What if the most effective thing Meta ever did for Thinking Machines Lab was raid it?
That’s the uncomfortable question sitting at the center of one of the more fascinating talent stories in AI right now. Meta has reportedly poached seven of TML’s founding members — and yet, Thinking Machines Lab is not limping. It’s hiring. Aggressively. And the names it’s pulling in are serious.
A Revolving Door That Spins Both Ways
Here’s what actually happened: Meta held acquisition talks with Thinking Machines Lab around early 2025. Those talks didn’t close a deal — but they did apparently open a pipeline. Meta has since picked off TML founders one by one, including Soumith Chintala, who was appointed CTO of TML in early 2026 before that relationship evolved further.
On paper, that sounds like a slow bleed. In practice, TML has used the attention — and the departures — as a signal to the broader research community that this is a place worth watching. When a company like Meta is circling you, trying to buy you, then trying to absorb your people, other researchers notice. The message reads less like “TML is losing” and more like “TML is where the interesting work is happening.”
Why This Matters to Anyone Building Bots
If you’re building AI-powered bots — whether that’s conversational agents, autonomous task runners, or anything in between — the institutions shaping foundational research matter more than most people realize. The models you’ll be prompting, fine-tuning, or building on top of in two years are being designed by the people moving between these organizations right now.
Meta’s AI research has historically fed into open-weight models that the bot-building community has relied on heavily. LLaMA changed what was possible for developers who couldn’t afford closed API costs. But TML is positioning itself as a different kind of player — one that’s actively recruiting from Meta’s own talent pool while Meta is simultaneously trying to absorb TML’s founders.
That’s not a one-sided talent war. That’s a feedback loop.
The Talent War Is the Product Roadmap
In AI, where a team goes is often a better signal than any press release. When seven founders leave a startup for a big tech company, the usual read is “startup failed to hold its own.” But TML flipped that script by continuing to recruit researchers away from Meta even as Meta was picking off its founders.
Think about what that means structurally. TML is saying: we can absorb the loss of founding members and still attract top-tier researchers from the very company taking them. That’s a statement about culture, about research direction, and about what kind of work people actually want to do when they have options.
For those of us building on top of whatever these labs produce, that competitive tension is genuinely useful. It means more research output, more open experimentation, and more pressure on both sides to ship things that matter.
What to Watch Next
- TML’s research output over the next 12 months will tell you whether the new hires are producing or just settling in.
- Meta’s response matters too — if they keep pulling from TML, it signals they see something specific worth absorbing, not just general talent.
- The acquisition talks that didn’t close last year could easily resurface. A company that tried to buy you and failed has unfinished business.
The Bigger Picture for AI Builders
Talent concentration in AI is a real problem for the ecosystem. When the same handful of researchers cycle between five major labs, the diversity of ideas narrows even as the headcount grows. What makes the TML situation interesting is that it represents genuine friction — a smaller organization pushing back against absorption, using the pressure to attract people rather than fold under it.
As someone who spends most of my time thinking about how bots get built and what they’re built on, I find that friction healthy. The best tools I’ve used came out of labs that had something to prove. TML, whether by design or circumstance, now has plenty to prove.
Meta may have taken seven founders. But it also handed TML a story — and in the current AI space, a good story is one of the most solid recruiting tools you can have.
🕒 Published: