DeepMind Alumni Have Launched 112 Startups in 18 Months

David Silver's $1.1 billion seed round for his new UK artificial intelligence lab Ineffable Intelligence dominated coverage last month. It was real money at an extraordinary stage. But framing it as an outlier misses the broader pattern that generated it.

New data from Evertrace, a research firm that tracks talent flows for venture investors, counts 112 former Google DeepMind researchers who have launched or are actively forming startups over the past 18 months, according to tech.eu. Silver is one of them. The pipeline is structural, not exceptional.

Of those 112 departures, 38 have formally started companies with public websites and LinkedIn pages. The remaining 74 have moved into what the industry calls stealth roles, a term that typically signals a pre-launch startup, though both Evertrace and tech.eu acknowledge the designation could not be confirmed in every individual case. Even discounting ambiguous cases, the scale is significant: more than a hundred trained researchers leaving a single lab within an 18-month window is a material shift in how advanced AI research gets distributed.

The geographic split

The numbers skew heavily toward the United States. Seventy of the 112 ventures are US-based; 28 are in the UK, where DeepMind was originally founded in 2010. Spain accounts for three startups. Switzerland, Germany and Canada each have two. Austria, Poland, Hong Kong, India and South Korea each have one, per tech.eu.

That most of the activity lands in the US rather than Europe might seem to complicate the "European tech ecosystem" framing, but Evertrace co-founder Jacob Houlberg draws a different analogy. "We are seeing that DeepMind is doing for AI what Klarna and Spotify did for European tech," Houlberg said in comments reported by tech.eu. "It's compounding into a founder factory. David Silver is the headline, but the pipeline behind him is the real story."

Among the alumni identified beyond Silver, Saining Xie stands out. An ex-DeepMind scientist and NYU computer science professor, Xie initiated the research work that seeded the diffusion transformer architecture in 2022. He now serves as chief science officer and co-founder at Advanced Machine Intelligence, the AI world-model company co-founded with former Meta chief scientist Yann LeCun. The pairing illustrates how DeepMind alumni are intersecting with parallel talent networks to build research organizations that draw from multiple elite lineages at once.

What this means for the field

The Klarna and Spotify comparison captures something real but obscures something important. Those companies applied well-understood technology to consumer markets with clear monetization paths. The current wave of DeepMind spinouts is attempting to commercialize foundational research in reinforcement learning, world models, and related areas of artificial intelligence, with inherently longer and less certain paths to revenue. What the analogy does accurately convey is the compounding dynamic: alumni hire alumni, share early investors, and form informal research networks that begin to replicate some of the density of a centralized lab.

For practitioners watching this from outside, the more consequential question is whether distributed startup structures can sustain the kind of long-horizon research that produced the work these founders are now commercializing. Compute access, proprietary data, and coordination at scale are genuine advantages that centralized labs retain over even well-funded startups. A $1.1 billion seed round is exceptional by any standard; it does not replicate the infrastructure of a parent organization embedded inside Google.

The 74 ventures still in stealth represent the next data point. When they emerge, the research directions they choose will reveal more about the true shape of this diaspora than any single fundraise can.

FAQ

What is the total number of DeepMind alumni startups identified in this data?
Evertrace counted 112 former Google DeepMind researchers who have launched or are forming startups over an 18-month period through early 2026, with 38 confirmed companies and 74 in stealth roles.

Who is David Silver and what is Ineffable Intelligence?
David Silver is a prominent DeepMind researcher associated with reinforcement learning work including AlphaGo. Ineffable Intelligence is his new UK-based artificial intelligence lab, which raised a reported $1.1 billion seed round in April 2026.

What is Evertrace?
Evertrace is a data firm that helps venture capital investors track talent movements and company formation. The firm compiled its DeepMind dataset from public sources including Companies House filings, patent records, and research grants.

What does a stealth role mean in this context?
A stealth role typically refers to an individual who has joined or is founding a company that has not yet publicly announced itself. Of the 112 alumni counted, 74 were in such roles, though the report notes these could not all be independently confirmed as new startup launches.