Google chief scientist Jeff Dean leaves after 27 years

### Google AI Pioneer Jeff Dean Leaves Company to Launch New Research Venture

Jeff Dean, one of Google’s most influential technology leaders and a key figure in the company’s development of large-scale computing and artificial intelligence, is leaving the technology giant after nearly three decades to launch a new AI-focused venture. Dean is establishing Discovery Loop with his longtime collaborator Sanjay Ghemawat, a Google Senior Fellow who worked alongside him on several of the company’s most important computing systems.

The founding team also includes prominent AI researchers Oriol Vinyals and Quoc Le, bringing together decades of experience in machine learning and large-scale computing. Discovery Loop has been established as a public benefit corporation with an ambitious goal of using artificial intelligence to transform how scientific and engineering research is conducted. The company plans to develop AI systems capable of proposing, carrying out and analyzing large numbers of experiments simultaneously, potentially allowing researchers to explore ideas much faster than traditional methods permit.

Instead of relying on experiments being conducted one after another by human researchers, the company envisions automated systems capable of running thousands of experimental processes in parallel. Dean’s departure represents a major moment for Google, where he has played a central role in the company’s technological evolution since joining in 1999. He was among Google’s earliest employees and joined the company when its workforce was only a few dozen people.

Over the following 27 years, Dean witnessed Google transform from a rapidly growing search company into one of the world’s largest technology organizations, employing more than 190,000 people. During his time at Google, Dean and Ghemawat were instrumental in creating foundational distributed computing technologies such as MapReduce and Bigtable. These systems became critical components of Google’s infrastructure and helped establish the foundations for processing enormous amounts of data across distributed computer networks.

Dean later became involved in Google Brain, helping strengthen the company’s position in artificial intelligence research and development. His work eventually extended into Google’s broader AI programs, including leadership related to the Gemini family of AI models. With Discovery Loop, Dean is now turning his attention toward a new approach that combines AI with automated experimentation. The company initially plans to apply its technology to machine learning research, where automated experimentation could help researchers test models, algorithms and ideas at a much faster pace.

Its longer-term ambitions extend well beyond AI, with potential applications in biology, drug discovery, materials science and semiconductor design. The company’s approach could allow researchers to investigate complex scientific questions through continuous cycles of experimentation and analysis, reducing the time required to move from an idea to a tested result. Dean described his departure as the beginning of a new chapter after nearly three decades at Google and reflected on his long-standing ambition to create technology that could reach millions of people.

Google CEO Sundar Pichai publicly congratulated Dean and Ghemawat on the new venture and confirmed that Google intends to support Discovery Loop as a founding investor and cloud partner. The backing signals that despite Dean’s departure, the relationship between Google and the new company is likely to remain significant. Discovery Loop’s launch comes as technology companies and research organizations increasingly explore ways to use AI not only to analyze information but also to conduct scientific discovery itself.

If successful, its automated experimentation model could reshape research by allowing machines to continuously generate hypotheses, perform experiments and evaluate results at a scale difficult for human teams to match. For Dean, the venture marks a shift from building the computing infrastructure that powered Google’s growth to developing AI systems that could potentially accelerate discoveries across multiple scientific fields.