• AI Weekly
  • Posts
  • The $12 billion A.I startup that couldn't keep its co-founder for a year

The $12 billion A.I startup that couldn't keep its co-founder for a year

Author

Joshua Clear
October 19, 2025

In partnership with

The Tech newsletter for Engineers who want to stay ahead

Tech moves fast, but you're still playing catch-up?

That's exactly why 100K+ engineers working at Google, Meta, and Apple read The Code twice a week.

Here's what you get:

  • Curated tech news that shapes your career - Filtered from thousands of sources so you know what's coming 6 months early.

  • Practical resources you can use immediately - Real tutorials and tools that solve actual engineering problems.

  • Research papers and insights decoded - We break down complex tech so you understand what matters.

All delivered twice a week in just 2 short emails.

The Billion-Dollar Ghost

Listen, I need to tell you about Andrew Tulloch because this story is absolutely wild and reveals something deeply broken about how AI startups work right now.

Here's what happened: In February 2025, Tulloch co-founded Thinking Machines Lab with Mira Murati (OpenAI's former CTO). They raised $2 billion—billion with a B—at a $12 billion valuation. For context, they had literally zero product at the time. Just vibes and résumés.

By August, Meta came knocking. Zuckerberg offered Tulloch up to $1.5 billion over six years to jump ship. And here's the kicker—Tulloch said no. He declined a nine-figure payday to stick with his startup's mission. His LinkedIn went viral. People were calling it a feel-good story about choosing purpose over profit.

Then in October—just eight months after founding the company, barely weeks after they launched their first product—Tulloch left anyway to join Meta.

What the Hell Just Happened?

The timing is what gets me. Thinking Machines had just shipped Tinker, their API for fine-tuning large language models. Investors had literally just wired $2 billion into this thing. And one of the two co-founders peaces out for "personal reasons."

Cold comfort for those investors, as Axios dryly noted.

Now, we don't know what Meta ultimately paid him. Maybe they matched the original offer. Maybe they found some other angle. What we do know is that Zuckerberg wanted Tulloch badly enough to personally recruit him twice, and Meta's been throwing around $100 million bonuses to land top AI talent.

Why This Matters (And Why You Should Be Worried)

This isn't just tech gossip. It exposes three uncomfortable truths about AI startups right now:

First, valuations are completely divorced from reality. A company with no product gets valued at $12 billion purely because the right people are in the room. That's not investing—that's paying ransom to keep talent away from Big Tech.

Second, there's no real moat here. You can't patent expertise. You can't lock people in with golden handcuffs when Meta's offering literal billions. The entire value proposition walks out the door the moment someone accepts a better offer.

Third, the talent pool is absurdly small. There might be only a few hundred people on Earth who can actually build frontier AI systems at the highest level. We're not talking about thousands of qualified engineers—we're talking about the people who trained GPT-4, who built PyTorch, who understand distributed training at massive scale. Tulloch was one of them.

The Real Product

Here's what I think is actually happening: These AI startups aren't building products. They're building acqui-hire vehicles with extra steps.

Think about it. Murati and Tulloch could've just joined Meta or Google or Microsoft directly. Instead, they raised $2 billion, operated for eight months, and then... one of them goes to Meta anyway. The VCs paid a massive premium for the privilege of maybe keeping these people around a bit longer.

It's like those key person clauses in insurance policies, except the premium is $2 billion and the policy doesn't actually prevent anything.

Where This Goes

Venture capitalists are now scrambling to figure out founder vesting schedules and "key man" protections. Good luck with that. When someone can make generational wealth by switching jobs, no vesting schedule matters.

The thing is, I actually don't blame Tulloch. If someone offers you enough money to fund small nations, and you've already built what you set out to build, and your personal circumstances change—why wouldn't you take it?

But for the AI startup ecosystem? This is a canary in a coal mine. If you can't keep co-founders around for a single year, what exactly are you building? And more importantly—what are you really paying for?

Sources

Reply

or to participate.