According to an official report by journalist Steven Levy in WIRED magazine, four of tech giant Google's top and most senior artificial intelligence scientists are leaving the company to launch a groundbreaking new startup called Discovery Loop. The venture, established as a public benefit corporation, will strive to achieve AI-powered scientific and engineering breakthroughs across a wide range of fields—including drug discovery, biology, chip design, and material design—by deeply automating the scientific method. This historic departure is considered a particularly devastating blow to Google, which is currently in the midst of an intense and fierce competition in the global AI market.
The Mysterious Lecture at Y Combinator and the Official Reveal
On July 25, Jeff Dean spoke to approximately 6,000 aspiring founders at Y Combinator’s Startup School event, held at the Chase Center in San Francisco. Dean, a legendary engineer at Google who co-founded the Google Brain division, served as the chief scientist of Google DeepMind and Google Research, and acted as the technical co-lead of Google’s flagship AI model, Gemini, is widely regarded as one of the most influential figures in the software world. Alongside his long-time collaborator, Sanjay Ghemawat, Dean wrote significant portions of Google's core computing and search infrastructure. A well-known joke in the computer science community suggests that Jeff Dean’s resume lists the things he hasn't done, simply because that list is far shorter.
During the interview at the Chase Center, Dean was asked what areas startup founders should think about when launching new companies. Dean replied that the most important thing is to choose a topic that excites the founder and would be useful to the world. He then shared his personal interest: a fully 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," Dean explained on stage. According to him, running thousands of these automated loops could unlock groundbreaking advances in multiple domains like science, biology, chip design, and even making improved AI models themselves. Unbeknownst to the crowd in attendance, Dean was actually speaking about his own secret startup, which he had organized just days prior. Now, the departure has become official: after nearly 27 years at Google, Dean is leaving the company alongside Ghemawat and two other senior AI scientists to found Discovery Loop.
The Founders: The Ultimate AI "Super-Team"
Aside from Dean and Ghemawat, who were among Google's earliest hires and are considered pillars of the company's infrastructure (Steven Levy compared their departure to Mick Jagger and Keith Richards leaving the Rolling Stones to start a new band), the founding team of Discovery Loop includes two other prominent figures. The first is Oriol Vinyals, VP of research at DeepMind and a technical lead for the Gemini program. The second is Quoc Le, a co-founder of Google Brain and the key scientist behind AutoML-Zero—a project that uses machine learning algorithms to autonomously build AI products.
In an interview conducted by Steven Levy with the co-founders just before the official launch, Dean shared that the idea was born only a few weeks earlier. He was quickly joined by the other three scientists, who are not only long-time colleagues but close friends who vacation together. "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," Dean noted.
The Operating Model: The Startup That Will Be Its Own First Customer
The four founders have mapped out a roadmap wherein Discovery Loop will be its own first customer. The autonomous experiment loops they develop will initially be directed toward improving the company's own advanced machine learning algorithms. This improved software will subsequently become the core of an AI system capable of tackling complex problems in subsequent domains. "I’m very excited about automating machine learning," said Quoc Le. "It might be that we will discover a different transformer architecture."
From there, Discovery Loop aims to achieve extraordinary breakthroughs across a variety of fields, including chip design, biology, drug discovery, and material design. Oriol Vinyals explained: "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." In the initial phase, the team envisions that these ideas will be co-developed with humans, but the ultimate goal is deep and full automation of the entire process. If the model succeeds, the founders contend that it is possible a small number of individuals using Discovery Loop could out-invent the world’s largest research organizations. These small teams are expected to be customers of Discovery Loop, alongside the startup running its own independent loops for discovery and self-monetization.
Secret Fundraising and Discrete Weekend Meetings
For the initial fundraising round, the founders prepared a simple pitch deck presenting their professional backgrounds and a general outline of how their approach could succeed. Dean emphasized that they did not use AI to create this pitch deck; however, the background and reputation of the team members spoke louder than any PowerPoint slide ever could.
Renowned venture capitalist Vinod Khosla met with the team at his Sand Hill Road office on a Saturday to prevent any leaks or early hints regarding Google's impending loss. Khosla was highly enthusiastic about the concept, stating: "With this team, I wouldn’t need to know what they were doing before I backed them... It’s the ultimate superstar team." Khosla went on to explain the unique nature of the venture: "Humans have been using AI to do research, not using AI to be a researcher. The fundamental thing [in Discovery Loop] is that AI itself is the researcher."
Other meetings took place at Jeff Dean's home. Ultimately, Khosla Ventures and Radical Ventures (an AI-focused venture capital firm) invested in the venture, alongside several other funds. The founders chose not to disclose the funding amount or the valuation at which the investment was made. Jordan Jacobs, managing partner at Radical Ventures who will join the startup’s board, commented on the investment, saying: "Building something to solve all those problems is a powerful idea. These people have been doing this kind of work in the past so they know what they’re doing."
The Complex Exit Negotiations with Google
Negotiating the exit from Google was particularly painful and complex. Jeff Dean shared that Alphabet CEO Sundar Pichai tried over multiple meetings to convince the team to stay at the company and keep their badges. Ultimately, the team members decided they preferred the fun and freedom that come with starting a new startup. Vinyals explained the decision: "In a large organization there is always a lot of inertia you have to overcome to make any radical changes. We want to build something different."
Despite the departure, Google ultimately gave the exiting engineers its blessing, alongside a financial investment and an agreement to provide compute power during the startup's first year of operations. In an official statement, Alphabet CEO Sundar Pichai wrote: "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."
Current Status and the Roadmap Ahead
Due to the dramatic impact of the executives' departure on Google, Dean and his partners took great pains to maintain absolute stealth mode. As of the time of reporting, the company has not yet begun hiring employees (which could potentially lead to further engineering defections from Google) nor has it rented its own office space. When the founders were asked who serves as the company's CEO, a brief silence ensued, after which Dean sheepishly replied: "I think I’m the CEO. Everyone pointed at me."
Discovery Loop’s mission is not entirely unique in the technological landscape. Almost every major figure in AI today—including Dario Amodei of Anthropic, Sam Altman of OpenAI, and Jensen Huang of Nvidia—has claimed at one point or another that AI will become an engine of scientific discovery and solve problems that humans cannot crack on their own. However, Discovery Loop’s founding team is betting that the unique knowledge and skills they have accumulated over the years will allow them to win the race and build the system that actually achieves these breakthroughs. Beyond this ambitious goal, the company faces a narrower, more immediate test of success: will the work they perform as a startup be significant enough to compensate for the massive talent gap they are leaving behind at Google?