coming up

swyx 41:07 ... I was preparing for it. Yeah, yeah. I know.

Jamin Ball 41:07 Um-

swyx 41:07 It, it is-- This is the appropriate time to bring it up.

Jamin Ball 41:10 Yeah.

swyx 41:10 Yes.

Jamin Ball 41:10 And it's just as things get... You know, you go back to, like, the early days of the internet. Like, how many bytes were traveling across the internet in 2000? How many bytes are traveling across the internet today? Like, many, many, many, many orders of magnitude more. And guess what? The price to transfer bytes over the network has, like, dropped by, like, many... Not, I mean, many orders of magnitude.

swyx 41:29 It's switching to meter.

Jamin Ball 41:30 Exactly.

swyx 41:30 It's switch-- It's the, the other switching to meter.

Jamin Ball 41:31 It's P and Q, right?

swyx 41:33 Yeah.

Jamin Ball 41:33 Price has gone way down-

swyx 41:34 Yeah

Jamin Ball 41:34 ... quantity has gone way up.

swyx 41:36 Yeah.

Jamin Ball 41:36 And I think you're gonna see the same thing, uh, with people buying and serving models and tokens.

swyx 41:42 Yeah. So, so, uh, generally pro supporter of that. I, I do want to remind people that that is a true objec-objective fact. This is not, like, VC bullshit. Like, this is, like, actually how the, the businesses are running-

Jamin Ball 41:54 Okay

swyx 41:54 ... and it, it is surprising to us-

Jamin Ball 41:56 Yeah

swyx 41:56 ... that, that I get while, yeah, we, we read about this i-in theory, but, like, to actually see this in financials-

Jamin Ball 42:01 Yeah

swyx 42:02 ... is, like, re-really ridiculous.

Jamin Ball 42:03 And you'll continue to see it. Think about, like, Jevon decision models.

swyx 42:06 Yeah, yeah.

Jamin Ball 42:06 You know, and, like, the price point of, hey, instead of you want a, a binary classifier, uh, you don't need a language model to do that. Just go use a decision model.

swyx 42:14 Yeah.

Jamin Ball 42:14 Super cheap, super efficient. I think you're gonna continue to see efficiencies like that. You know, I go back to S3 and EC2, AWS's big initial products, storage and compute. You've seen, you know, 99% price reductions in both of those services from the day they were launched. But, like, what's happened to the EC2 and S3 business, or just the broader storage and compute, you know, more CPU compute? Uh, it's, it's gone up exponentially. I mean, all the hyperscalers reported today, they're all over $100 billion run rate businesses today, if they were individual companies. Google Cloud's grown over 100%. It was just crazy.

swyx 42:48 [laughs]

Jamin Ball 42:48 Azure's growing almost 50.

swyx 42:50 Yeah.

Jamin Ball 42:50 Amazon reported today, we're recording on Thursday, uh, 39, almost 40%. Like, these businesses are just accelerating and growing at-

swyx 42:57 Beast mode

Jamin Ball 42:58 ... it's, it's crazy.

swyx 42:59 Beast mode.

Jamin Ball 42:59 It's crazy.

swyx 42:59 There's, there's... You should drop everything you're doing and just go start one of these things.

Jamin Ball 43:03 Yeah, yeah.

swyx 43:04 Uh, yeah, it's October 29th for those, those, uh, keeping track. And, and yes, a, a fun personal anecdote, I was in my hedge fund when, uh, the day they broke out Ama- AWS from Amazon.

Jamin Ball 43:13 Hmm, no kidding.

swyx 43:14 And at the time we were not long Amazon because we were worried about, uh, AWS margins-

Jamin Ball 43:19 Yeah

swyx 43:19 ... and also, like, oh, is it just startups? And what if, like, you know, we go through, uh, a, a startup winter and, like, all that?

Jamin Ball 43:26 Yeah.

swyx 43:26 It was so nonsensical.

Jamin Ball 43:28 Mm-hmm.

swyx 43:28 So, like, it, it... The, the funny thing about finance is that you can do all these works and make worse decisions.

Jamin Ball 43:33 Yeah. Yeah.

swyx 43:34 [laughs]

Jamin Ball 43:35 Yeah.

swyx 43:35 We definitely did that.

Jamin Ball 43:36 Total- totally.

swyx 43:37 [laughs]

Jamin Ball 43:37 Totally. I mean, that, that's the thing, like, market timing's impossible. Like-

swyx 43:42 And also-

Jamin Ball 43:42 You know

swyx 43:43 ... fixating on the wrong thing.

Jamin Ball 43:44 Yeah. That's right.

swyx 43:44 Which a lot of people are, are having to encounter for the first time.

Jamin Ball 43:46 That's right.

swyx 43:47 You have 2,000 data points-

Jamin Ball 43:49 Yeah

swyx 43:49 ... that were released-

Jamin Ball 43:50 Mm-hmm

swyx 43:50 ... in this thing-

Jamin Ball 43:51 Mm-hmm

swyx 43:51 ... and you'll fixate on one thing, and it'll be like that is-

Jamin Ball 43:53 Yeah

swyx 43:53 ... that's the everything that you need to know. And

swyx 43:55 like, well, no. Like, as you, you lost sight of the bigger picture.

Jamin Ball 43:57 Yeah. I mean, that's, that's for me. I mean, again, there's a lot of things that I believe that others would disagree with me on.

swyx 44:01 Yeah, yeah.

Jamin Ball 44:02 But I believe there's one question that matters, and that's can Anthropic stay at the frontier?

swyx 44:05 Yeah.

Jamin Ball 44:06 And to me, if I believe the answer to that question is yes, and I think everything they've done historically has demonstrated that, I think they'll build a great business and will continue to.

swyx 44:14 But okay, but then, like, there's a right price for that.

Jamin Ball 44:17 Of course, of course. And this is, this is not like, will it be priced right or this or that.

swyx 44:21 Yeah.

Jamin Ball 44:21 That's more of, hey, if you're able to take a longer term view on will the be- will the business be successful over time, I think to me, that's the most important question. Now, whether you think the IPO price is cheap, expensive, just right, I mean, we don't even know what that's gonna be yet. Um, prices can change in public companies for idiosyncratic reasons, reasons based on individual company performance, or they could change for macro correlated risk. Like, the reality is, is if Anthropic went public, um, a month ago, you know, the, the big route that you've seen in public markets recently, especially in semis and the AI trade, Anthropic would also be down, even if the business was doing well. So stocks can go up and down based on individual company performance as well as just-

swyx 45:00 Market data

Jamin Ball 45:00 ... market data. Exact- exactly. Exactly.

swyx 45:02 Yeah. So, you know, uh, you should be prepared for those kind of discussions if you're getting into this. I think, um, actually just as a cur- a side question, there's a lot of these, like, um, non-public market trackers of the stock.

Jamin Ball 45:14 Hmm.

swyx 45:15 Like, like, "Here's my proxy of OpenAI stock-"

Jamin Ball 45:17 Yeah. Yeah, yeah

swyx 45:17 ... "here's my proxy of Anthropic stock." Do you guys put any weight into that or, like, zero? You don't even-

Jamin Ball 45:21 I don't.

swyx 45:21 Yeah.

Jamin Ball 45:21 Yeah, I don't.

swyx 45:22 Yeah. I'm just... I, I'm wondering if I should pay attention to those things more closely-

Jamin Ball 45:26 Yeah

swyx 45:26 ... because they, they do give some indication of, like, well, here's the stock price reaction when a model just re- released.

Jamin Ball 45:32 No, I mean, it might, it might show some market sentiment.

swyx 45:34 Yeah.

Jamin Ball 45:34 Right? But I think Ramp had some data. I think I copied a Ramp chart, um, the other day. I mean, this was Anthropic spend-

swyx 45:41 Yeah

Jamin Ball 45:41 ... um, on some of their most recent models. The dark blue is Opus 5.5. And like, hey, what, what the Ramp data shows is-

swyx 45:49 People switch.

Jamin Ball 45:50 Uh, yeah, exactly.

swyx 45:51 Yeah.

Jamin Ball 45:51 And w- and with, with Opus 5.5, you saw some re-acceleration. And so would it surprise me if you saw a re-acceleration of revenue and run rate in October, and October won't be in Q3, that's the first month of Q4? No, it wouldn't surprise me, 'cause I do think a lot of the third-party data is showing and suggesting that.

swyx 46:08 Yeah.

Jamin Ball 46:08 Um-

swyx 46:09 Well, so okay, let me... Uh, this is, this chart is super interesting, right? Because you, you just said, like, when you, when you say stay at the frontier-

Jamin Ball 46:15 Mm-hmm

swyx 46:15 ... there's different levels with that.

Jamin Ball 46:16 Yeah. Yeah.

swyx 46:17 Stay at the frontier meaning we're at one of the top two, one of the top six-

Jamin Ball 46:21 Mm

swyx 46:21 ... we're in the top 20, whatever.

Jamin Ball 46:22 Mm-hmm. Mm-hmm.

swyx 46:22 Um, usually people mean top two.

Jamin Ball 46:25 Yeah.

swyx 46:25 Usually, but sometimes you can expand and say top six.

Jamin Ball 46:28 Yeah, yeah.

swyx 46:28 Google now has a, has a claim-

Jamin Ball 46:30 Yeah

swyx 46:30 ... to be a number one-

Jamin Ball 46:31 Totally

swyx 46:31 ... in some metrics.

Jamin Ball 46:32 Totally.

swyx 46:32 Uh, Meta doing super, super well. Uh, SpaceX doing super, super well.

Jamin Ball 46:36 Yep.

swyx 46:36 Not at, not number one-

Jamin Ball 46:38 Right, right

swyx 46:38 ... but number three, number four.

Jamin Ball 46:40 Yeah. Yeah.

swyx 46:40 You could say that.

Jamin Ball 46:41 Yeah.

swyx 46:41 So look at that, like, summer slump in-

Jamin Ball 46:44 Yeah

swyx 46:44 ... Anthropic revenue.

Jamin Ball 46:46 Yeah.

swyx 46:46 Um, was that because OpenAI was briefly ahead?

Jamin Ball 46:49 Yeah. I mean, I, I think Astro was a great model.

swyx 46:51 Yeah.

Jamin Ball 46:51 And I think a lot of people started using Astra.

swyx 46:55 Yeah.

Jamin Ball 46:55 Um, and it wasn't just Astra, right? It was, what was Google's models? Um, what was the one they just released, Carbon or Argon? Uh-

swyx 47:01 Uh, Carbon, yeah, Carbon, Argon-

Jamin Ball 47:02 It was yesterday.

swyx 47:02 Yeah. Yes, yes, next Thursday. Uh, they haven't released it yet. This is my gripe with Google. They never release anything. They just announce-

Jamin Ball 47:08 Yes

swyx 47:08 ... a wait list.

Jamin Ball 47:09 Yes.

swyx 47:09 And then, like, goddammit, like, just [laughs]

Jamin Ball 47:11 I think, I think, I think it was also you saw a massive explosion of, you know, Kimi K3, the GLM series, DeepSeek models. You saw, uh, open weight models get really powerful, um, and y- and get good quickly. So I think you kind of saw all of that, but it, it's, it's crazy, like, the far right bar, like, what was non-Opus 5.5 or Sonnet 5.5, like, that has gone, that's halved-

swyx 47:36 Mm

Jamin Ball 47:36 ... or more-

swyx 47:37 Mm

Jamin Ball 47:37 ... maybe more. You know, in, what is that? Two, three months.

swyx 47:41 Yeah, yeah.

Jamin Ball 47:41 And so, like, you do... Im- imagine if they didn't have Opus 5.5 or Sonnet 5.5, right? Like-

swyx 47:46 Yeah

Jamin Ball 47:47 ... who knows what the revenue growth and trajectory could have been?

swyx 47:49 Yeah.

Jamin Ball 47:49 And so, like, that's kinda why I do think it's important to be, have a frontier class-

swyx 47:54 Yeah

Jamin Ball 47:54 ... style model.

swyx 47:55 And so I, I will say from watching these labs for the past three, four years, they used to have more of a reserve.

Jamin Ball 48:03 Yeah.

swyx 48:03 Like, I will have internal models that I can just release whenever I feel like it.

