Andrew Tulloch Leaves Meta Less Than a Year After Reported $1.5 Billion Pay Package

Andrew Tulloch, the AI researcher Mark Zuckerberg reportedly lured away from Thinking Machines Lab with one of the largest individual pay packages in tech history, has left Meta after roughly ten months on the job. Semafor first reported the departure on September 9, noting that Tulloch had waited until the day after Meta shipped its Muse agentic assistant before making his exit official, and that neither he nor Meta offered a public explanation for the timing or the reason.

Tulloch's path to Meta was itself a story about the price of talent. He co-founded Thinking Machines Lab in 2025 with former OpenAI chief technology officer Mira Murati, then left that startup in October 2025 after the Wall Street Journal reported Zuckerberg had offered him a package potentially worth $1.5 billion over six years, following a failed attempt to acquire Thinking Machines outright. Meta initially rejected the framing of that report before Tulloch ultimately joined the company's TBD lab, the unit inside Meta Superintelligence Labs run by Scale AI founder Alexandr Wang.

"Inaccurate and ridiculous" — Andy Stone, Meta Spokesperson, disputing the Wall Street Journal's reporting on Tulloch's pay package

The reversal is notable set against what Thinking Machines Lab did after losing him: the startup went on to raise roughly $2 billion at a $12 billion valuation just weeks after his departure, evidence that the talent war of the past year has inflated valuations on both sides of these exits, for the labs bleeding researchers and the ones absorbing them.

Multiple outlets covering the Meta exit, including Semafor, TechCrunch and Gizmodo, converge on the same unresolved detail: where Tulloch goes next is unconfirmed. Unverified social-media claims point to Anthropic, but no named source at either company has corroborated that, and machine.net is not treating it as established fact.

What is established is the pattern: a researcher commands a package in the nine-figure-to-billion-dollar range, stays under a year, and moves again, all while the labs on either end of the trade use the moment to reset their own valuations. That undercuts the industry's working assumption that eye-popping compensation functions as retention. If a headline-grabbing multi-year package can't hold a researcher through even one full product cycle, the real currency in AI hiring may be less about lock-up terms and more about which lab looks best positioned to win at any given moment — a dynamic that rewards continuous poaching over long-term retention deals.