Top AI talent commands huge stakes

Andrej Karpathy’s move to Anthropic this week and Ilya Sutskever’s disclosure that his OpenAI stake is worth about $7 billion have put unusual numbers on a familiar Silicon Valley reality: a small group of AI researchers now carries compensation packages that look closer to founder-scale wealth than ordinary employee pay. The figures surfaced as OpenAI prepares for a possible public offering that Reuters reported could come as soon as the second half of 2026 and value the company at up to $1 trillion. At the same time, companies including Meta, Anthropic, OpenAI and Safe Superintelligence have been competing for a limited pool of researchers with experience building frontier models. The result is a labor market in which one hire, one defection or one retained founder can move billions of dollars in implied value. ### Why are a few AI researchers suddenly worth so much? Euronews reported on May 19 that top AI researchers are drawing million-dollar salaries and, in some cases, equity stakes worth far more, as major labs and large technology companies compete for scarce technical talent. The article tied that escalation to a narrow group of people with direct experience in large language models, scaling laws, pretraining and model safety. (cnbc.com) Ilya Sutskever provided the clearest public example when he testified in federal court this month that his OpenAI stake is worth roughly $7 billion. That testimony, reported by multiple outlets and cited by Euronews, made concrete how much value early technical contributors can still hold even after leaving a company. ### What does Karpathy’s move to Anthropic show? (euronews.com) Andrej Karpathy said on May 19 that he is joining Anthropic, returning to frontier-model research after earlier roles at OpenAI and Tesla. Anthropic said he would work on the pretraining team and help build a group using Claude to accelerate pretraining research. CNBC reported that Karpathy wrote on X that “the next few years at the frontier of LLMs will be especially formative” and that he was excited to “get back to R&D.” His move adds another prominent OpenAI alumnus to a rival lab that has been expanding aggressively. (thenextweb.com) ### Where does Meta fit into this bidding war? Meta explored talks involving Safe Superintelligence, the startup co-founded by Sutskever after he left OpenAI, according to reporting cited by Euronews and later accounts of Meta’s recruitment push. (cnbc.com) Safe Superintelligence was reported in 2025 to have reached a $32 billion valuation despite having no public product, showing how investors were pricing founder reputation and technical credibility as much as current revenue. That matters because Safe Superintelligence is not just another startup. Sutskever was OpenAI’s former chief scientist, and his standing in the field has made his new company both a recruiting target and a benchmark for how much strategic value one researcher can carry into a new venture. ### Why does OpenAI’s possible IPO sharpen the focus on talent? (euronews.com) Reuters reported in October 2025 that OpenAI was laying the groundwork for an initial public offering that could value it at up to $1 trillion, and a Reuters-backed report this week said the company is preparing to confidentially file in the coming weeks. A public filing would force closer scrutiny of how much of OpenAI’s value rests on its researchers, its computing arrangements and its corporate structure. (euronews.com) OpenAI told investors in a document viewed by CNBC, and later described by Reuters, that its ties to Microsoft could pose a business risk because Microsoft provides “a substantial portion” of its financing and compute. OpenAI said separately that Microsoft “will remain a critical long-term partner.” (money.usnews.com) ### What should readers watch next? The next concrete milestone is OpenAI’s expected IPO filing window in the coming weeks, which Reuters reported on May 20. That filing, if it appears, would give investors their first detailed look at the company’s finances, risk factors, partner dependence and ownership structure. (economictimes.indiatimes.com) Anthropic’s integration of Karpathy will also be watched closely because the company said he starts this week on pretraining work. In a market where a single researcher can carry a multibillion-dollar signal, hiring announcements are no longer just personnel news. (cnbc.com) (msn.com)