Jamin Ball 48:06 Mm-hmm.

swyx 48:07 And, um, and then I think they, towards the GPT-5, 6 era, everyone just released whatever they had when they had it.

Jamin Ball 48:15 Right. Right.

swyx 48:16 Now I feel like the reserves are building up again. I don't know why, but, like, I, it definitely seems like, like, as you saw with the internal model, uh, it, it seems like they, they have more wood behind the shed-

Jamin Ball 48:25 Wood

swyx 48:25 ... for stuff like this.

Jamin Ball 48:26 Yeah.

swyx 48:26 Like, I can re- I, I'm w- sitting on Fable 5.5-

Jamin Ball 48:29 Yeah

swyx 48:29 ... I can release it the day before our IPO.

Jamin Ball 48:30 Yeah.

swyx 48:31 And like-

Jamin Ball 48:31 Yeah

swyx 48:31 ... like, that'll, that'll make my IPO go well.

Jamin Ball 48:34 Yeah.

swyx 48:34 You know? [laughs]

Jamin Ball 48:35 I mean, look, I, I, I think there's, I think there's good reasons, you know, that the models aren't released right away. You know, on one hand, there is just a lot of, like, safety-

swyx 48:43 There's safety checking, yes

Jamin Ball 48:44 ... and alignment and all that. You know-

swyx 48:45 But as a PM, as an e- as a, as a executive, like, playing this game-

Jamin Ball 48:49 Yeah

swyx 48:49 ... I can play it ... pretty well if I have a backlog

Jamin Ball 48:51 Totally. I mean, but, and it's, and it's also think about, um, you know, OpenAI's Patch the Planet initiative-

swyx 48:56 Oh, yeah

Jamin Ball 48:56 ... which is basically, "Hey, let's use these super powerful models to go find vulnerabilities"

swyx 49:02 Not even-

Jamin Ball 49:02 Yeah

swyx 49:02 ... it was 5/6. It was 5/6 Sole-

Jamin Ball 49:04 That's right

swyx 49:04 ... they found a PHP vulnerability.

Jamin Ball 49:05 Yeah. Oh, I didn't realize it was 5/6 Sole.

swyx 49:06 It wasn't frontier.

Jamin Ball 49:07 Yeah, which is crazy.

swyx 49:08 It was just-

Jamin Ball 49:08 So-

swyx 49:08 It was just they spam a bunch of like, like a last gen model

Jamin Ball 49:11 Exactly, exactly. And then let's, let's find all these vulnerabilities, like the PHP one, let's go create the patches so that we can instantaneously release the patch while announcing the vulnerability. The challenge is attackers are using these models, are using really powerful open weight models, and they're finding these vulnerabilities themselves.

swyx 49:27 Yeah. Yeah, it's a race.

Jamin Ball 49:28 And it's, it's a race. And it used to be, uh, you know, you had weeks, months from, like, vulnerability discovered to, like, exploit. I mean, now it's collapsed, and it's basically zero. You know, it's zero days. Like, vulnerability announced or discovered, exploit avail- like, it's, it's almost negative.

swyx 49:45 Yeah.

Jamin Ball 49:46 Um, and I think in, you know, OpenAI's announcement of the PHP thing, I think hidden in there was... I think they basically said, like, "Oh, and we, we- there are more serious vulnerabilities that we know about that we're working on patching-

swyx 49:58 For sure

Jamin Ball 49:58 ... right now"

swyx 49:59 Yeah

Jamin Ball 49:59 Which, I mean, more, more, more, more critical than, you know, something that touches 70% of the internet, um-

swyx 50:06 Yeah

Jamin Ball 50:06 ... is, is, is pretty serious.

swyx 50:08 Like, the 70% of the internet is just a proxy for old.

Jamin Ball 50:10 Yeah. Yeah, yeah, yeah.

swyx 50:11 And, like, of course it's older-

Jamin Ball 50:12 Mm-hmm

swyx 50:12 ... and it's gonna be more vulnerable.

Jamin Ball 50:14 That's right.

swyx 50:14 Yeah, of course, yeah. You know-

Jamin Ball 50:15 That's a good way to phrase it

swyx 50:16 [laughs] That's a-

Jamin Ball 50:17 Yeah

swyx 50:17 ... um, I think, uh, there's a couple of takes. I'm not that interested in, like, dwelling on, on that stuff only-

Jamin Ball 50:22 Hmm

swyx 50:22 ... because there's gonna be smarter people than me on that.

Jamin Ball 50:25 Yeah, of course.

swyx 50:25 The- there's one angle of this which is kind of funny, which is like, well, we're spending a billion tokens so, so finding all the vulnerability-

Jamin Ball 50:32 Hmm

swyx 50:32 ... and we're gonna spend a billion more to fix-

Jamin Ball 50:33 To fix it

swyx 50:34 ... the vulnerabilities.

Jamin Ball 50:35 Hmm.

swyx 50:35 What if we just didn't do either? And it, which [laughs] would like net the same position-

Jamin Ball 50:38 Yeah [laughs]

swyx 50:39 ... as to where we're at.

Jamin Ball 50:39 Yeah, yeah, yeah.

swyx 50:40 Anyway.

Jamin Ball 50:40 Well, I think cat's out of the bag on that one.

swyx 50:42 Yeah, cat's out of the bag, so-

Jamin Ball 50:42 You can't put the toothpaste back in the tube

swyx 50:44 So that-

Jamin Ball 50:44 Yeah

swyx 50:44 ... but that is one argument people make, like blame the AI instead of blame the-

Jamin Ball 50:47 Yeah

swyx 50:47 ... like kill the messenger.

Jamin Ball 50:48 Right.

swyx 50:48 The, the other part which is interesting is, like, I, I... And this might actually, uh, be an issue for basically nation state actors-

Jamin Ball 50:56 Mm-hmm

swyx 50:56 ... uh, who actually had all these exploits already.

Jamin Ball 50:59 Yeah

swyx 50:59 And now, like, well, you know, civilians are finding them, goddammit. Like-

Jamin Ball 51:02 Yeah, yeah

swyx 51:03 ... I now have to work harder.

Jamin Ball 51:03 Yeah, exactly

swyx 51:04 Like, all my zero days are gone. Like-

Jamin Ball 51:05 Yes

swyx 51:06 ... I, I had them for 10 years, like, uh-

Jamin Ball 51:07 Yes

swyx 51:07 ... whatever.

Jamin Ball 51:08 Yes.

swyx 51:08 But anyway, so, so these are fun anecdotes or consequences of this.

Jamin Ball 51:13 Hmm.

swyx 51:13 If you want to really do super Straussian and Machiavellian, if you're a nation state actor, what do you do-

Jamin Ball 51:18 Hmm

swyx 51:18 ... to, to, to preserve your edge?

Jamin Ball 51:20 Hmm.

swyx 51:20 Because, like, you cannot have the av- channels that you're relying on lock up.

Jamin Ball 51:25 Yeah.

swyx 51:25 Uh, so then you-

Jamin Ball 51:27 Interesting

swyx 51:27 ... then, then you send a Saudi decoy-

Jamin Ball 51:29 Yeah

swyx 51:29 ... to put the thing behind your TV-

Jamin Ball 51:30 That was crazy

swyx 51:31 ... so that people-

Jamin Ball 51:32 [laughs] That's crazy

swyx 51:32 ... so that, so that they get distracted by that-

Jamin Ball 51:35 Yeah

swyx 51:35 ... so that your real spy-

Jamin Ball 51:37 Yeah

swyx 51:37 ... gets in. [laughs]

Jamin Ball 51:38 Hmm. That's some-

swyx 51:41 This is-

Jamin Ball 51:41 ... War Games.

swyx 51:41 Yeah. I, I, I watch too much TV to, to, to think that what we are told is the full story.

Jamin Ball 51:47 Yeah. Yeah.

swyx 51:48 Right?

Jamin Ball 51:48 I think that's fair.

swyx 51:48 Right?

Jamin Ball 51:49 I think that's fair. Too much is at stake.

swyx 51:51 Yeah, it's too much is at stake. Um, okay, so coming back to the actual, again, finance-

Jamin Ball 51:54 Yeah

swyx 51:54 ... like I really

swyx 51:55 want to be a primer for everyone-

Jamin Ball 51:57 Yes

swyx 51:57 ... who are, are new to the economics of models and model labs and what, what have you. Um, one thing that have- the people have said is that the R&D per model-

Jamin Ball 52:06 Hmm

swyx 52:06 ... let's say, like, what does, what does s- Fable 5, Sonnet 5, Opus 5 cost?

Jamin Ball 52:12 Mm-hmm.

swyx 52:13 Let's call it a billion dollars.

Jamin Ball 52:15 Mm-hmm. Mm-hmm.

swyx 52:15 That lasted me six months.

Jamin Ball 52:17 Mm-hmm.

swyx 52:17 Right? Because you just said the dep- the depreciation, it last, it lasts six months.

Jamin Ball 52:21 Yeah, yeah.

swyx 52:22 And so isn't that worrying, right? Like, a- a- about, like, well, these things are the fastest depreciating assets in history.

Jamin Ball 52:28 Yeah.

swyx 52:28 This is something that me and Sarago have, have talked about before.

Jamin Ball 52:31 Yeah. Yeah, yeah, yeah.

swyx 52:32 Um, and I think the, the net is, is no-

Jamin Ball 52:35 Hmm

swyx 52:35 ... but, like, it's a qualified no-

Jamin Ball 52:36 Mm-hmm

swyx 52:37 ... of, well, as long as my next model run is better than the f- than the, than the current st- state-of-the-art, we're good.

Jamin Ball 52:44 Yeah. Yeah.

swyx 52:45 But it's a little bit of a house of cards if, if the next one doesn't work out.

Jamin Ball 52:48 Yeah.

swyx 52:48 Which, um, briefly aff- affected OpenAI.

Jamin Ball 52:51 Yeah.

swyx 52:51 And so a lot of people know by now that, you know, like GPT-4.5-

Jamin Ball 52:56 The big-

swyx 52:56 ... the big, the big run wasn't, wasn't as successful.

Jamin Ball 52:59 Yep.

swyx 52:59 Um, and they, they, you know, they, they've, they've since, since fixed it, but, like, that a slump-

Jamin Ball 53:03 Yeah

swyx 53:03 ... in a, in a run-

Jamin Ball 53:05 Yeah

swyx 53:05 ... or a failed run-

Jamin Ball 53:06 Yeah

swyx 53:06 ... uh, can really, really screw with the, the ROI on a thing.

Jamin Ball 53:08 And look, I think there's a really good analogy to the chip industry, and I remember speaking to someone, you know, much older and much smarter than myself, who did a lot of semiconductor investing in, you know, in the 90- 80s and 90s-

swyx 53:19 Hmm

Jamin Ball 53:19 ... um, about all these chip companies.

swyx 53:22 They've seen it all.

Jamin Ball 53:22 And, you know, they've seen it all, and they were highly skeptical.

swyx 53:24 [laughs]

Jamin Ball 53:25 Um, and, you know, sometimes you have industry vets that are highly skeptical and, you know, they miss the next big opportunity 'cause they never think the, the incumbent can be disrupted. But the point of view they had was it's very different to tape out a chip that is competitive with the leading chips at the time from building a machine that can continuously tape out-

swyx 53:46 Yes

Jamin Ball 53:46 ... chips that compete with, you know-

swyx 53:49 It's the model factory

Jamin Ball 53:50 ... the variable bins.

swyx 53:50 Yeah.

Jamin Ball 53:50 The, the chip factory.

swyx 53:51 Yeah.

Jamin Ball 53:51 And so he was like, this, this individual's point of view was maybe a chip startup can build one chip, but guess what? They're not gonna have the machine to constantly pump out a new chip every year and stay at the frontier, um, from the chip standpoint, and that's why Nvidia's king. It's not because they have the best chip now, it's they have the best factory.

swyx 54:07 Process for chips.

