Google's 30th employee just quit to start a company that wants to automate the scientific method itself. If that sentence doesn't stop you for a second, you're not paying attention.
On July 25, 2026, Jeff Dean stood on stage in front of 6,000 aspiring founders at Y Combinator's Startup School in San Francisco. Someone asked him what problems young founders should chase. Dean answered like a man with nothing to hide: pick something you're excited about, something useful. Then he described his own dream project, an automated version of the scientific method. "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 told the crowd, according to Wired.
What the room didn't know is that Dean was describing a company he had already quietly organized. Eleven days later, on August 5, it became official. Dean is leaving Google after nearly 27 years to launch Discovery Loop, a public benefit corporation built to use AI to accelerate scientific and engineering discovery.
This isn't a solo act. Dean is bringing three heavyweight collaborators with him: Sanjay Ghemawat, a Google senior fellow and one of Dean's longest-running technical partners; Quoc Le, a founding member of Google Brain and the scientist behind AutoML-Zero; and Oriol Vinyals, a VP of research at DeepMind and a technical lead on Gemini.
For context on just how deep this cuts, Dean joined Google in 1999 as its 30th employee. He co-founded Google Brain, became chief scientist of Google DeepMind and Google Research after the Brain-DeepMind merger, and served as a technical colead on Gemini. Along with Ghemawat, he wrote major chunks of the infrastructure that still runs Google's crawling, indexing, and query-serving systems. Wired's Steven Levy put the exit in blunt terms, comparing it to Mick Jagger and Keith Richards leaving the Rolling Stones to start a new band. That's not hyperbole. It's arithmetic.
The pitch is bigger than a chatbot for lab work. Discovery Loop says it wants to run thousands of experiments simultaneously, using AI to automate the entire research loop, propose, test, evaluate, repeat, at a scale no human team could match. In its launch materials, the company put it this way: "While science and engineering have tremendously advanced society over past centuries, progress has traditionally relied on slow, sequential human iterations, creating a significant bottleneck. Discovery Loop is developing advanced AI systems that leverage massive computational scale to fundamentally transform the speed and efficiency of innovation by automating complete experimental loops."
The founders plan to be their own first customer. According to Wired, the initial experiment loops will target improving the company's own machine learning algorithms before expanding into other domains like chip design and drug discovery. Quoc Le said he's especially excited about automating machine learning itself, floating the possibility that the process could surface "a different transformer architecture." That's a polite way of describing recursive self-improvement, AI systems helping build better AI systems with less human involvement in the loop. It's one of the more consequential ideas in the field, and Discovery Loop is putting it at the center of its roadmap.
The funding round is being co-led by Radical Ventures and Khosla Ventures, with Kleiner Perkins, Lightspeed, and Doerr Capital also participating. Here's the twist: Alphabet is investing too. CNBC confirmed the departure is happening on friendly terms, with a company representative saying Google will put money into Dean's new venture. Sundar Pichai addressed it directly in a company blog post: "After an incredible 27-year run, Jeff Dean is at a moment where he wants to try something new, and we're excited to support him in that."
Dean's exit is one piece of a larger reshuffle. Demis Hassabis is stepping down as CEO of Google DeepMind to become chairman of that unit and chief scientist of Alphabet. Koray Kavukcuoglu, DeepMind's technology chief, is being promoted to head Google's AI division and will lead development of Gemini 4. Alphabet shares dropped about 4% after the news broke, per CNBC. The timing isn't subtle. Google just reported negative cash flow for the first time on record, driven by capital expenditures that could reach $205 billion for the year, all while fighting OpenAI and Anthropic for frontier-model supremacy.
Losing one legendary engineer is a headline. Losing four of them at once, including the guy who literally helped build the infrastructure your search results run on, is a structural problem. Google is framing this as generous, friendly, even strategic, complete with an equity stake in the new venture. Maybe that's true. But you don't reshuffle your ent
META BLOCK