# Frontier labs need a repeatable model factory, not one winning model

Markdown copy of https://realalpha.app/s/frontier-labs-need-a-repeatable-model-factory-not-one/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

- Speaker: [Jamin Ball](https://realalpha.app/people/jamin-ball/) · Jamin Ball is a partner at Altimeter and author of Clouded Judgement. · Invests in enterprise software and AI companies, and writes Clouded Judgement on software metrics.
- Host: [Shawn Wang (swyx)](https://realalpha.app/people/swyx/) · Co-founder, AI Engineer · Wrote “The Rise of the AI Engineer” and founded Smol AI, maker of AI News.
- Show: Latent Space
- Episode: [Why Would Anthropic Go Public Now? — Jamin Ball, Altimeter](https://realalpha.app/v/P735f1xlpL0/), Nov 7, 2026
- Moment: 51:58 to 57:00 of the episode (5 min), playing on https://realalpha.app/s/frontier-labs-need-a-repeatable-model-factory-not-one/; the episode from there: https://realalpha.app/v/P735f1xlpL0/?t=3118

## Quote

> “the question isn't, can you build one model? It's, can you **attract compute and talent** required”
> — [Jamin Ball](https://realalpha.app/people/jamin-ball/) at [54:31](https://realalpha.app/v/P735f1xlpL0/?t=3271) · Latent Space

## Takeaways

1. The advantage is recurring model development, not one standout model.
2. Frontier labs need compute and talent to ship new generations.
3. Ball says Anthropic's compute spend doubled as revenue grew 14-fold.

## About this segment

Jamin Ball compares frontier labs to a chip factory: one strong model is not enough without the capacity to make the next one. He argues Anthropic's revenue has grown much faster than its training compute.

From “Why Would Anthropic Go Public Now? — Jamin Ball, Altimeter”, Nov 7, 2026. [Watch the full episode](https://realalpha.app/v/P735f1xlpL0/)

## Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

[52:03](https://realalpha.app/v/P735f1xlpL0/?t=3123) people have said is that the r&d per model let's say like what this what does fable five sonnet five plus five cost let's call it a billion dollars that lasted me six months right because you just said the depreciation it lasted that six months yeah yeah and so isn't that worrying right like about like well these things are the fastest depreciating assets in history yeah this is something that me and sarah go have talked about before yeah yeah yeah yeah um and i think the the net is there's no but like it's a qualified no of well as long as my next model run is better than the current state of the art, we're good.

[52:44](https://realalpha.app/v/P735f1xlpL0/?t=3164) Yeah, yeah. But it's a little bit of a house of cards if the next one doesn't work out. Yeah. Which briefly affected OpenAI. Yeah. So a lot of people know by now that, you know, like GPT-4-5, the big run wasn't that successful. Yeah. And they've since fixed it. But like that, a slump in a run, a failed run, can really, really screw with the ROI on the thing. 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 and, you know, in the 80s and 90s, about all these chip companies.

[53:22](https://realalpha.app/v/P735f1xlpL0/?t=3202) They've seen it all. And, you know, they've seen it all. And they were highly skeptical. And, you know, sometimes you have industry vets that are highly skeptical. And, you know, they miss the next big opportunity because they never think 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 chips that compete with, you know, the chip factory.

[53:51](https://realalpha.app/v/P735f1xlpL0/?t=3231) And so he was like, this individual's point of view was maybe a chip startup can build one chip. But guess what? They're not going to have the machine to constantly pump out a new chip every year and stay at the frontier. from the chip standpoint. And that's why NVIDIA is king. It's not because they have the best chip now, it's they have the best factory. Process for chips. Exactly. It's the machine that builds the machine.

[54:11](https://realalpha.app/v/P735f1xlpL0/?t=3251) And so it's kind of one of these, I mean, this is the bear case on all these NeoLabs, which is, you know, and you talk to many smart folks and what they might tell you is, hey, if you want to really compete and have like a SOTA model at the frontier, you can't 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.

[54:33](https://realalpha.app/v/P735f1xlpL0/?t=3273) And so as a Neolab, 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 or models or like new types of models, decision models like Jev.

[55:10](https://realalpha.app/v/P735f1xlpL0/?t=3310) but I think the big labs, they are on the treadmill of continuously pumping out new models because guess what? If Anthropic stops, OpenAI and Gemini will continue and then you'll just kind of 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.

[55:43](https://realalpha.app/v/P735f1xlpL0/?t=3343) That could be because they're doing more economically valuable work or they're doing the work faster. But what you're seeing from now kind of black and white is, hey, as Anthropic has 2x'd their compute spend to build and train these models, they've 14x'd their revenue. And I'm not really sure what would cause that trend to stop. I think models are only going to take on longer running tasks, more economically valuable tasks.

[56:15](https://realalpha.app/v/P735f1xlpL0/?t=3375) And if you can't tell, I'm a hyper optimistic person. And so maybe I'm just drinking the Kool-Aid too much. But 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, much cheaper models, you know, and kind of span everything from frontier to, you know, distilled versions of their own models that are smaller in and, you know, cents per million input output tokens, not dollars.
