Google's Chief Scientist Jeff Dean Leaves to Launch AI Startup Discovery Loop

Jeff Dean, the chief scientist at Google DeepMind, is leaving the company to launch a new AI startup called Discovery Loop, along with three other top-tier Google AI researchers. Having worked at Google for the past 27 years, Dean is one of the most influential executives at the tech giant and is widely revered by employees, as per multiple news reports. He is credited with writing the code for several of Google’s key projects involving search and computing, with his frequent collaborator, Sanjay Ghemawat, who is among the four co-founders of Discovery Loop. Besides Dean and Ghemawat, who were reportedly among Google’s first hires, the other two co-founders of Discovery Loop are Oriol Vinyals, VP of research at DeepMind and a technical lead for Gemini, and Quoc Le, a DeepMind scientist who led Google’s AutoML-Zero project that uses machine learning algorithms to autonomously build AI products. Discovery Loop will be focused on using AI to power superhuman advances in a variety of fields, including chip design, biology, drug discovery, and material design. The exits of four of top Google AI brains came along with another announcement by Google on Wednesday, August 5, that Demis Hassabis will be stepping down from the role of DeepMind CEO. Hassabis has been elevated to chair of the AI unit and chief scientist of Google’s parent company, Alphabet. Announcing Discovery Loop! I am very excited to announce that, along with my longtime friends and collaborators @Sanjay_Ghemawat , @OriolVinyalsML and @quocleix , we are founding Discovery Loop ( @DiscoLoopAI ), a Public Benefit Corporation whose mission is to automate machine… pic.twitter.com/ancoplyNvN — Jeff Dean (@JeffDean) August 5, 2026 The leadership shakeup at Google DeepMind comes at a time when the tech giant is looking to keep pace with rivals such as OpenAI and Anthropic in the frantic AI model race. Google stock fell 4 per cent following Wednesday’s announcement. Discovery Loop’s founding mission echoes efforts that are already underway across the tech industry. Several prominent AI figures such as Dario Amodei, Sam Altman, and Jensen Huang have previously claimed that AI will become an engine of scientific discovery by solving problems that humans could never crack on their own. However, the decades of experience between Discovery Loop’s founders may help the team build an AI system that actually delivers such breakthroughs. “We were working on different things, but we were all starting to see the possibility of AI being able to automate scientific and engineering loops,” Jeff Dean was quoted as saying by Wired . Automating machine learning could lead to the discovery of new AI architectures beyond present-day transformer models, as per the group. “As we create the core intelligence and the discovery loop fundamentals, we’ll be able to quickly generalize to other domains. One of the things that we’ll be obviously very focused on is how these models come up with new ideas to try. That’s not something that currently they’re super strong at,” Vinyals, another Discovery Loop cofounder, said. I also want to give a huge thanks to the incomparable @JeffDean after an incredible 27-year run at Google. He’s off to start his own public benefit corporation with @Sanjay_Ghemawat focused on accelerating discoveries across ML, science, & engineering. @Google will support as a… — Sundar Pichai (@sundarpichai) August 5, 2026 Commenting on the exits, Google CEO Sundar Pichai said in a statement, “Over 27 years, Jeff and Sanjay helped to drive some of the most significant technology transitions, from our early search infrastructure to the neural networks that helped create the modern AI era.” “We’ll continue to work with Discovery Loop as a founding investor and Cloud partner, and collaborate on a research framework for ML systems and related infrastructure advances,” Pichai added. Based on an initial roadmap laid down by the startup, Discovery Loop is setting out to be its own first customer as it aims to develop autonomous experiment loops that will be directed towards improving the company’s advanced machine learning algorithms. The improved software will then become the core of an AI system that can tackle problems in subsequent domains. The startup is betting on a future where a few dedicated individuals may use Discovery Loop to out-invent the world’s largest research labs and organisations. It may also design specific loops to seek and monetise breakthroughs. Discovery Loop is currently set up as a public benefit corporation, with Google having a stake in the startup. Google will also provide the startup with compute power for one year as part of the arrangement. Khosla Ventures and an AI-oriented VC firm called Radical Ventures have also invested in Discovery Loop, along with a handful of other firms. The exact details related to how much funding and at what valuation are unclear. Dean is considered to be a legendary engineer at Google. He is the co-founder of Google Brain, and was later appointed as the chief scientist of Google DeepMind and Google Research. Dean also served as technical co-lead of the company’s flagship AI model Gemini. In a town hall with over 6,000 would-be founders at Y Combinator’s Startup School at San Francisco’s Chase Center held last month, Dean was asked what areas startup founders should think about tackling. He replied that the most important thing would be to pick something they were excited about and would be useful in the world. “You propose an experiment, you implement what you need to run the experiment, you evaluate the experiment, and then you get results from that,” he said. Demis Hassabis, who is a Nobel Prize winner, stepping down as DeepMind CEO is not significant considering that he was reportedly not the one running the frontier AI lab on a day-to-day basis anyway. That job was carried out by Koray Kavukcuoglu, DeepMind’s chief technology officer (CTO), who is now formally taking the reins of Google DeepMind. Hassabis will continue to run Isomorphic Labs, Alphabet’s drug discovery unit.