Jamin Ball 54:09 Exa- exactly.

swyx 54:09 Yeah. It's the machine that builds the machine that-

Jamin Ball 54:11 And so it's-

swyx 54:11 ... basically

Jamin Ball 54:11 ... kind of one of these, I mean, this is the bear case on all these Neo labs, which is, you know, and, and you talk to many smart folks, and what they might tell you is, "Hey, if you wanna really compete and be- have, like, a soda model at the frontier, you can't, you can't lo- you can't, uh, go more than one order of magnitude of compute less than the labs. You just can't compete if you don't have, you know, one order of magnitude of compute of the big labs." Um, and so as a Neo lab, you know, the question isn't can you build one model, it's can you attract, compute, and talent required to continuously pump out new generations of models every year to ultimately compete in your domain? And so I think for the labs, the reality is, is I think it's very hard- for anyone unknown players to catch up to kind of like the generalized frontier big models. I think you'll continue to see domain-specific models, um, or models or like new types of models, decision models like Jev. Um, but I think the, I think the big labs, they are on the treadmill of continuously pumping out new models because guess what? If, if Anthropic stops, OpenAI and Gemini will continue, and then you'll just kind of see revenue, you'll see revenue fall off, which is why I do think it's so important to stay at the frontier. Now, what do I expect and like what have I seen from Anthropic now in the data? It's as their model families get better and better, they get more efficient, but they also capture more revenue.

swyx 55:43 Hmm.

Jamin Ball 55:43 That could be because they're doing more economically valuable work, or they're doing the work faster. Um, but what you're seeing from n- now kind of black and white is, hey, as Anthropic has 2x'd their compute spend, uh, to build and train these models, they've 14x'd their revenue. And like I'm not really sure what would cause that trend to stop. Um, I think models are only gonna take on longer-running tasks, um, more economically valuable tasks. And, you know, if you can't tell, I'm a hyper-optimistic person-

swyx 56:18 [laughs]

Jamin Ball 56:18 ... and so may- maybe I'm just drinking the Kool-Aid too much. Uh, but I d- I do think we're nowhere close to hitting a limit of the dollar kind of like economic value that leading models can capture. I think what's maybe more interesting is, you know, OpenAI, you know, Luna class models kind of also trying to capture and compete at the low end of the market with much smaller, uh, much cheaper models, you know, and kind of span everything from, um, frontier to, you know-

swyx 56:49 Small, yeah

Jamin Ball 56:50 ... distilled versions of their own models that are smaller in and, you know, cents per million input/output tokens, not dollars.

swyx 56:58 Yeah. Um, uh, cosign all of that. I think, uh, that is, uh, a good way to think about it. Um, I, I, I think, uh, you know, fun- I'll just comment this 'cause it's in my head. Uh, S-1 Terra has been pretty, pretty successful, S-1 Luna. Like that, that whole series is very interesting.

Jamin Ball 57:13 Mm.

swyx 57:13 The approaches of the different labs is also super interesting.

Jamin Ball 57:16 Yeah.

swyx 57:17 Um, Anthropic to this day has no decision model.

Jamin Ball 57:19 Mm-hmm.

swyx 57:19 And they're like-

Jamin Ball 57:19 Mm-hmm

swyx 57:20 ... "Well, we're good."

Jamin Ball 57:20 They're good.

swyx 57:20 Like, "We'll use TypeSafe, whatever."

Jamin Ball 57:22 Yeah, yeah, yeah.

swyx 57:22 Um, and like what they don't have a standalone voice API.

Jamin Ball 57:25 Right. Right.

swyx 57:26 They, they, they have a voice model.

Jamin Ball 57:27 Yeah.

swyx 57:27 They just don't, don't pub- market it.

Jamin Ball 57:29 Right.

swyx 57:29 They don't-- They've never done images.

Jamin Ball 57:31 Right.

swyx 57:31 You know, it's like the, it's a lot of enterprise focus-

Jamin Ball 57:33 Yes

swyx 57:33 ... coding focus.

Jamin Ball 57:34 Yes.

swyx 57:34 And OpenAI has been a little bit more of like, "We'll do everything-

Jamin Ball 57:37 Yeah

swyx 57:37 ... and we'll try Sora. We shut down Sora."

Jamin Ball 57:40 Yes, yes.

swyx 57:40 Remember when Sora was like the number one app in-

Jamin Ball 57:41 Yeah, exactly

swyx 57:42 ... in the US?

Jamin Ball 57:43 Exactly. Yeah, exactly. Yeah.

swyx 57:44 Um, and then Gemini has been the most like-

Jamin Ball 57:46 Yeah

swyx 57:46 ... we will explore everything. Um, and so like we appreciate them because if they didn't try it, no one else would.

Jamin Ball 57:51 Yeah.

swyx 57:51 Right? Like they have the funds for it.

Jamin Ball 57:53 Yeah.

swyx 57:53 Um, but like I think, uh, often about Anthropic in terms of like, well, okay, enterprise and coding, and they, they're first to it. They nailed it.

Jamin Ball 58:01 Mm-hmm.

swyx 58:01 They st- continue to nail it, right? And everyone's chasing after them.

Jamin Ball 58:04 Yep.

swyx 58:04 Um, is that the story going forward? Is, is, uh, is this just more of the same or

swyx 58:09 like w- how, what should we expect?

Jamin Ball 58:10 I think it's-- I think a lot of this boils down to like a new type of innovator’s dilemma, you know, which I almost call like the compute innovator’s dilemma. Like let's say you're OpenAI, and you have Sora. You have many different model families. You have image models. You have everything. Well, the reality is, is how do you choose what to serve? Maybe Sora demand explodes. Are you gonna allocate a lot of your Codex compute to Sora?

swyx 58:36 To slop videos? Yeah. [laughs]

Jamin Ball 58:37 Um, because you probably will have to. Like if you actually wanna make the Sora-

swyx 58:40 I mean, that's exactly why they shut down Sora

Jamin Ball 58:41 ... yeah, experience good-

swyx 58:42 Yeah. [laughs]

Jamin Ball 58:42 ... you need to. And at the end of the day, what a lot of people have decided is, well, guess what? A lot of the, the highest ROI, most economically valuable work has been in coding, so it's hard to prioritize anything else other than that. It's why there's probably opportunities for other companies and startups to build models and build products in those domains because guess what? What is OpenAI and Anthropic chasing? They're not chasing the incremental $10 billion market cap. They're chasing the next trillion. And so the reality is, is for a startup or a Neo lab, great, you could go build a, build a company, build a product, build a model where you exit for $10 billion, and it's a home run, grand slam. Um, that's awesome. And simultaneously, it's too small for OpenAI and Anthropic to matter. And so like what has Anthropic done just a bang-up job at? They said, "Coding is our thing. We're gonna be great at coding. We're gonna build the best coding harness in Claude Code, and we're gonna continue to nail it." I think everyone is kind of starting to ask themselves, "Well, when do we..." Maybe, maybe you asked a version of this earlier. "When do we tap out the coding TAM?" Um, I think one point of view is-

swyx 59:41 I have asked a very, very many times this question

Jamin Ball 59:43 ... but I think one point of view is, hey, coding is actually the best generalizing-

swyx 59:48 Knowledge work

Jamin Ball 59:48 ... model and framework.

swyx 59:48 Yes.

Jamin Ball 59:49 And like that can be applied... You know, you think of it as coding, but when I use Claude Code, like I'm not coding, [laughs] but I'm using Claude Code. When you use Codex, are you, are you coding? No, these are-- they're generalizing, um, and coding is a good framework for generalizing knowledge work. I do think you're seeing a big push by both OpenAI and Anthropic into life sciences and drug discovery. I think they've both been quite public, um, about their interests, um, to kind of like capture that domain as, as a new field.

swyx 1:00:21 Yeah.

Jamin Ball 1:00:21 And I do think those are just different, um-

swyx 1:00:24 Yeah, it would seem like Anthropic is investing more than OpenAI on-

Jamin Ball 1:00:26 I, totally

swyx 1:00:26 ... specifically the life sciences stuff.

Jamin Ball 1:00:27 I'd agree.

swyx 1:00:28 Yeah.

Jamin Ball 1:00:28 Yeah, I'd, I'd agree. I'd agree. Um, but, uh, now I can't remember the question you asked. Um, but I, but I do think you're starting to see these companies evolve the scope of what they want to go after, and it has been very targeted on coding given it's been the best generalizing, you know, you wanna generalize knowledge work. These models have generalized the best as they've been stronger in coding. Um, but it also can give it some, a soft underbelly to other areas.

swyx 1:00:54 Yeah.

Jamin Ball 1:00:54 Maybe it's visual thinking and reasoning. You're seeing great models there like Ellurian. Maybe it's decision models like Jev and TypeSafe, right? Like there's, there's a lot of different, uh- Soft underbellies they may have, which is opportun- which is great. It's opportunities. Like-

swyx 1:01:07 Yeah

Jamin Ball 1:01:07 ... you'd love to see better, cheaper models in other domains. You know, you don't need to drive the Ferrari engine to the grocery store. Hey, maybe you- a Toyota will do just fine.

swyx 1:01:16 [chuckles] Yeah.

Jamin Ball 1:01:16 And I think that will continue.

swyx 1:01:18 Um, and good to Lorien shout-out. We just interviewed Andrew-

Jamin Ball 1:01:20 Oh, no kidding

swyx 1:01:21 ... uh, on his benchmark and, and, um, and Glimpse, I think was his-

Jamin Ball 1:01:23 Glimpse, yeah

swyx 1:01:24 ... the new... Are you guys invested or?

Jamin Ball 1:01:25 We are.

swyx 1:01:25 Oh, yeah. Oh, wow.

Jamin Ball 1:01:26 Shameless promotion.

swyx 1:01:27 Wow, okay. All right.

Jamin Ball 1:01:27 [laughs] Yes.

swyx 1:01:28 Uh, it's, it's, it's weird to... Like, this is a whole different discussion, but Altimeter I don't get as like, well, you will lead the giant rounds in, like, uh, you know, the, in topics of the world, but you also care about Lorien.

Jamin Ball 1:01:38 Yeah.

swyx 1:01:39 It's, like, way smaller-

Jamin Ball 1:01:40 Yeah, yeah, totally

swyx 1:01:40 ... and, like, you know, probably worth, like, a, a minute of your time.

Jamin Ball 1:01:43 Yeah.

swyx 1:01:43 [laughs]

Jamin Ball 1:01:43 Well, no, like, like, we're, we're the ton of my time. I mean, I would say, like, we barbell.

swyx 1:01:47 Yeah.

Jamin Ball 1:01:47 I mean, we're lifecycle investors.

swyx 1:01:49 [laughs]

Jamin Ball 1:01:49 We invest early, we invest late, and we invest in the public markets.

swyx 1:01:52 Yeah.

Jamin Ball 1:01:53 And I think there's a lot of value. Like, the way we think about it is some of the best founders today, they need late-stage capabilities early on.

swyx 1:01:59 Yeah.

Jamin Ball 1:01:59 And late-stage capabilities isn't just capital. It's relationships with execs. It's relationships with public companies. I mean, it's... Companies are built on, on hyper mode now, right? It's not, "Let me go raise a round, and let me, you know, add 10 million of ARR in 18 months, and I'll go raise my next round-

swyx 1:02:15 No

Jamin Ball 1:02:15 ... in a few years later." It's like you're, you're growing to 100 million super fast. You're, you're raising nine figures of, of money super quickly, and it's, and it's hard. I mean, in many ways, you, like, you have to, like, break the laws of physics to how, like, cultures can be maintained-

swyx 1:02:27 Like lawyers-

Jamin Ball 1:02:28 ... hyper-

swyx 1:02:28 ... lawyers are not set up for this.

Jamin Ball 1:02:29 Yeah. No, it's hard.

swyx 1:02:30 They're like, "What do you mean why don't we come here on a weekend to... because, like, Monday might be different?"

Jamin Ball 1:02:33 Yeah, exa-

swyx 1:02:34 [laughs]

Jamin Ball 1:02:34 ... exactly, exa- exactly. I mean, it's... You know, you used to think about as a board member and investor, "Hey, let me think ahead." Like, let, let-

swyx 1:02:40 Next, next quarter

Jamin Ball 1:02:41 ... let me make sure we hire the VP of sales and the CFO, you know, six months before, uh, before we need to, so we're not behind. Now, it's like, hey, six months before you need to may have been before the company was even incorporated, you know? And so how do you, how do you do that? Like, when do you hire the right execs so you're not doing it too late? How do you hire ahead but not too ahead? I mean, it's-- I feel like the pressure of building companies has never been greater. Now, the size of the prize has never been greater. Um, but it is. It's this... You know, it might appear as easy as ever to build a company, but I actually think it's the opposite. It's almost as hard as ever to build a good company, and maybe more importantly, to build a durable company, which is, I think, what, what, what everyone aspires to do.

swyx 1:03:17 Yeah. Uh, I'll also shout out the, on, you know, on the model factory thing, um, we also interviewed Poolside-

Jamin Ball 1:03:21 Oh, nice

swyx 1:03:22 ... which got acquired by NVIDIA.

