Jeff Dean exits Google with top researchers to start AI-powered science startup

Jeff Dean, a foundational figure at Google who joined the company in 1999 as its 30th employee, is departing to lead a new startup focused on automating scientific discovery through artificial intelligence.

Joining Dean at the helm of the venture are three other prominent researchers: Sanjay Ghemawat, a senior fellow and accomplished engineer; Quoc Le, a longtime AI specialist who helped establish Google Brain; and Oriol Vinyals, a research scientist from Google DeepMind. The group is founding Discovery Loop, structured as a public benefit corporation.

The Vision#

Discovery Loop aims to transform how scientists and engineers conduct research by deploying AI to run thousands of experiments in parallel, dramatically compressing timelines that have historically depended on sequential human effort. The startup intends to automate much of what has been a labor-intensive experimental process, potentially enabling researchers to conduct work at unprecedented scale and speed.

Beyond experiment automation, Discovery Loop is also exploring recursive self-improvement—using AI to help design and build more capable AI systems with minimal human intervention. As the founding team explained, progress in science and engineering has long been bottlenecked by the slow pace of human iteration. They argue that AI-driven automation can eliminate that constraint.

Market Timing#

Applying AI to accelerate research discovery has long been an aspiration in scientific and technology circles, yet it remained largely theoretical until recently. Discovery Loop is entering a moment when commercial viability appears achievable.

The company has attracted backing from Radical Ventures and Khosla Ventures, which are co-leading its initial fundraising round. Kleiner Perkins, Lightspeed, and Doerr Capital also joined the investment syndicate. Notably, Alphabet, Google's parent company, is among the financial backers.

Dean's Legacy at Google#

Over his 27-year tenure, Dean shaped much of Google's technological foundation. He contributed to the search engine's core infrastructure—crawling, indexing, and query serving systems—and later played a major role in advancing the company's AI and machine learning initiatives, including work on multimodal models like Gemini. His departure represents a significant loss of internal expertise.

"We think there is opportunity for AI to more fully automate what has traditionally been a very human-intensive experimental loop," Dean . "You will get both a higher quantity and a higher quality of experiments, and that will lead to scientific breakthroughs and advances."