Jeff Dean's Discovery Loop Eyes $50B Valuation Weeks After $10B Round

Jeff Dean, the legendary Google engineer who departed as the company's chief scientist last month, is already back at the fundraising table for his new AI startup Discovery Loop — this time targeting a valuation of roughly $50 billion. That's a fivefold jump from the approximately $10 billion valuation the company was seeking just weeks earlier when it was raising $1 billion. The speed of the revaluation underscores just how feverishly competitive the AI funding market has become, where top technical talent commands almost limitless investor appetite.

Discovery Loop's mission is ambitious even by AI-lab standards: the company aims to automate scientific and engineering discovery by running thousands of parallel experiments simultaneously, targeting the hardest unsolved problems in those fields. Dean has framed the venture in explicitly idealistic terms, saying at a recent talk that he and his cofounders may make decisions that sacrifice company financial interests in favor of broader societal benefit.

The founding team is formidable. Sanjay Ghemawat, Dean's longtime collaborator since joining Google in 1999, helped architect foundational distributed-computing systems. Quoc Le, a founding member of Google Brain, made landmark contributions to deep learning and large language models. Oriol Vinyals, formerly of Google DeepMind, is known for breakthroughs in sequence modeling and reinforcement learning. Together, their résumés span Google Search, Ads, Gmail, and Gemini — a pitch deck one former Google product leader described as among the most impressive ever assembled. Dean, Le, and Vinyals all rank among the most-cited AI researchers on Google Scholar.

The initial funding round was led by Radical Ventures and Khosla Ventures, with Lightspeed, Kleiner Perkins, and Doerr Capital participating. Alphabet itself joined as a founding investor and cloud partner — an unusual endorsement from Dean's former employer. Vinod Khosla publicly compared the investment to his early bet on OpenAI, predicting the team could generate new science improving billions of lives rather than just shipping software features or benchmark scores.

Discovery Loop joins a cohort of pre-product AI labs — including Safe Superintelligence and Thinking Machines Lab — that have raised extraordinary sums purely on team pedigree and vision. The pattern would have been unthinkable in prior funding cycles but now reflects a hardening investor thesis: a handful of elite technical teams will define the next generation of frontier AI, and missing them early is the real risk.