Jamin Ball 1:03:23 Yes.

swyx 1:03:23 They were pitching this model factory really hard. And, and, and I do think, like, uh, the business that builds the business is kind of like the, the, the-

Jamin Ball 1:03:29 Mm-hmm

swyx 1:03:29 ... the general-

Jamin Ball 1:03:29 Mm-hmm

swyx 1:03:30 ... uh, takeaway that I want people to know. Um, one of the things I wanted to-- I was diving towards, I, I didn't, I didn't quite start it yet, but I'll start it now, which is typically when you have an S1, you have a clear sense of the business, uh, let's call it market segments, product segments-

Jamin Ball 1:03:43 Mm-hmm

swyx 1:03:43 ... the org chart, the whatever.

Jamin Ball 1:03:45 Mm-hmm.

swyx 1:03:46 Um, and we got Claude Code-

Jamin Ball 1:03:48 Yeah

swyx 1:03:48 ... Claude Cowork-

Jamin Ball 1:03:49 Yeah

swyx 1:03:50 ... miscellaneous-

Jamin Ball 1:03:51 Yeah

swyx 1:03:51 ... uh, or platform, whatever.

Jamin Ball 1:03:52 Yeah, yeah, yeah.

swyx 1:03:53 Uh, and I'm like, okay, what about legal? What about design? Like, all this stuff that you've added. Claude.ai, the consumer business.

Jamin Ball 1:03:59 Yeah, yeah.

swyx 1:04:00 For a while, that was a thing.

Jamin Ball 1:04:01 Mm-hmm. Mm-hmm.

swyx 1:04:02 The, the, um, do we care? Do we-- Is this all, like, one revenue bucket?

Jamin Ball 1:04:06 Mm.

swyx 1:04:06 You know?

Jamin Ball 1:04:07 I, I think we do care.

swyx 1:04:10 Like, the, what, the products, right?

Jamin Ball 1:04:11 Yeah.

swyx 1:04:11 Like this, this Claude-

Jamin Ball 1:04:12 Yeah

swyx 1:04:12 ... Code, and what else?

Jamin Ball 1:04:14 I, I-

swyx 1:04:15 [laughs]

Jamin Ball 1:04:15 And this is where I do think, and there was, there was a section of the S1, um, where they broke out some of these, like, vertical applications they wanna go after. I mean, it, it, it's kinda funny you bring this up. Like, I had a hypothesis, you know, years ago that, like, hey, maybe the labs will stop releasing models, and instead they'll just build applications and products powered by their models.

swyx 1:04:36 Yes.

Jamin Ball 1:04:36 And if they're powered by the best models that are never released publicly, hey, maybe th- they have the right to build all the best applications. Uh-

swyx 1:04:42 That you kill Harvey by only using Claude, uh, Fable Six Legal.

Jamin Ball 1:04:46 Right. Or it's, "Hey, you know what, Cursor? Guess what? We have... Well, you know, years ago, we got the best model, and we're not releasing it. So guess what? We're gonna have the best coding agent because our model that's the most powerful isn't released publicly." And I do think it's why

Jamin Ball 1:05:01 some of the leading application companies today are moving towards in-housing their models.

swyx 1:05:06 Yes.

Jamin Ball 1:05:07 They wanna control their own destiny. You need to. I think it's gonna be really hard as an application company over the long haul to survive without a research team internally. Um, it's one of the, like, an investment theses that I have that if we're, you know, investing in applications, like, I wanna believe there's two things, really strong product org, but also a really strong research org. And I also wanna believe that they're aligned, and as the research gets better, the product gets better, which sounds counterintuitive, but there are a lot of research orgs that are just doing research for the sake of research versus-

swyx 1:05:35 Yeah

Jamin Ball 1:05:36 ... closing the loop and actually, like, having a, a, a positive cycle of, of product improvements with research improvements.

swyx 1:05:43 I'm, I'm not, uh, you know, I'm not in Anthropic, but I have the benefit of being friends with fair- a fair amount of people in Anthropic, uh, and I would say research is king at Anthropic. [laughs]

Jamin Ball 1:05:51 Yeah, totally. Totally.

swyx 1:05:52 By far.

Jamin Ball 1:05:53 Totally.

swyx 1:05:53 You know, and, and so part of my whole thesis around model labs versus agent labs is-

Jamin Ball 1:05:58 Mm-hmm

swyx 1:05:58 ... uh, model labs will, you know, respect the model and AG- and try to build AGI, and products will be a final last mile-

Jamin Ball 1:06:05 Yeah

swyx 1:06:06 ... so the delivery mechanism, whereas the agent labs will, will respect the customer the most and build whatever they need, and th- and they'll, they'll backwards integrate to models as and when.

Jamin Ball 1:06:14 Mm-hmm.

swyx 1:06:14 But they'll never be, probably never be at the frontier for, for those things.

Jamin Ball 1:06:17 Yeah.

swyx 1:06:17 And both can coexist.

Jamin Ball 1:06:18 Yeah, totally.

swyx 1:06:18 Both, both have a market.

Jamin Ball 1:06:19 Totally.

swyx 1:06:19 Um, but yes-

Jamin Ball 1:06:20 Maybe, maybe to your question, I do expect to see a much bigger push by both labs, Anthropic and OpenAI, into vertical applications and kinda, like, diversifying out of just delivering models. I, I think-

swyx 1:06:32 I, I think the, the actions of Anthropic show so far have shown no serious intent at it.

Jamin Ball 1:06:37 That's true. Yeah.

swyx 1:06:37 Because Claude Design is just a tab inside of-

Jamin Ball 1:06:39 That's right

swyx 1:06:40 ... Claude.

Jamin Ball 1:06:41 That's right.

swyx 1:06:41 It's not, like, a whole Adobe, you know, replacement.

Jamin Ball 1:06:44 Mm-hmm.

swyx 1:06:44 Claude Legal is a markdown read-

Jamin Ball 1:06:45 Mm-hmm

swyx 1:06:45 ... file, you know, markdown repos.

Jamin Ball 1:06:47 Mm-hmm. Mm-hmm.

swyx 1:06:48 Uh, which, like, that is the AGI pill is if, if your research is king and, like, you know, o- one model does everything, that is what you do.

Jamin Ball 1:06:53 Yeah.

swyx 1:06:53 Right? So-

Jamin Ball 1:06:54 It's also just classic enterprise and org management.

swyx 1:06:56 Yeah.

Jamin Ball 1:06:56 You have a cash king product-

swyx 1:06:58 Yeah, yeah

Jamin Ball 1:06:58 ... that think your sales reps, what do they wanna sell? The thing they know how to sell, that they've been selling, that is crushing it.

swyx 1:07:04 Yeah.

Jamin Ball 1:07:04 What's harder to sell? Like-

swyx 1:07:05 New things

Jamin Ball 1:07:06 ... new things, new SKUs, and it's-- So it's inertia, it's momentum, and it's just classic, like, big company boring stuff.

swyx 1:07:12 But it's just like, you know, I, I think for someone, like, I've been through maybe 15 years of IPOs.

Jamin Ball 1:07:18 Mm-hmm.

swyx 1:07:19 And For an IPO stage company, much less a $2 trillion company-

Jamin Ball 1:07:24 Yeah. Yeah

swyx 1:07:24 ... you've never seen like, "Well, we have three SKUs." [laughs]

Jamin Ball 1:07:27 Yeah. Yeah, exactly. Exactly. But, but I think to Anthropic's credit, they've been really good at focus, and like they haven't been, you know, enamored with the next thing or-

swyx 1:07:36 No robots.

Jamin Ball 1:07:38 Yeah.

swyx 1:07:38 No images.

Jamin Ball 1:07:39 You know, it's... And again, you look at OpenAI, it's tough. Like-

swyx 1:07:41 [laughs]

Jamin Ball 1:07:41 ... who are they competing with? They're competing with NVIDIA, they're competing with Google, they're competing with Meta, they're competing with Anthropic.

swyx 1:07:49 Yeah.

Jamin Ball 1:07:49 You know, like they're- they have different lines of business that are all competing with giants, like the biggest tech company in the world, and it's like it's, it's a tall task to take on one at a time, let alone all of them at a time.

swyx 1:07:59 Mm.

Jamin Ball 1:07:59 But hey, that's, that's the game you take on if, you know, when you're a trillion-dollar company.

swyx 1:08:04 Yeah.

Jamin Ball 1:08:04 You know, the stake- the stakes are just

Jamin Ball 1:08:05 higher.

swyx 1:08:05 Yeah. Um, okay, I wanna round out the Anthropic discussion and, and then we can go to general commentary on, on other stuff.

Jamin Ball 1:08:10 Yeah.

swyx 1:08:11 We, we've, uh, cheated a bit with Ailorian and, and TypeSafe, but, uh, have we covered the main debates that you've, you've been engaged in? You know-

Jamin Ball 1:08:18 There's, um, there's two others. So, um, this is gonna be very like financially in the weeds. Um, there- If you look at the-

swyx 1:08:29 You're gonna talk about convertibles.

Jamin Ball 1:08:30 Yes, exactly.

swyx 1:08:31 Yes.

Jamin Ball 1:08:32 Ha. [laughs] Okay. So, um, on the balance sheet, you know, earlier in the episode we talked about, hey, 140 million of cash raised, uh, capital raised, 134 billion of cash. That's what the s- the balance sheet shows you. What the balance sheet also shows is a... I'm, I'm, I've, I'm gonna... I might mess up the exact numbers 'cause I don't have 'em in front of me, but I think about 108 billion of convertible notes payable as a liability. And so the question is, well, is that debt? Did they raise debt? Um, and what you also see is a huge portion below the operating income line of like other income and expenses that makes it look like they have a huge net loss. They're both related, and let me explain. Um, so any company and any startup founder listening to this would understand and appreciate this, like there's different ways to raise capital. You can raise preferred equity, you can raise converts, you can raise SAFEs. There's a whole different host of options to raise money. Anthropic, just like every other startup, did everything. They raised some preferred equity, they raised some converts, um, they raised some SAFEs. And along the way, there's different reasons that certain investors or the company might prefer one or the other. When it comes to strategic investors, there's even more reasons why they might prefer one or the other, and I think it's, it's very well known that some of the strategic investors, Amazon and Google in particular, were very large investors and very big owners of Anthropic early on, right? And-

swyx 1:09:50 And... Yeah. Okay, go ahead. Sorry.

Jamin Ball 1:09:52 And, and so early on, Anthropic and Google, '23, '24, '25, a lot of the dollars that they provided the company were in the form of converts and SAFEs and notes. Um, those, the value, the, the, the value that those converted at, at the time, was pegged more towards the valuation those companies were raising at at the time. Why might they prefer a convert? Well, they're public companies, uh, Google and Amazon. If they have an ownership in a third party that exceeds a certain limit, they now have to start reporting more on that look-through ownership, right? If you own 20% of another company, you almost have to-

swyx 1:10:29 It's consolidated financial statements.

Jamin Ball 1:10:30 It can... Exactly.

swyx 1:10:31 Yeah.

Jamin Ball 1:10:31 And that company's losing a lot of money, you now have to report that loss on your books.

swyx 1:10:34 Oof. Oof.

Jamin Ball 1:10:34 And so it's, they did... You know, strategics don't want to hit certain ownership thresholds, so, "Hey, give us a convertible, give us a no- we'll give you cash. You haven't given us shares yet, and so we don't have that equity ownership. We don't, we haven't gone above those thresholds." Maybe Anthropic didn't wanna give 'em voting shares. I don't know, right? "Let's, let's have your shares convert into non-voting shares 'cause we don't want the governance of a s- of a supplier or a strategic." Like I'm, I'm totally speculating and hypothesizing.

swyx 1:11:01 These are all reasonable-

Jamin Ball 1:11:02 But, you know-

swyx 1:11:03 ... reasons to do it

Jamin Ball 1:11:03 ... reason- reasonable reasons why, hey, they were converts. But the conversion price was at a much lower valuation relative to what it is today, okay?

swyx 1:11:12 Mm-hmm.

