PALO ALTO, CALIFORNIA — For 27 years, Sanjay Ghemawat's name sat quietly behind some of the most consequential engineering in the history of the internet. As a co-author, with Jeff Dean, of the Google File System, MapReduce, Bigtable and Spanner, Ghemawat helped build the technical scaffolding that let Google's search engine — and much of the modern cloud computing industry that followed it — operate at planetary scale. This week, that quiet, decades-long partnership took a public turn: Ghemawat and Dean are leaving Google to co-found Discovery Loop, a new AI startup aimed at automating scientific and engineering research, backed by Alphabet itself as a founding investor.
Google confirmed the departures on Wednesday in a joint memo from CEO Sundar Pichai and DeepMind CEO Demis Hassabis, titled "Next Chapter of Our AI Momentum." Joining Ghemawat and Dean in the new venture are Oriol Vinyals, a vice president of research at Google DeepMind who helped design the company's chatbot technologies, and Quoc Le, a co-founder of Google Brain whose research helped inspire today's generation of AI chatbots. Dean is expected to serve as CEO.
From Distributed Systems Pioneer to AI Founder
Ghemawat, an Indian-American computer scientist born in West Lafayette, Indiana, joined Google in 1999 after working at Digital Equipment Corporation's Systems Research Center, where he first began collaborating with Dean. Their partnership at Google produced some of the field's most cited systems papers: the Google File System (2003), MapReduce (2004), Bigtable (2006) and Spanner (2012), each of which became foundational infrastructure not just for Google but for the broader distributed-computing industry that grew up around cloud platforms like Amazon Web Services and Microsoft Azure. Wired magazine has described Ghemawat as one of the "most important software engineers of the internet age." He was elected to the National Academy of Engineering in 2009 and, together with Dean, received the ACM Prize in Computing in 2012.
What Discovery Loop Aims to Build
Discovery Loop, organized as an independent public benefit corporation, aims to automate the iterative cycle of scientific research — proposing experiments, running them, evaluating results and iterating, potentially thousands of times over. "Imagine a future where a handful of people can conduct scientific research and engineering tasks much more rapidly, and with higher quality, than massive teams of scientists," the company said in describing its mission. At launch, the company plans to focus on automating machine-learning research and engineering itself, before expanding into areas such as hardware design, drug discovery and clean energy, according to reporting by the Wall Street Journal.

The new venture is based in Palo Alto and has raised seed funding from Radical Ventures and Khosla Ventures, among other Silicon Valley investors — with Google itself as a founding investor and, notably, as the startup's cloud computing provider for at least its first year. The departing researchers have declined to disclose the size of their funding round, but by most accounts the exit was an amicable one: Pichai wrote that Dean and Ghemawat "helped drive some of the most significant technology transitions" during their tenures, from Google's early search infrastructure to the neural networks underpinning the modern AI era.



