The full quote:

> When entrepreneurs walk into the offices of Andreessen Horowitz (a16z), a big American venture-capital firm, the odds these days are that their startups are using AI models made in China. “I’d say 80% chance [they are] using a Chinese open-source model,” says Martin Casado, a partner at a16z.

This is very different from what the author portrays. It may be the case that many pre-funded startups are using Chinese open-source models (somewhere in their workflow). But what percent of startups that survive more than a year (either with funding or revenue) are still doing this?

I imagine their pitch is: "look at how well we're doing using open source Chinese models! We'll do even better once we raise money to be able to afford frontier models!"

The way the author presents this quote makes me think he had a preferred narrative and found quotes to back it up. Or he's just a very uncareful reader.

Casado says he was misquoted and walked this back: https://x.com/martin_casado/status/1990462245541982546

"Well, not quite. I'd say 20-30% use open source. Of those I'd say 80% use Chinese based models. So closer to 16-24%."

Reads pretty differently.

I don't really think the author is misrepresenting the quote in any way, it just seems like many people are reading it as a hard statistic instead of a reference to a properly attributed quote.

Just anecdotally though, my company is not a startup, well established and well known and has already started investigating, purely for dev purposes (not product), using Chinese models - this was spurred by costs rising much faster than expected.

So while I agree that I don't think it's anywhere near the 80% level across the board - I wouldn't be surprised if it starts moving that way.

There may be confusion between 'product tech' and 'development tooling'. It's entirely plausible most of the startups that are A: building AI tech, B: early stage, and C: raising from top Sand Hill Road VCs, are currently prototyping their product tech concepts starting from open weight models.

A VC partner meeting with early-stage founders is focused on the viability, uniqueness and defensibility of the IP tech stack not what tooling the coders are using. The developers could be using Claude or GPT 5.6 to develop a tech stack based on open weight models.

Could it be that the startups that embed LLMs in a product will prefer Openweights for a more stable economic model, while SWE who use LLMs as a tool prefer SOTA to produce code?

Found the narrative, it was right there at the bottom. Now I remember reading some of his other stuff and it's very much in the same vein.

> I care about having open technology that can be run in the public interest, aligned with the public’s values. Threads like public AI, federated services, and open research have traction but need backing. Getting there in the US needs more nuanced strategy and support than we’re seeing today.

This quote sounds like it’s about companies building products on AI to serve a tweaked or harnessed version of that to others not the developers in those companies model of choice for coding.

There's probably some selection bias in that stat, too. Based purely on vibes, I wouldn't be surprised if startups pitching to a16z are using open Chinese models slightly more often than startups pitching to other VCs.

It's also important to note that several PE firms have signed contracts with model trainers to specifically use their model in their owned companies. I know Anthropic signed a deal worth hundreds of millions to acquire users just a few months ago.

Yep, and following that with

> and Chinese models are poised to take the lead.

makes it sound like the second part is a continuation of the first quote from the same source, but actually the second link is just some random person’s substack post from almost a year ago.