Jamin Ball 1:11:13 And so when you look at the balance sheet, what... And balance, balance sheet have to balance. You have assets equal equity plus liabilities. They have, it has to balance. And so when you raise a convert, you get cash. Your assets go up. Well, if your balance sheet's gonna have to balance, your liabilities also have to go up, and it makes sense. A convert is a promise for shares in exchange for cash, and so Anthropic took on that liability, and at the time, the liability matched the cash raised. Now, what's happened to Anthropic's valuation since 2023 and 2024?

swyx 1:11:49 200x.

Jamin Ball 1:11:49 It's, it's gone way up.

swyx 1:11:51 Yeah.

Jamin Ball 1:11:51 And so now what happens is the dollar value of your liability, because you promised shares at a price, the dollar value of that liability has skyrocketed because the valuation of the business has gone up. What does that mean? Okay, now the liability looks big. It looks like $108 billion debt that Anthropic raised. The reality is, is they didn't raise $108 billion of debt. There was about $11 billion in those combined convertible notes. About $6 billion has already been... Or no, $5 billion has been converted to equity. There's 6 billion outstanding. The market value of that 6 billion is 108 billion because the company's value has appreciated meaningfully. And so there's this debate of like, well, they have 108 billion of debt on the balance sheet. They haven't raised 141. They've raised 141 plus 108. Um, that debate is just misguided. Um, there is no debt on the balance sheet, and the 11 billion of converts is inclusive in the 141 billion of total capital raised. Um, and so debate one of, well, have they really only burned-

swyx 1:12:54 Can I, can I just say good job summarizing this?

Jamin Ball 1:12:56 Yeah.

swyx 1:12:57 Uh-

Jamin Ball 1:12:57 But, but it, it actually goes deeper.

swyx 1:12:58 Yes.

Jamin Ball 1:12:59 Um, that's just the balance sheet.

swyx 1:13:00 Okay.

Jamin Ball 1:13:00 Okay? Now let's look at the income statement, the P&L.

swyx 1:13:03 Yeah.

Jamin Ball 1:13:04 They have an operating profit. You could say great, GAAP operating profit. But what comes below operating profit? Net income. Someone could say, well, in the first six months of 2026, they had negative $80 billion-

swyx 1:13:19 Oh my God [laughs]

Jamin Ball 1:13:20 ... of net income. So you could say, and this is a debate, this is what people are saying. Again, it's very misguided. They're losing $80 billion.

swyx 1:13:26 [laughs]

Jamin Ball 1:13:27 Like-

swyx 1:13:27 It's true

Jamin Ball 1:13:28 ... what a terrible business.

swyx 1:13:29 Yeah.

Jamin Ball 1:13:29 Technically, by GAAP accounting, it is. But okay, let's go back, and I'm gonna go back to the balance sheet. The liability has gone way up. That liability is now 108 billion. Okay, let's think about the, the balance sheet. Assets equals liabilities plus equity. As liabilities have gone up, two things have to happen. Either your assets have to go down or your equity has to go down. Well, your assets haven't gone down. That means your equity, the, the value of your equity, has to go down. This is where it might get a little complex. [laughs] But how the, how the three statements link is-

swyx 1:14:03 Yeah, people can take Accounting 101. It's, yeah

Jamin Ball 1:14:05 ... net income-

swyx 1:14:06 Yeah

Jamin Ball 1:14:07 ... on the P&L flows to retained earnings on the equity value of your balance sheet. And so what has to happen as that liability has gone up, retained earnings has to go down by the same amount. Which basically means net income has to go down by the same amount, because net income on the P&L flows to retained earnings on the balance sheet.

swyx 1:14:27 It's, it's all mathematically linked.

Jamin Ball 1:14:28 And so-

swyx 1:14:29 You, you can't break it, yeah

Jamin Ball 1:14:29 ... it, it's all mathematically linked. You can't break it. But what's happened is in order for those statements to link and for those statements to tie, you need a huge decrease in value on your income statement. And what ha- what happens is, guess what? Anthropic's value has just skyrocketed recently. Um, they've raised big rounds over the last nine months. Um, and so as those converts have appreciated in value, the value that those notes have appreciated shows up as a loss on the P&L. And of the $81 billion of loss in net income on the first six months of this year, 72 billion is from a series of converts for Anthropic, uh, for AWS and Google, and 9 billion is for a separate set of safes and notes. And so 100% of the loss in the first six months is just the value appreciation of their converts. If Anthropic never issued converts to Amazon and Google, just did straight equity, that loss would p- it wouldn't be there.

swyx 1:15:31 Yeah.

Jamin Ball 1:15:31 And so it's kinda one of these, like, technically from GAAP accounting, they do show an $80 billion net loss. Do I view that as an accurate representation of the business? No. Because the reality is, is 100% of that loss is just the appreciation of their equity-

swyx 1:15:47 Yeah

Jamin Ball 1:15:47 ... which everyone would say is, is a good thing. And as those notes convert to equity, those go away, because you no longer have this dynamic of your liability going up as your stock price goes up, which means your net income has to go down over here. So, you know, all the financial statements tick and tie. And I, yeah, sorry if that got too, like, financial-

swyx 1:16:09 No

Jamin Ball 1:16:09 ... technical in the weeds

swyx 1:16:09 ... that was a really good recap of the discussion. That should be a de- definitive answer for anyone, uh, who has come across that topic. Uh, there's another discussion about the circular financing-

Jamin Ball 1:16:19 Yeah

swyx 1:16:19 ... of, of Google and Amazon. We don't have to go into that. I mean, I think that is small in comparison to the numbers we're talking about today.

Jamin Ball 1:16:25 Yeah.

swyx 1:16:26 Um, and uh, in general, like, you know, book value accounting is not that relevant-

Jamin Ball 1:16:31 Mm-hmm

swyx 1:16:31 ... c- compared to overall enterprise value.

Jamin Ball 1:16:33 Mm-hmm.

swyx 1:16:33 Um, and-

Jamin Ball 1:16:34 Oh, and maybe one other thing related to this.

swyx 1:16:35 Yeah.

Jamin Ball 1:16:35 I think there are a lot of people that say, "Okay, well, the only reason the liability's gone up is because the value's gone up." Well, is that... Does that mean that Anthropic issued shares today, and did they give, did they give Amazon and Google $108 billion worth of value, but did they give them those shares at the price-

swyx 1:16:52 Yeah, they, they, they just-

Jamin Ball 1:16:52 ... from four years ago? Did they just give them shares at a huge discount? It's like, no, that didn't happen. What happened is they raised that money years ago in normal financings, like a normal startup would. And over time, it's just appreciated in value.

swyx 1:17:04 Yeah. So, so, you know, I, I think a lot of people, if, uh, they, they're just not financially educated to that level-

Jamin Ball 1:17:09 Yeah

swyx 1:17:09 ... 'cause they've never had that exposure, so-

Jamin Ball 1:17:11 Yeah

swyx 1:17:11 ... it is what it is. It's you just educating. What, what was the second, uh, debate that you were referring to? You said you had two debates.

Jamin Ball 1:17:16 Oh, those related. Yeah, so it was, it was-

swyx 1:17:17 Oh, this is the other one

Jamin Ball 1:17:18 ... it was the balance sheet.

swyx 1:17:19 Oh, wow, okay.

Jamin Ball 1:17:19 Have they raised more debt? And then it was the income statement.

swyx 1:17:22 Yeah.

Jamin Ball 1:17:22 Are they actually losing money?

swyx 1:17:23 As an investor, why do you get involved in these [laughs] kinds of debates?

Jamin Ball 1:17:27 Oh, this is more of, like... I mean, I'm-

swyx 1:17:29 You're just having fun.

Jamin Ball 1:17:29 You know-

swyx 1:17:29 Okay, all right

Jamin Ball 1:17:30 ... yeah, I, I think Anthropic... Look, Al- Altimeter is a firm. We do venture investing, growth investing, and public investing.

swyx 1:17:36 Yeah.

Jamin Ball 1:17:36 And I spend a lot of my time on the venture investing, the early stage. However, I just, I consider myself, like, a student of finance, and Anthropic-

swyx 1:17:43 I know

Jamin Ball 1:17:43 ... is such an iconic company.

swyx 1:17:45 You shouldn't have to educate people on the three statements of accounting.

Jamin Ball 1:17:48 Yeah.

swyx 1:17:48 Like, that's like [laughs]

Jamin Ball 1:17:49 Yeah, yeah. But, but like, hey, I mean, I think a lot of people look at the balance sheet, or the income statement, the P&L.

swyx 1:17:54 Yeah.

Jamin Ball 1:17:54 But the reality is, is there's a lot to learn from the balance sheet and the cash flow statement, too. Um, and so this was more of a, hey, as, as just someone who's intellectually curious, I mean, Anthropic is, is a really important company to understand. I think if the Anthropic S-1 came out and the financial profile that was pictured was doomy and it didn't look like a good business, I think we have a lot to be worried about.

swyx 1:18:18 Yeah.

Jamin Ball 1:18:18 Um, because if they're not delivering profits and high ROI, what does that mean for everyone downstream of them? Like, that's, it's... I think it's really important to understand the Anthropic P&L. But what I understood from the Anthropic P&L, we... I, here, let me just rip through these really quickly. Um, here's what you're seeing about their gross margins. Lots of leverage in the gross margins. They're improving rapidly.

swyx 1:18:38 Which, uh, one, one nuance on this is this is, uh, the half that is not through cloud p- cloud partners.

Jamin Ball 1:18:43 Exa- yeah, uh, this, this... No, this is just the GAAP gross margin.

swyx 1:18:47 Right. Okay.

Jamin Ball 1:18:48 Is, is that... Okay. Is that-

swyx 1:18:49 I, I, I, I don't know. I, I read something about, like, how obviously cloud partner margin would not go-

Jamin Ball 1:18:54 So-

swyx 1:18:54 ... into gross margin at all.

Jamin Ball 1:18:55 Yeah. I, I-

swyx 1:18:55 And that's 50% of revenues

Jamin Ball 1:18:57 ... how they count, I believe, is a lot of the costs that they pay out the rev share is in sales and marketing-

swyx 1:19:02 Yeah, yeah

Jamin Ball 1:19:02 ... which is below the line, which is, which I think what you're saying.

swyx 1:19:04 Which, by the way, $3 billion-

Jamin Ball 1:19:05 Exactly

swyx 1:19:05 ... of sales and marketing. [laughs]

Jamin Ball 1:19:06 I think it's, like, 40-something percent of the revenue comes from cloud partners.

swyx 1:19:10 Yeah. [laughs]

Jamin Ball 1:19:11 It's, it's not a trivial amount.

swyx 1:19:12 48.

Jamin Ball 1:19:12 48? Okay.

swyx 1:19:13 Yeah, yeah.

Jamin Ball 1:19:13 Yeah, I was close.

swyx 1:19:14 Insane.

Jamin Ball 1:19:14 Yeah. So you see gross margin profile improving, and, and sure, maybe it'd be lower if, uh, you kinda factored in the third party, you know, the, the cloud providers, but, like, the trend is going up. You see operating margin going up, positive GAAP operating margin. You see fri- free cash flow margin Going up. And so to me, what I see is a highly profitable business that is probably going to flip to reinvest in compute and build-out. I think the way compute gets build, um, is gonna look different in the future. It's gonna be expensive. There's gonna be more upfront payments, right? Like, I do think you should expect to see the profitability profile of these labs, um, kind of go the other way in the short term as they shift to investing in growth and investing in compute. Uh, but the biggest takeaway for me was it's a highly profitable business that's still operating in growth mode, that's seeing some re-acceleration probably, you know, in the third-party data from the new models, which gave me a lot more confidence for the rest of the industry.

swyx 1:20:10 Yeah.

