Summary: On June 17, 2025, Sam Altman confirmed on a podcast that Meta offered OpenAI employees $100 million signing bonuses and even higher annual compensation. He said none of OpenAI’s best people had accepted, while reporting also noted Meta hired senior researchers such as Jack Rae and Johan Schalkwyk. The exchange became a reference point for the 2025 AI talent war and the limits of cash-based recruitment.
On June 17, 2025, Sam Altman confirmed on the Uncapped podcast — hosted by his brother Jack Altman — that Meta had offered OpenAI employees $100 million signing bonuses, with annual compensation above even that figure. It was the first on-the-record confirmation of the numbers behind Meta’s recruitment spree, straight from the OpenAI side.
His next line was the story. Altman said he was glad that “none of our best people have decided to take him up on that.” An unprecedented pile of money, and the researchers it targeted simply said no. The exchange put the central tension of the mid-2025 AI talent war on the record: what money can and cannot buy in a market obsessed with AGI. Both TechCrunch and The Verge covered the episode the day it dropped.
What Altman actually said
Describing Meta’s approach, Altman was blunt: “$100 million signing bonuses, more than that in compensation per year.” Then came the punchline: “none of our best people have decided to take him up on that.” The two lines became headline material within hours and turned the episode into a standard citation for every later story about the talent war.
Altman paired the rejection with a dig. Meta was poaching with cash, he argued, because “they’re not a company that’s great at innovation,” and a team assembled on pay rather than mission would struggle to build a great culture. His own staff, he said, believed OpenAI remained the better bet on reaching AGI — framing retention as a mission question, not a compensation one.
Meta’s poaching scoreboard
Reading Altman’s remarks as a total failure, however, overstates the case. TechCrunch’s reporting noted that Meta’s attempts to take OpenAI veteran Noam Brown and Google’s Koray Kavukcuoglu came up short, but the company did land Jack Rae from Google DeepMind and Johan Schalkwyk from Sesame AI, both joining the superintelligence effort led by Alexandr Wang.
The realistic scoreboard: zero headline stars, but a meaningful haul of senior researchers from adjacent labs. For Meta, the spree was a continued attempt to rebuild research capacity after its Scale AI investment. For OpenAI, the inner circle held — which, given the numbers involved, counted as a genuine defensive win.
Why the money didn’t move them
Altman’s own explanation leaned on mission and culture: his staff believed OpenAI had the better shot at AGI, and people who moved for pay would find their incentives hard to align with a long-term goal. It was, of course, a winner’s narrative — but it pointed at a real retention problem. In an industry where compensation had already inflated past the point of differentiation, adding another zero to the check bought very little at the top.
Observers at the time argued that elite researchers were actually calculating compute access, influence over the research roadmap, publication freedom, and whether the company would back their technical direction. A $100 million check buys acceptance; it does not reliably buy any of those. That Meta’s offers still missed their targets said the market’s pricing unit was no longer cash.
Context: the Scale deal and the counterpunch
Placed on a timeline, the remarks came after Meta’s massive Scale AI investment and a drumbeat of hiring reports — a public chapter in Meta’s effort to rebuild its AI bench. Altman did not stay defensive in the same episode: TechCrunch noted that OpenAI was then rumored to be preparing an open model release, and possibly building a social app to rival Meta on its own turf.
The back-and-forth kept escalating over the following weeks, but the June 17 conversation left a clear timestamp: before the superintelligence arms race surfaced at full scale, the first exchange happened in the labor market — and both sides could plausibly claim gains.
What it signaled for AI hiring
The episode became shorthand for 2025’s compensation inflation: signing bonuses denominated in nine figures, declined by the people they were aimed at. The practical lesson for other labs was that retention at the top hinged on compute, research direction, and shipping authority — not incremental cash.
For ordinary engineers the spillover was real too. Labs backfilling around star hires kept pushing mid-level AI salaries higher through the second half of 2025. Looking back, the podcast episode is where the market’s pricing logic for that year became legible: mission and compute were the currency, and money was merely the entry ticket.