Ilya Sutskever leaves OpenAI, Jakub Pachocki steps up as Chief Scientist

The man who was once Sam Altman’s closest intellectual sparring partner is walking out the door. Ilya Sutskever, OpenAI’s co-founder and until Tuesday its Chief Scientist, is leaving the company. Altman announced the departure in a note to staff that read almost like a eulogy for a working relationship that defined the company’s early years. “Ilya is easily one of the greatest minds of our generation, a guiding light of our field, and a dear friend,” Altman wrote. He added that Sutskever is leaving to work on “something personally meaningful,” though no details were given. It’s a clean but heavy break — the kind that leaves you wondering what wasn’t said.

Sutskever’s exit closes a particularly fraught chapter. He was a central figure in the November 2023 boardroom drama that briefly ousted Altman, only to reverse course days later under intense employee and investor pressure. Sutskever himself signed the letter demanding Altman’s return, a move he later expressed regret about. Since then, his public presence at the company has been practically invisible. No big research announcements, no keynotes, none of the speculative AGI philosophizing that once made him famous. His departure formalizes a rift that anyone paying attention could see had been widening for months.

Into that vacuum steps Jakub Pachocki, a name that carries far less public recognition but immense internal weight. Pachocki has been at OpenAI since 2017 and was the research lead behind GPT-4 and the OpenAI Five Dota 2 bot — projects that proved large-scale reinforcement learning could actually work outside a lab notebook. Altman called him “easily one of the greatest minds of our generation,” a phrase he also used for Sutskever. That’s not accidental. It’s a signal to the research community and to investors that the intellectual lineage remains unbroken, even if the original architect is gone.

Pachocki holds a PhD in theoretical computer science from Carnegie Mellon and has been quietly steering the company’s fundamental research toward scaling laws and optimization techniques. The question now is whether he can fill a role that was as much philosophical as it was technical. Sutskever shaped how OpenAI thought about AI safety, timelines, and the very definition of AGI. Pachocki inherits not just a title, but an unresolved tension between speed and caution that has defined OpenAI since its founding. What Sutskever plans to build next remains a mystery — but if history is any guide, it won’t be boring.