Leaked list alleges Meta assembled $300M plan to recruit ~40 top AI engineers from OpenAI, Apple and others

The fight here is not really about résumés. It is about a tiny number of people who can move a frontier model, a data pipeline, or a training stack fast enough to matter. That is why the viral claim about Meta building a roughly $300 million target list for about 40 elite AI engineers got traction so quickly. The exact list itself is not publicly verified. But the broader picture behind it is very real — Meta spent much of 2025 raiding rival labs, and the known compensation numbers were already wild. (bloomberg.com) ### What is the confirmed part? Meta did launch a new superintelligence effort in June 2025 after frustration with its AI progress. Zuckerberg personally got involved in recruiting, and the company began pulling in researchers and leaders from OpenAI, Google DeepMind, Scale AI, Apple, and others. That part is not rumor anymore — it showed up across multiple later hires and departures. (bloomberg.com) ### Where does the $300 million claim come from? The $300 million figure appears to come from a social-media post describing an internal-style recruiting list for around 40 people. I could not independently verify that list from primary reporting. So the safe read is this: the leaked figure is an allegation, not(bloomberg.com)ers and engineering leaders. (techcrunch.com) ### Why are people treating it as plausible? Because the public numbers are already enormous. Sam Altman said in June 2025 that Meta had been making “giant offers” to OpenAI staff, including talk of $100 million signing bonuses and even more in annual compensation. Meta pushed back on the idea that these we(techcrunch.com)performance conditions. But nobody really disputed the core point that the offers were unusually large. (techcrunch.com) ### Why does Ruoming Pang matter so much? Pang is the cleanest concrete example because he was not just any engineer. He ran Apple’s foundation-models team — the group behind the models feeding Apple Intelligence and other on-device AI work. Bloomberg later said Meta landed him with a package worth more tha(techcrunch.com)re like an extension of an existing playbook. (bloomberg.com) ### Was OpenAI actually hit? Yes. Meta hired multiple notable OpenAI researchers in June 2025, including Jiahui Yu, Shuchao Bi, Shengjia Zhao, Hongyu Ren, and Trapit Bansal. That matters because these were not generic “AI employees.” They were people tied to core model research and reasoning work — the part of the stack that is hardest to replace quickly. (bloomberg.com) ### Why are these people worth so much? Because frontier AI is weirdly bottlenecked. A company can buy GPUs, lease data centers, and license data. But there are far fewer people who know how to make giant training runs work, improve post-training, tune reasoning behavior, or build efficient on-device systems. (bloomberg.com) That last part is an inference from the hiring behavior and pay structure, but it is the logic the market keeps signaling. (bloomberg.com) ### Is this just about Meta versus OpenAI? Not really. Apple got hit too, and later reports showed additional Apple AI departures to Meta after Pang left. That suggests the pressure was not a one-off poach but part of a broader drain from companies whose AI efforts looked slower or less decisive. Once one big package clears, the next candidate’s price goes up. (bloomberg.com) ### What is the bottom line? The leaked list may or may not be authentic in its exact form. But the market it describes is already here. Meta showed that a top AI lab will spend like a hedge fund for a tiny circle of researchers — and now every rival has to decide whether to match, overpay, or watch key people walk.