Jamin Ball 1:20:10 So long as the revenues show up, the compute obligations will be hit, um, the supply chains will, you know, continue to be saturated and constrained, and companies will continue to perform. And OpenAI and Anthropic, the reality is, is they're a huge percentage of the overall AI revenues, and so everyone looks to them as a proxy for the rest of the industry. So if they keep growing, which again, I think a lot of the bears would say they're gonna stop growing because open weight models are gonna eat their lunch, like this, this, and that. Um, but I think those are the important kind of questions.

swyx 1:20:40 Yeah. Very important, uh, very good questions. I, I think, like, yeah, anyway, uh, do you, do you dabble in, like, forward projections of numbers? Like, what, what, like we, we, what, what should we expect from FY27? What is the-

Jamin Ball 1:20:52 Oh, I think it's too hard

swyx 1:20:52 ... the street expect? Okay. All right.

Jamin Ball 1:20:53 Uh, yeah, and you, if, yeah, I, I think that's, that's too hard.

swyx 1:20:56 Because the, the, the delta is very high.

Jamin Ball 1:20:59 Yeah.

swyx 1:20:59 Right? Like, given the year they just had.

Jamin Ball 1:21:01 It's, it's a business that is so hard to project because the future is so, like, unknown and unknowable. Um, you know, they'll go public, the sell side research analysts at all the banks will, will build their models. We'll kind of have a forward, a forward number. I don't know where that's gonna shake out. Um, but I, I, I do think you'll, you know, you'll, you'll see, um, a lot of optimism, um, about what the future of this business could look like. But again, it's would it surprise me if it was much higher than expected or some... You know, anything could happen. It's anyone trying to predict the future, at le- especially in the short term-

swyx 1:21:38 Mm-hmm

Jamin Ball 1:21:38 ... it's, it's as, it's as hard as ever. I think businesses at the frontier are the most predictable because they're the least likely to be disrupted. But, um, I don't know. We live, we live in a highly uncertain time.

swyx 1:21:49 Yeah. The, the kind of A+ financial analyst a- as a PM, uh, uh, uh, in my hedge fund I would have wanted is to do physical limits-

Jamin Ball 1:21:59 Uh-huh

swyx 1:21:59 ... uh, calculations-

Jamin Ball 1:22:00 Mm-hmm

swyx 1:22:00 ... on, like, okay, you have all these inference optimizations, w- where, uh, like how much juice is there left-

Jamin Ball 1:22:05 Yeah

swyx 1:22:05 ... on, on the current thing?

Jamin Ball 1:22:06 Yeah.

swyx 1:22:07 Um, let's, let's say the next gen has, has this level of inference. How much can we charge on that?

Jamin Ball 1:22:11 Mm-hmm.

swyx 1:22:11 And, uh, given the competitive profile and e- exhibited, um, pricing history of, of these things.

Jamin Ball 1:22:16 Yeah.

swyx 1:22:17 Uh, and then finally, how much GPU supply-

Jamin Ball 1:22:20 Yeah

swyx 1:22:20 ... uh, has been locked in and, and secured so that that translates to this number of tokens, and that's what it can sell for. 'Cause that is, like, the physical limits-

Jamin Ball 1:22:26 Mm-hmm

swyx 1:22:26 ... of what, where these things can go.

Jamin Ball 1:22:28 Mm-hmm.

swyx 1:22:28 Because I think a lot of the financial analysts, and I used to do this, you, you would just kind of have this spreadsheet of, like, here's the past performance-

Jamin Ball 1:22:35 Yeah

swyx 1:22:35 ... and you click, click the right corner and just drag it out.

Jamin Ball 1:22:37 Yeah. Exactly.

swyx 1:22:38 That's not how it works.

Jamin Ball 1:22:39 Can't do that. I mean-

swyx 1:22:39 That's not good ana- that's not good analysis, right?

Jamin Ball 1:22:41 Totally. I mean, it's... I think there's limitations everywhere.

swyx 1:22:44 Yeah.

Jamin Ball 1:22:44 Everyone knows, like, "Hey, Nvidia can only produce so many chips." You know?

swyx 1:22:47 Yes.

Jamin Ball 1:22:48 Challengers, Cerebras TPUs can only produce so many chips. But I think there's other constraints. There's only so much land power shell, um, that's approved and permitted.

swyx 1:22:56 Correct.

Jamin Ball 1:22:56 Um-

swyx 1:22:56 So you do that build up, right?

Jamin Ball 1:22:57 And at the same time-

swyx 1:22:58 And then you're like, "Okay, who has what share?"

Jamin Ball 1:23:00 When we were... We, we talked a little bit about, like, debt and credit markets. Like, right now, you wanna go raise multi-billion dollar debt for GPUs? No bid. No bid. They're... Like, debt markets are closed. They're, you know, big debt deals that aren't happening, right? Almost the equivalent of, like, an IPO getting pulled. Big debt deals aren't happening.

swyx 1:23:17 So why is there 100 billion in equity but no 100 billion in debt?

Jamin Ball 1:23:20 Well, I think right now-

swyx 1:23:21 'Cause I go, I'm gonna turn around and do those, put those-

Jamin Ball 1:23:24 It's-

swyx 1:23:24 ... money right in the data center

Jamin Ball 1:23:25 ... the credit funds, they're all looking at the look-through exposure, and any neoCloud or a hyperscaler that's raising debt is the biggest question the credit providers are gonna say is, "Well, who's the offtake?" Because the reality is, is that's actually who I'm underwriting-

swyx 1:23:39 Yeah

Jamin Ball 1:23:39 ... because that's the customer. That's the one paying. That's the end payer of the bill, and if the end payer of the bill does poorly, the bills don't get paid. And right now, the end payer of the bills are Anthropic and OpenAI. So many of the data centers and the, the, the, the being built by the neoClouds are for OpenAI and Anthropic, and a lot of the credit providers are saying, "I'm feeling kind of full. You know? I just did a ton, I just did a ton of debt, um, for either one of these." And didn't matter if it was CoreWeave or Nebius or an upstart like Avolta or someone else. It's like, "Hey, I'm feeling full." And I think debt markets just are gonna go through a little bit of indigestion. They need to figure out, "Well, do I feel good? Do I feel great? Hey, some of these projects that were promised, are they being built on time? Is the revenue flowing back to me?" And I do think the credit and debt markets will ebb and flow, um, because there are just a thousand reasons why, you know, a project could get delayed, right? A neoCloud could go sign a contract with OpenAI. They could sign with, uh, Nvidia. They could go have Dell package them up. They could raise the debt. And then, you know what? There could be local community uprising-

swyx 1:24:47 Mm-hmm

Jamin Ball 1:24:48 ... and picketing. I'm lit- I'm not even being kidding. And then it's delayed. Then what? Well, the debt was raised. Someone's gotta pay the interest. What happens?

swyx 1:24:58 Yeah.

Jamin Ball 1:24:58 You know? And you have creative delayed draw facilities where you don't, you know, you, the debt is committed, but you don't draw it down. But, like, all of these have limits, you know? Like, and you get super in the weeds on, like, SLAs and, and who was responsible for it. Was it a delay in the supply chain? Was it a delay in the construction, you know, the data center developer? Was it a delay by the operator? Was it, you know, something else like-

swyx 1:25:22 Yeah

Jamin Ball 1:25:23 ... community unrest? And, you know, that's all, that's everything that, you know, credit markets are, are thinking through.

swyx 1:25:29 Yeah.

Jamin Ball 1:25:30 And then there's a lot of other super technical reasons, you know, that we don't need to get into that, you know, make it difficult with credit default swaps and hedges going through the roof. And so now it's for debt providers, it's tough to hedge the exposure they have. And so as, as the hedges get more expensive, your willingness to take on more risk goes down because you can't hedge out of it. Like it's, it gets super complicated.

swyx 1:25:47 Th- this is all the function-- This is all symptoms of the debt market running out of money.

Jamin Ball 1:25:51 Yeah. Well, I don't, I don't know if... Debt markets are super deep.

swyx 1:25:53 Yeah, yeah.

Jamin Ball 1:25:53 I think it's more of like an appetite.

swyx 1:25:55 Yeah.

Jamin Ball 1:25:55 Um, and of course, there's a price for everything. Hey, great. You know what? Go take out $5 billion of debt at 20%.

swyx 1:26:00 But you don't see this-- You see this at equity layer and, and like the-- So like, like again, how come there's $100 billion of equity-

Jamin Ball 1:26:06 Yeah

swyx 1:26:06 ... that's gonna go into Anthropic and, and Anthropic cannot raise $100 billion-

Jamin Ball 1:26:10 Well, I think-

swyx 1:26:10 ... or $50 billion for it?

Jamin Ball 1:26:11 I, I think, um, I would-- I don't know if I would say they can't do it.

swyx 1:26:15 Yeah.

Jamin Ball 1:26:15 Um, I don't know if-- Well, one, they, they need to be rated. Um, do they want to?

swyx 1:26:21 Yeah.

Jamin Ball 1:26:21 You know, my g- my guess is Anthropic and OpenAI are happy that NeoClouds exist to take the balance sheet risk. Maybe they don't want to take that balance sheet risk, 'cause again, it could be things-

swyx 1:26:31 Exactly

Jamin Ball 1:26:31 ... out of their control-

swyx 1:26:32 Okay

Jamin Ball 1:26:32 ... that blow them up. You have a huge debt stack in your cap, in your cap table, and things out of your control go wrong-

swyx 1:26:39 Someone else's problem, yeah

Jamin Ball 1:26:40 ... like you don't, you're not the one holding the bag. And again, this is like the beauty and the curse of, of why NeoClouds can be great businesses or they could be brutal businesses if the contracts aren't airtight, if things aren't structured well, your balance sheet business, you know. And, and just like fintech balance sheet businesses can ebb and flow, you know, there's a, there's a ton of... Oops, my computer may have died.

swyx 1:27:04 It's fine. We'll edit.

Jamin Ball 1:27:04 Uh.

swyx 1:27:04 We'll edit the screen.

Jamin Ball 1:27:05 Okay. You know, balance sheet businesses, right? Like there were a ton of fintech companies in 2020 that did super well when interest rates were low, that got blown up when interest rates were high. And balance sheet businesses can be really tricky and very cyclical, and you don't wanna own a balance sheet business when a market cycle tips over. So, hey, on, at, on some level, I'm sure, you know, especially for internal chip efforts, right, like the OpenAI team will-- might- may wanna develop their own data centers for their Jalapeno chips, right? And Anthropic, same thing for their chips. Um, but I'm sure they're also very happy for other people to take on that risk, um, because it's someone else's core competency, and, you know, a lot of debt could add a lot of blow-up risk. Um, and there was no debt on, on the, uh, Anthropic balance sheet. Now, they have facilities in place that they could draw down, and so would it surprise me if you start to see debt on their balance sheet in the future? No. Um, I mean, but again, it's like, you wanna build a gigawatt, right? $50 billion, $60 billion, $70 billion. That's, that's not a [chuckles] that's not a trivial amount just for... And that's just, just one gigawatt. Um, I think both OpenAI and Anthropic have said they kinda wanna get to five by the end of the year. Who knows if it's, if it's that or something more or less than that.

swyx 1:28:17 Are they about equal?

Jamin Ball 1:28:19 I don't know.

swyx 1:28:20 Okay.

Jamin Ball 1:28:20 Um-

swyx 1:28:21 That's interesting to me-

Jamin Ball 1:28:23 Yeah

swyx 1:28:23 ... 'cause I, I think Anthropic, uh-

Jamin Ball 1:28:24 OpenAI was ahead.

swyx 1:28:25 Correct.

Jamin Ball 1:28:25 Yeah.

swyx 1:28:26 Right, by, by a lot. Um, and, and-

Jamin Ball 1:28:27 Yeah, exactly

swyx 1:28:27 ... but I think Anthropic's closed the gap a lot. Uh, and OpenAI at, at least started, like Stargate was a plan, the first stage in a plan to get to 30-

Jamin Ball 1:28:34 Mm-hmm, mm-hmm

swyx 1:28:35 ... 30 gigawatts.

Jamin Ball 1:28:35 Mm-hmm, mm-hmm.

swyx 1:28:36 And so-

Jamin Ball 1:28:36 Yeah, by 2030.

swyx 1:28:37 Yeah. So and so I was like, okay, well that's-- I, I haven't heard that plan from Anthropic.

