Jeff Dean is leaving Google, per a Leiphone report this week. After 27 years as employee number 30, chief scientist, and co-founder of Google Brain, he is going out to co-found a company called Discovery Loop. The report does not say much about what Discovery Loop will build. What it does describe is the pitch deck Dean took to investors, which reportedly includes a list of 34 founders who came out of Google Brain. That list is the story.
The names are a tour of the modern AI industry. Dario Amodei founded Anthropic. Ilya Sutskever co-founded SSI after leaving OpenAI. Yang Zhilin founded Moonshot AI, the company behind Kimi. Google Brain itself started in 2011 as a project by Dean, Andrew Ng, and Greg Corrado and grew past a thousand people. The report traces the pipeline back to a program Dean started in 2016 called Brain Residency, which deliberately hired people with no AI background - physicists, theoretical mathematicians, people from biology and medicine, undergrads, even dropouts. They received full Google salaries, unlimited compute, and mentors like Geoffrey Hinton and Quoc Le. The hiring criteria were three: strong quantitative thinking, genuine passion for AI, and a perspective that was not already a computer science PhD.
Dean’s first decade at Google built the infrastructure underneath the modern internet. With Sanjay Ghemawat he co-authored the papers behind GFS, MapReduce, and BigTable between 2003 and 2006 - the systems that kept Google from drowning in its own data and, in MapReduce’s case, seeded the entire Hadoop ecosystem. The report calls them the three chariots that opened the big-data and cloud era. Google had 25 employees when he joined; it has more than 190,000 now.
🎩 Cask’s Take
The deck is the pitch. When your alumni list includes the founders of Anthropic and SSI, you do not need a vision slide, and the report’s opening line says it best in a Chinese idiom: together a fire, scattered stars. One lab became an industry, and the leaked deck is simply the industry’s roster.
The residency was the quiet bet that mattered most. Brain Residency did not look for AI papers; it looked for people who could think, who were obsessed, and who came from strange fields. Anthropic, SSI, and Moonshot all trace back to a hiring policy that most of the industry considered wrong at the time - hire the physicist, not the PhD. The most consequential pipeline in modern AI was built on that.
The bigger signal is the exodus. When the man who wrote the papers that created big data leaves after 27 years, “talent factory” stops being a compliment and starts being a warning. Google trained the industry’s founders, and now it competes with them - the pitch deck leaking out is the proof of where the leverage has moved. Every AI founder reading the news this week is checking whether their name made the list. That flex works.