Leadership Crisis at DeepMind: Gemini Falls Behind as Top Talent Flees to New Ventures
Key Takeaways
- ▸Jeff Dean and other top Google researchers departing DeepMind to start Discovery Loop signals loss of confidence in Gemini's future trajectory
- ▸Gemini has fallen from peak performance (Gemini 3 Pro in Nov 2025) to 8th-9th place in model rankings as of mid-2026
- ▸DeepMind faces a severe talent retention crisis with multiple high-profile departures from key AI research teams including reinforcement learning experts
Summary
Google announced a major DeepMind leadership overhaul on August 5th, with significant departures that industry observers say signal deep organizational dysfunction. Jeff Dean, former Google Chief Scientist and Gemini co-lead, is leaving to start Discovery Loop with other top researchers including Sanjay Ghemawat, Quoc Le, and Oriol Vinyals—all among Google's elite technical talent. Koray Kavukcuoglu, the remaining Gemini co-lead, is taking over from Demis Hassabis as DeepMind's leader.
The departures underscore mounting competitive pressures as Gemini has fallen significantly behind rival AI models. Once positioned as one of the top three AI models globally when Gemini 3 Pro peaked in November 2025, the model has since declined to 8th or 9th place in current rankings, falling behind Claude Opus 4.5, GPT 5.6, Fable 5, and other competitors. Industry sources suggest the upcoming Gemini 3.5 Pro performs roughly at Claude Opus 4.5 level—a meaningful step backward—while Gemini 3.6 Flash trails models like Muse Spark and Grok 4.5.
While Gemini struggles, Google Cloud Platform (GCP) is emerging as the clear winner from the reshuffling, with year-over-year revenue growth exceeding 100%. As top AI talent exits DeepMind to launch new ventures funded by external investors, many are choosing to build on GCP infrastructure, effectively redirecting significant computing resources away from the Gemini division toward Google's more profitable cloud business.
- Google Cloud Platform is accelerating (>100% YoY revenue growth) as departing researchers and external AI labs choose GCP for computing infrastructure
- Internal compute allocation at Google now prioritizes GCP profitability over Gemini model development, reflecting a fundamental strategic shift
Editorial Opinion
The exodus of Jeff Dean and his cohort represents a critical failure in Google's AI strategy, despite the company's historical dominance in foundational LLM breakthroughs. The cruelest irony: as Gemini falls behind competitors, Google Cloud emerges as the real winner—profiting from the very talent and resources leaving the Gemini division. This suggests Google's organizational structure is fundamentally misaligned with the pace and intensity required to compete in frontier AI, and that breakthrough models may increasingly come from external labs rather than from inside Google's walls.
