Big companies love to overpay for star talent. But even by Silicon Valley standards, this deal was on another level.
Mark Zuckerberg reportedly offered a single AI researcher a pay package worth up to $1.5 billion over six years. That works out to roughly $250 million a year — for one person, not a whole executive team.
The researcher took the deal. Then he walked away less than a year later.
Here’s the full story of Andrew Tulloch, the man who turned down, then accepted, then left one of the biggest pay packages in corporate history.
Corporate America Has a Long History of Overpaying for Talent
Paying huge money to poach a star from a rival isn’t new. Companies have done it for decades, and it rarely works out the way they hope.
In 2011, JCPenney handed Ron Johnson nearly $53 million in stock to leave Apple, where he had spent 11 years building the Apple Store into a retail powerhouse. Johnson also gave up about $80 million in unvested Apple equity to take the job. He lasted just 17 months. JCPenney’s sales collapsed, its stock lost half its value, and the company eventually filed for bankruptcy in 2020. The Apple stock Johnson walked away from would be worth close to $2 billion today.
A year later, Yahoo lured Marissa Mayer away from Google to become CEO. She earned around $405 million over five years, but the turnaround never happened. Yahoo’s core business was eventually sold to Verizon for about $4.5 billion.
Back in 1995, Disney paid Michael Ovitz roughly $1 million a year plus stock options worth about $110 million to leave the talent agency he’d co-founded. He lasted 14 months and left with a $109 million payout, sparking years of shareholder lawsuits.
Not every mega-deal blows up, though. Google paid Ruth Porat around $70 million to leave Morgan Stanley in 2015. She’s still at the company today. And in 2017, Uber paid Dara Khosrowshahi roughly $200 million to leave Expedia and take over as CEO after Travis Kalanick’s exit.
Those were massive deals. But they were all for CEOs running entire companies.
What Facebook did next was different. It offered nine-figure money to a single engineer most people had never heard of.
Meet the $1.5 Billion Man
Andrew Tulloch isn’t a household name, but inside the small world of elite AI researchers, he’s been a legend for years.
Tulloch is Australian. He studied math at the University of Sydney and later at Cambridge. He spent 11 years at Facebook before leaving in 2023 to join OpenAI. In early 2025, he helped Mira Murati, OpenAI’s former chief technology officer, launch a new AI startup called Thinking Machines Lab.
Here’s a story that shows just how highly regarded he’s always been.
Back in 2016, OpenAI was a tiny research lab trying to build its founding team. OpenAI president Greg Brockman tried to recruit Tulloch and even discussed it directly with co-founder Elon Musk. At the time, Tulloch was earning about $800,000 a year at Facebook. OpenAI could only offer around $300,000 total. Tulloch passed and stayed put.
Nine years later, $800,000 would look like pocket change next to what Facebook was about to offer him.
The $1.5 Billion Offer
By 2025, the biggest tech companies weren’t just racing to build the best AI models. They were fighting over the handful of researchers who might actually be able to build them. Facebook — officially renamed Meta in 2021, though most people still call it Facebook — was spending enormous sums trying to catch up.
Zuckerberg wanted Tulloch back badly.
According to The Wall Street Journal, Meta first tried to buy Thinking Machines Lab outright. When that failed, it turned its attention directly to Tulloch and reportedly offered him a pay package that could be worth up to $1.5 billion over at least six years, depending on bonuses and stock performance.
Spread evenly, that’s about $250 million a year. Break that down further, and it comes to roughly $961,500 per workday, or around $120,000 an hour — close to $2,000 every single minute on the clock.
Meta pushed back hard on the number. A company spokesperson called the Journal’s description of the offer “inaccurate and ridiculous.”
Tulloch initially said no. He stayed at Thinking Machines Lab.
He Changed His Mind — For Less Money
A few months later, in October 2025, Tulloch reversed course and agreed to rejoin Meta. People close to the deal say the package he actually accepted was smaller than both his original offer and the $1.5 billion figure that made headlines. The exact terms were never made public.
Whatever the number was, it was enough to pull a co-founder away from one of the hottest AI startups in the world — a company that had raised $2 billion just months earlier and was valued at around $12 billion.
Tulloch joined Meta’s TBD Lab, a small, high-priority research group sitting at the center of Zuckerberg’s push to make Meta a leader in artificial intelligence. The lab was run by Alexandr Wang, the young billionaire who co-founded Scale AI and was brought in to help lead Meta’s AI overhaul.
Then, 11 Months Later, He Left
In September 2026, Tulloch told colleagues he was leaving Meta. His second run at the company had lasted roughly 11 months.
He reportedly delayed his exit on purpose, waiting until Meta finished rolling out its new family of open-source AI models and launched its AI assistant, Muse. Once that shipped, he made his move.
Neither Tulloch nor Meta has explained exactly why he left. A Thinking Machines Lab spokesperson later said he had “decided to pursue a different path for personal reasons.” It’s still unclear where he’s headed next, or whether he plans to keep doing AI research at all.
His exit also isn’t an isolated case. It’s part of a wider wave of departures from Meta Superintelligence Labs, the division Zuckerberg built to lead the company’s AI comeback.
Why This Story Matters
Tulloch’s story says a lot about where the AI industry is right now. Companies aren’t just competing over products anymore — they’re competing over people, and the price tags attached to top researchers have gotten genuinely absurd.
Even a disputed offer in the hundreds of millions of dollars wasn’t enough to keep one researcher at a company for a full year. Whether that means the money didn’t matter as much as everyone assumed, or the work environment mattered more, is something only Tulloch really knows.
One thing is clear: in the current AI talent war, even a nine-figure paycheck doesn’t guarantee loyalty.