Jamin Ball 1:28:40 But again, like let's, let's h- let's do some fun math and really quick on the fly. Let's see if I can do it.

swyx 1:28:44 Ooh.

Jamin Ball 1:28:44 Getting tired. Um, OpenAI, let's say, I think when they came out with that, it was maybe a year ago, and they were what, like one and a half gigawatts, something like that. So whatever, round that down to zero for just the sake of the argument. Let's say OpenAI and Anthropic, each one are-- go from 0 to 30 over five years. So you're gonna add 60 gigawatts of AI capacity, just the two labs. What's a good assumption for their total capacity? I don't know, 40%, 50%. Um, let's say it's 40%. And I don't know, now we're getting too tough in math in my head, but, you know, let's say that gets you to, I don't know, 110, 120 gigawatts of total capacity brought online over the next five years. Let's say 100, just so the math is easier.

swyx 1:29:22 Yeah.

Jamin Ball 1:29:22 And let's say it costs $50 billion to build out a gigawatt. That's $5 trillion.

swyx 1:29:27 [chuckles]

Jamin Ball 1:29:28 $5 trillion over the next five years, right? And so call it a trillion a year. You're, you're talking about a non-trivial percent of GDP, you know, real GDP. That's like 5% plus or minus of GDP. Like, that is a massive-

swyx 1:29:50 Bonkers

Jamin Ball 1:29:50 ... massive build-out. You know, rivals the railroad build-out. You know, I, I, I think I've seen some charts somewhere that kind of compare those expected numbers, but like fi... You know, when you're measuring things in like whole number percents of GDP, uh, you're-- that's, that's huge. And so like the credit markets, debt markets need to be super, super, super deep and super, super open for that to happen. You need global build-out. Um, you know, it's kinda funny. I, you know, you know, SemiAnalysis number for next year is like 50, and you know, I think Brad at the All-In Summit said, "Oh, maybe it's lower." And I think there was some trolling going on back and forth. And I think the end result of that was like, hey, maybe the 50 gigawatt number, I don't wanna speak for anyone, was more of like what's constructed. You know, Brad was more of what's constructed and live generating tokens, because there can be a gap, um, in what's constructed and ready to go versus what's actually live and generating tokens. But I think it's like a real debate in the world, you know? And then you could see a pretty wide range of outcomes. You know, do we really add 50 gigawatts of capacity next year as an industry, um, or is it 20 or 30? You know, like who knows? I just know everyone you talk to in the supply chain is saying, "You know, we, we have, we have orders or indications for orders that are well over 100% of our capacity."

swyx 1:31:06 Yeah.

Jamin Ball 1:31:07 And like I don't have-

swyx 1:31:07 That's true

Jamin Ball 1:31:07 ... line of sight to that slowing down for the next few years.

swyx 1:31:10 Yeah.

Jamin Ball 1:31:11 And-

swyx 1:31:12 Yeah, and, and the limit like land, shell, power, all, all these-

Jamin Ball 1:31:15 Yeah

swyx 1:31:15 ... but also silicon, like straight up-

Jamin Ball 1:31:16 Totally

swyx 1:31:17 ... like compute memory. Every-

Jamin Ball 1:31:18 Totally

swyx 1:31:18 ... every part of the supply chain-

Jamin Ball 1:31:19 Totally

swyx 1:31:20 ... the-- that's the work. That's why SemiAnalysis is so powerful right now because-

Jamin Ball 1:31:22 I mean, look at these recent-

swyx 1:31:23 It is

Jamin Ball 1:31:23 ... Kimi models, Kimi K3. You know how much memory you need gigabytes of memory just to serve that? You're not running that on your laptop.

swyx 1:31:29 Yeah.

Jamin Ball 1:31:29 You know, you need like a rack to serve one model.

swyx 1:31:32 Yeah.

Jamin Ball 1:31:32 Um, you know, again, that's, that's, those are just the big models, right? Like you're also gonna have tons of small models, um- But I know the big models are getting bigger, which only require more memory, um, and more compute to run

swyx 1:31:43 I used to say they were capping out. I used to say-

Jamin Ball 1:31:46 Mm

swyx 1:31:46 ... like the max size is one to two, and then, like we would just optimize from there.

Jamin Ball 1:31:49 Yeah.

swyx 1:31:49 I've been wrong. [laughs]

Jamin Ball 1:31:50 What's, I mean, what's interesting is it does feel like the open weight models are capping more around, like, 100 billion active parameters-

swyx 1:31:56 Yes, that's right. Yeah

Jamin Ball 1:31:56 ... and-

swyx 1:31:57 That's mostly a, like, an inference speed thing.

Jamin Ball 1:31:58 Totally.

swyx 1:31:59 But, like, I think-

Jamin Ball 1:31:59 Totally

swyx 1:31:59 ... uh, it's actually really interestingly tied to the generations of GPUs that they have.

Jamin Ball 1:32:03 Yeah. Yeah, exactly.

swyx 1:32:04 Right? Because, like, you- you- you-

Jamin Ball 1:32:05 Exactly

swyx 1:32:05 ... like wait for the Vera Rubins to, to roll out-

Jamin Ball 1:32:07 Mm-hmm

swyx 1:32:07 ... and then whatever China has.

Jamin Ball 1:32:09 Yeah, exactly.

swyx 1:32:10 And, and, well, then, then they'll get it, you know?

Jamin Ball 1:32:12 Yeah. Totally.

swyx 1:32:12 Like, uh, then so it- it- they- they're really memory constrained at the inference layer, and that actually shapes the model architecture-

Jamin Ball 1:32:17 Totally

swyx 1:32:17 ... which is, like, very kind of cool.

Jamin Ball 1:32:18 Totally.

swyx 1:32:19 Um, oh, so that is Anthropic.

Jamin Ball 1:32:21 Yes.

swyx 1:32:22 Uh, powerhouse tour.

Jamin Ball 1:32:23 Yeah.

swyx 1:32:23 Congrats. Uh, I don't know how you felt coming into this, but you, you did great.

Jamin Ball 1:32:27 Oh, this was fun.

swyx 1:32:27 Uh-

Jamin Ball 1:32:28 Yeah

swyx 1:32:28 ... and, like, you know, obviously more to do and more to analyze, but I just definitely wanted on the podcast to do this-

Jamin Ball 1:32:32 Yeah

swyx 1:32:32 ... because I, I know you.

Jamin Ball 1:32:34 Yeah, of course.

swyx 1:32:35 And-

Jamin Ball 1:32:35 No, it's been fun

swyx 1:32:36 ... and, like, I think when else were we-

Jamin Ball 1:32:37 We've been meaning to do this for a while

swyx 1:32:38 ... when, when else were we gonna get this in a pod?

Jamin Ball 1:32:39 Glad we could finally make it happen.

swyx 1:32:39 Yeah,

swyx 1:32:40 yeah. Exactly.

Jamin Ball 1:32:40 Yeah.

swyx 1:32:40 Exactly. Um, any other commentary? When you don't spend time on, you know, the labs, what are you, what are you spending time on?

Jamin Ball 1:32:45 Ooh, man. It's, um, again, in a world that's highly unknown and unknowable and uncertain, I kind of fall back to what are my core beliefs that I do have conviction in, and that, that is the world's gonna consume more models, world's gonna consume more power, world's gonna consume more compute, networking, and storage, and we're gonna need to build a lot more data centers to serve the intelligence we're building. Those are kind of like my six core truths, and so, like, a lot of what I'm looking for are what are businesses that are a direct derivative of one of those, one of those core beliefs that I have.

swyx 1:33:13 Yeah. And any, any other expressions of those core beliefs that you've, uh, done recently?

Jamin Ball 1:33:17 Done recently? Um, yeah, certainly. Um, there's a s- chip company still in stealth that's super exciting, maybe we can talk about later.

swyx 1:33:25 Okay.

Jamin Ball 1:33:26 Um, there's, you know, a power company, American Terawatt. They want to build a new grid, an industrial grid, to serve industrial offtake. Big, huge, bold vision. Um, you know, we talked about Alorian and Glimpse and kind of the visual thinking-

swyx 1:33:39 Visual reasoning

Jamin Ball 1:33:39 ... visual reasoning.

swyx 1:33:39 Anything on the compute market side? Uh, so effectively, like, slicing out compute, uh, in order to increase the efficiency.

Jamin Ball 1:33:46 Uh, more, more of like the-

swyx 1:33:46 Trading futures. Trading futures on compute, all that stuff

Jamin Ball 1:33:49 ... oh, I haven't, I ha- I haven't spent as much time there.

swyx 1:33:51 Yeah.

Jamin Ball 1:33:51 Uh, I haven't spent as much time there.

swyx 1:33:53 Uh, you know, um, we spent some time with Ansh, who's-

Jamin Ball 1:33:56 Okay. Yeah, yeah, yeah, yeah

swyx 1:33:56 ... doing National Compute and, and AMP Grid and-

Jamin Ball 1:33:58 Yeah

swyx 1:33:58 ... and SF Compute and all those, those guys. Uh, in, in New York we launched a compute trading track where-

Jamin Ball 1:34:03 That's right

swyx 1:34:03 ... where, where, like, mostly it's just to increase utilization-

Jamin Ball 1:34:06 Yeah

swyx 1:34:06 ... is the broader goal.

Jamin Ball 1:34:07 Yeah.

swyx 1:34:07 But that's how you can magically create a whole bunch more compute, which is better use the ones-

Jamin Ball 1:34:12 Yeah

swyx 1:34:12 ... we already have.

Jamin Ball 1:34:13 Totally.

swyx 1:34:13 Uh, and so if, you know, if the MFU, whatever choice of utilization is at 30 to 40% right now, taking it up to 60%, you've doubled-

Jamin Ball 1:34:22 Yeah

swyx 1:34:22 ... the number of GPUs you have.

Jamin Ball 1:34:23 Yeah, totally.

swyx 1:34:24 And so I think it's-

Jamin Ball 1:34:25 I mean, I think there's gonna be so much cool innovation. It's why, like, this is the most exciting time to be in tech, to be investing, to be around all these startups, 'cause, like, you're just... You're seeing innovation at the silicon layer. You're seeing innovation at the capital markets layer. You're seeing innovation at the product layer. I mean, like, it's like every-

swyx 1:34:42 Yeah

Jamin Ball 1:34:42 ... creative destruction everywhere.

swyx 1:34:43 I, I notice you're not doing much on the application layer.

Jamin Ball 1:34:46 You know, look, I, I would say-

swyx 1:34:47 The, the agent layer. Yeah

Jamin Ball 1:34:48 ... no, cer- cer- certainly are. Again, like I, I, I... I mean, I'm, I'm, I'm speaking more personally-

swyx 1:34:53 Yeah

Jamin Ball 1:34:53 ... than necessarily on the, the Altimeter behalf. Um, want to believe there's a strong product research org combo. Um, there's certainly a lot of great teams, and we think, you know, early on in market cycles, you know, we kind of use these, like, triangle analogies where you have, um, compute or infra, like, like hardware infra, um, software infra, software apps. And in a mature market, the biggest, the biggest TAM is the apps, and, you know, the smallest TAM is the, the hardware. Um, early on in cycles it's inverted, and, you know, the biggest markets are at the, at the, at the both top or bottom of the triangle, more of the, the physical hardware and then the infrastructure and then the apps. And, like, eventually the triangles invert, and so, hey, at some point in the future, I think the apps and the agents will be the biggest part of the TAM. We're just early on, and there's lots of opportunity in-

swyx 1:35:49 Yep

Jamin Ball 1:35:49 ... the other parts of-

swyx 1:35:50 Ob- objectively, that is correct.

Jamin Ball 1:35:51 Yeah.

swyx 1:35:52 I, I, I do feel like... And, and, and by the way, you can see various versions of this in all of computer history.

Jamin Ball 1:35:56 Yeah.

swyx 1:35:57 Like-

Jamin Ball 1:35:57 Yeah

swyx 1:35:57 ... right down to, like, the, uh, uh, Western Micro, West... Like, the, like, when hard disks were-

Jamin Ball 1:36:03 Yeah

swyx 1:36:03 ... were, were getting built out or whatever. Anyway, so I, I, I don't think those are the best examples. Western Digital?

