Andrew Tulloch, co-founder of Thinking Machines Labs, has left to join Meta, marking another twist in the ongoing AI talent wars. Just months ago, reports suggested Tulloch turned down a $1.5B compensation package from Meta — one of the largest in tech history. His departure follows Thinking Machines’ record-breaking seed round and the launch of its first product, a model-tuning toolkit that some market observers viewed as underwhelming.

A new article from Thinking Machines Labs may hint at where the record-setting startup is heading. The piece explores why LLMs remain nondeterministic even with temperature set to zero tracing the issue to floating-point math across multiple cores. With reports that 95% of AI pilots fail, Thinking Machines seems to be signaling that unpredictability is more than a quirk it’s a core obstacle to scaling AI reliably.

The AI sector has entered an era of unprecedented valuations — with Thinking Machines raising the largest seed round in history and nearly half of all venture capital now flowing into AI. As M.G. Siegler noted on spyglass, OpenAI leads with a $500B secondary valuation, while Anthropic, xAI, and others are pushing past the $100B mark. But as the race tilts from technology to market share and data access, smaller players remain crucial, driving innovation and attracting the att

Meta’s pursuit of AI dominance hit a dramatic note last week with reports that CEO Mark Zuckerberg offered a staggering $1.5 billion compensation package to Thinking Machines Labs co-founder Andrew Tulloch—shortly after the company reportedly declined an acquisition offer. While Meta has denied the claims, the incident underscores the current reality: top-tier AI talent is now the most valuable asset in tech. In a twist of “Airony,” the industry viewed as a threat to human jo