Jeff Dean and Google Executives Found Discovery Loop, an AI Discovery Startup

Jeff Dean, chief scientist at Google DeepMind and a co-founder of Google Brain, has founded a startup called Discovery Loop together with a team of high-profile senior Google executives, Wired reports. The new company aims to find AI breakthroughs in science, primarily in drug development and semiconductor chip design.

Who is behind Discovery Loop?

Dean has spent more than two decades at Google and is one of the principal architects of modern machine learning. In 2011 he founded Google Brain, one of the first major corporate AI research centers in the world, which later became part of Google DeepMind. Under his leadership, the lab created TensorFlow and other foundational tools. It was also at Google Brain, in 2017, that the paper "Attention Is All You Need" was written, the architectural basis of all modern language models.

Dean's co-founders at Discovery Loop are other senior Google executives. Wired describes them as "high-profile" but did not name them at the time of publication.

What will Discovery Loop do?

According to Wired, the startup will apply AI to accelerate scientific discovery in several key areas:

• Drug development — finding candidate molecules and speeding up the path from hypothesis to trials
• Chip design — automating stages of semiconductor architecture development that take years today
• Other scientific fields — a specific list has not been disclosed publicly

"The startup will look for AI breakthroughs in everything from drug development to chip design," says the description of

Discovery Loop published by Wired.

The company's name reflects its core concept: a closed loop of scientific inquiry in which AI formulates a hypothesis, helps design an experiment, analyzes the result, and moves on to the next iteration, faster and more systematically than a human can. Successful AI projects have already used this logic: according to Google DeepMind, AlphaFold predicted the structures of more than 200 million proteins, and AI-driven electronic design automation (EDA) systems are already shortening chip development cycles. Discovery Loop aims to generalize this approach as an independent entity free of corporate constraints.

Why does this matter for the AI industry?

A team of this caliber leaving Google is not a common event. A comparable precedent in recent history is the departure of Dario Amodei and his colleagues from OpenAI and the founding of Anthropic in 2021. Both cases illustrate a persistent trend: leading AI researchers are convinced that the next wave of breakthroughs will come not from inside corporations but from narrowly focused structures that iterate quickly.

The segments Discovery Loop has chosen, AI-accelerated drug development and chip design, are among the priorities of the global venture market. The traditional path from molecule to market takes more than ten years on average, and the design cycle of a modern chip is comparable. AI that can shorten these timelines changes the economics of entire industries. By being independent of Google, Discovery Loop can work with pharmaceutical giants and semiconductor companies that would not hand sensitive data to a Big Tech competitor.

What this means

Discovery Loop is not an ordinary AI startup: it signals a redistribution of key AI talent beyond corporate labs. When one of Google AI's chief architects builds an independent company from scratch, it means the race for AI breakthroughs in fundamental science is entering a new phase, one where speed and focus matter more than scale and resources.