Jamin Ball 1:36:08 Western Digital.

swyx 1:36:09 Right?

Jamin Ball 1:36:09 Yeah, yeah, yeah.

swyx 1:36:10 Um, anyway, so, um, they'll, uh, they're... Yeah, I, I, I think that is true, but, like, to me, domain specificity is very interesting-

Jamin Ball 1:36:17 Mm-hmm

swyx 1:36:18 ... and to me, the app layer, the agent layer is supposed to capture that.

Jamin Ball 1:36:21 Yeah. Totally.

swyx 1:36:21 They, they're just like the biggest outcome here is Cursor-

Jamin Ball 1:36:24 Yeah

swyx 1:36:24 ... right now.

Jamin Ball 1:36:24 Yeah, yeah.

swyx 1:36:25 Uh, and that 60 billion-

Jamin Ball 1:36:26 Yeah

swyx 1:36:26 ... we're here talking about 2 trillion, I think.

Jamin Ball 1:36:27 Yeah, exactly.

swyx 1:36:28 Very, very different. [laughs]

Jamin Ball 1:36:29 Yeah, totally. Totally. I think I also just, like, skew a little bit more infra than, yeah, you know, just personal bias.

swyx 1:36:33 Yeah, yeah. We,

swyx 1:36:34 we're, we're infra guys at heart.

Jamin Ball 1:36:35 Yeah, exactly.

swyx 1:36:36 Uh, okay. Great. Um, my closing question was gonna be favorite, um, Brad story. I mean, Brad's a character.

Jamin Ball 1:36:43 Mm. Yeah.

swyx 1:36:44 Uh, and-

Jamin Ball 1:36:44 Yeah

swyx 1:36:44 ... we've, we've worked with him for, like, I don't know, three, four years now.

Jamin Ball 1:36:46 Six years.

swyx 1:36:47 Six years.

Jamin Ball 1:36:48 Yeah.

swyx 1:36:48 Okay, so, so Clouded Judgement ends, but, okay. All right. This is-

Jamin Ball 1:36:50 Right around... Yeah, yeah.

swyx 1:36:51 I, I feel like I'm getting my time- either my timeline is wrong, or I met you way earlier than I thought I met you.

Jamin Ball 1:36:56 No, I think we met earlier-

swyx 1:36:57 [laughs]

Jamin Ball 1:36:57 ... 'cause I, I was, I would think I was still at Redpoint when we first met.

swyx 1:36:59 Oh my God. Okay. All right. All right.

Jamin Ball 1:37:00 Yeah. You, where were you? Netlify? No.

swyx 1:37:02 Uh, Airbyte. Uh-

Jamin Ball 1:37:03 No, no. Before that.

swyx 1:37:04 Before that was Temporal.

Jamin Ball 1:37:05 Before that.

swyx 1:37:06 Netlify.

Jamin Ball 1:37:07 That's when we met.

swyx 1:37:08 Really?

Jamin Ball 1:37:08 Yeah.

swyx 1:37:09 Jesus.

Jamin Ball 1:37:09 Yeah.

swyx 1:37:09 That's, that's ancient history.

Jamin Ball 1:37:11 Yeah.

swyx 1:37:12 Uh, anyway, Brad's a character. I, I just think he's f- he's interesting, and-

Jamin Ball 1:37:16 Yeah

swyx 1:37:16 ... like, what's, what's, what- so what's a good story you have?

Jamin Ball 1:37:18 Oh. Oh man, you put me on the spot.

swyx 1:37:20 Yeah. [laughs]

Jamin Ball 1:37:21 Um, ho- hon- honestly, I would say favorite Brad story, just how much he's worked at Invest America. Like, and I don't know if people know-

swyx 1:37:32 Yeah

Jamin Ball 1:37:32 ... this initiative. It is the idea of, like, how can we get people invested in capitalism early on in their life? I think, like, right now in the world, right, I mean, you survey You know America, you know most people have a negative view of capitalism because they haven't participated in it.

swyx 1:37:47 Mm.

Jamin Ball 1:37:47 And like a lot of people view capitalism as a great thing, and it's the people who've participated in the equity upside of America and have kind of compounded, um, into capitalism. But most people don't have that luxury. And so how can we educate people with first-hand experience about the benefits of capitalism by helping them participate early on? So the whole idea is an investment account funded by the US government at birth that can be funded by anyone along the way. That could be philanthropists like, like Michael Dell, um, who made a $6 billion dollar, uh, donation. I think Elon Musk has pledged one share to every kid in America, which is-

swyx 1:38:21 Yeah

Jamin Ball 1:38:21 ... I think $10 billion. Uh-

swyx 1:38:23 Anthropic has some version of like the public benefit fund or something.

Jamin Ball 1:38:26 Hey, e- exactly. And so I, I think just seeing him work on that behind the scenes for years, um, and kind of seeing it come to fruition all very quickly. You know, it's the classic overnight success. And so there's not like one specific story, it's more of just seeing the grind that goes on.

swyx 1:38:42 It's a cause and he believes in. Yeah.

Jamin Ball 1:38:42 Yeah, ex- exactly.

swyx 1:38:43 And, uh, is it done by the way? I actually don't know the-

Jamin Ball 1:38:45 So it's passed. Yeah, it's passed into law.

swyx 1:38:46 It's passed, so it's, so what?

Jamin Ball 1:38:47 Um-

swyx 1:38:47 Like if I have a kid now, they, they get it?

Jamin Ball 1:38:49 You could open an account. Yeah.

swyx 1:38:50 Okay.

Jamin Ball 1:38:50 You open an account. Um-

swyx 1:38:52 Yeah

Jamin Ball 1:38:52 ... any new kid born gets $1,000 at birth. Um-

swyx 1:38:54 Yeah

Jamin Ball 1:38:54 ... parents can contribute to it, employers can contribute to it. Um, friends can contribute to it.

swyx 1:38:59 So it's like a, it's like a baby 401 [k] .

Jamin Ball 1:39:00 Exactly. [laughs]

swyx 1:39:01 [laughs]

Jamin Ball 1:39:01 Exactly. And you know, you have people like Michael Dell who say, "Hey, here's my donation, and I want it to be gifted to..." I can't remember the exact specifics, but I think it was like, "Hey, bottom 75%, uh, income by ZIP Code." And the money just-

swyx 1:39:14 That's better. Yeah

Jamin Ball 1:39:14 ... ex- ex- so you c-

swyx 1:39:15 Yeah

Jamin Ball 1:39:15 ... exactly. The money just... 'Cause you know-

swyx 1:39:17 It's like-

Jamin Ball 1:39:17 ... as, as Brad would say, "Look, what's w- 529 plans," which are kind of education plans. Like, hey, I can fund money that can be used to, for my kid's education one day. Like the reality is, is the people using 529 plans are generally like the wealthiest part of the country. Like 529 plans and the tax advantage wasn't made for the wealthy part of the country, but that's who uses it the most. Uh, what he wanted to do was, "Hey, let me create investment accounts for the people who actually need them." Um, and I think it's, you know, it's an awesome initiative that's just getting started. I think you're gonna see a lot more direct philanthropy. What's a big issue people have with philanthropy is, well, you're gonna donate a bunch of money, and then middlemen are gonna take half of it.

swyx 1:39:58 I, I have seen-

Jamin Ball 1:39:58 And there's-

swyx 1:39:58 ... literally directly this

Jamin Ball 1:39:59 ... tons of loss

swyx 1:40:00 ... it's, it's used for corruption, like straight up.

Jamin Ball 1:40:01 Ex- exactly. But what's better? Direct contributions. Direct to bank account of people who need it most. No middleman. Direct accounts that then people can see, appreciate, and value over time. Can't touch it till you're 18 years old. Um, and so you get to see the appreciation happen over time, maybe learn and, and participate in, in the upside of America along the way. Um, and I think it's, it's, it's really like it, it's caught on, and just seeing how passionate he is about it, um, has been kind of inspiring to see.

swyx 1:40:32 Okay. My one pushback is like I'm, you know, homeless on the streets.

Jamin Ball 1:40:36 Yeah.

swyx 1:40:37 Food stamps, everything.

Jamin Ball 1:40:38 Yeah. Yeah.

swyx 1:40:39 I have $1,000 sitting there.

Jamin Ball 1:40:40 Yeah. Yeah.

swyx 1:40:42 You then like, "Well, give it to me."

Jamin Ball 1:40:44 Yeah.

swyx 1:40:44 Like [laughs]

Jamin Ball 1:40:45 So the, yeah, the... You... Well, it, it's made, it's made more for kids now.

swyx 1:40:49 Yeah.

Jamin Ball 1:40:49 And so hopefully you're not... Well, maybe there, there are plenty of homeless kids.

swyx 1:40:51 Yeah, yeah, exactly. Right?

Jamin Ball 1:40:52 Um-

swyx 1:40:52 Like so, so like that, there's the thing like it's like, well, you know, did you know you have a bank account and you're participating-

Jamin Ball 1:40:57 So that's, that's-

swyx 1:40:57 ... in the growth of capitalism

Jamin Ball 1:40:58 ... that is the biggest thing.

swyx 1:40:59 I don't care, you know? Like [laughs]

Jamin Ball 1:41:01 So that, that's the biggest thing now which is, um, people need to claim their accounts. The government opened up accounts for every kid in America.

swyx 1:41:07 Yeah.

Jamin Ball 1:41:08 And parents need to claim them.

swyx 1:41:09 Yeah, yeah.

Jamin Ball 1:41:09 That it, it is sitting there. Now, the money's sitting there for newborns.

swyx 1:41:13 Yeah.

Jamin Ball 1:41:13 Um, you know, if you're 15 years old, money doesn't show up in your account. I think donations will, and so you could be a 15-year-old that does have a $250 donation from Michael Dell sitting in your account. Um, but they need to be claimed. And so like now the, I would say the focus has gone from, let's get this passed into law. Let's get it up and running. Let's get the infrastructure set up so people can donate. Now it's let's go convert those-

swyx 1:41:37 Yeah, this is execution

Jamin Ball 1:41:38 ... created accounts-

swyx 1:41:39 Yeah

Jamin Ball 1:41:39 ... to claimed accounts.

swyx 1:41:39 Yeah.

Jamin Ball 1:41:39 And that's just a, that's like a grassroots word of mouth game.

swyx 1:41:44 Yeah.

Jamin Ball 1:41:44 Because there are people whose accounts are sitting there with money in it-

swyx 1:41:46 Yeah

Jamin Ball 1:41:46 ... and it just needs to be claimed.

swyx 1:41:47 And you didn't know. Yeah, yeah.

Jamin Ball 1:41:48 Yeah.

swyx 1:41:49 Yeah. And by, by the way, most people have money sitting with the government that has not been unclaimed. You can look it up.

Jamin Ball 1:41:53 Yeah, unclaimed property.

swyx 1:41:54 Unclaimed.

Jamin Ball 1:41:54 It's like a great use case-

swyx 1:41:55 Yeah, yeah

Jamin Ball 1:41:55 ... for like a Muse and Instinct and-

swyx 1:41:57 Have a, have an agent go look up your name.

Jamin Ball 1:41:58 An agent, exactly.

swyx 1:41:59 Uh, you probably have a few thousand dollars sitting somewhere that someone returned to you, and it's just sitting there.

Jamin Ball 1:42:02 You just need to know your address.

swyx 1:42:03 Yeah.

Jamin Ball 1:42:03 Your prior addresses.

swyx 1:42:04 Yeah, yeah.

Jamin Ball 1:42:04 And then go have Instinct or Muse-

swyx 1:42:07 Yeah, it's highest ROI on this podcast is that.

Jamin Ball 1:42:08 Exactly.

swyx 1:42:08 Not, not the Anthropic stuff.

Jamin Ball 1:42:09 Exac- exactly. Exactly.

swyx 1:42:11 Um, cool. Thank you so much for just spending your time with us.

Jamin Ball 1:42:13 Yeah, of course.

swyx 1:42:14 Uh, this was great. Uh, we'll have you back. Uh, we'll see how the IPO goes. [laughs]

Jamin Ball 1:42:17 This was fun. Thanks for doing it.

swyx 1:42:19 Thank you. [outro music]