- Alphabet’s Google is consolidating its AI leadership in Mountain View, California, appointing Koray Kavukcuoglu — who moved from London to California in the past year — to run its sprawling AI research and operations; Demis Hassabis, DeepMind’s co-founder and long-time CEO, has stepped back from day-to-day operations to become chairman of Google DeepMind and Alphabet’s Chief Scientist; Sebastian Borgeaud, head of a key AI coding effort, also relocated from the UK to California; the reorganization attempts to reverse three years of friction since Google’s 2023 merger of Google Brain (Mountain View) and DeepMind (London) under the Google DeepMind banner, a combination that left researchers operating on separate continents with complicated decision-making and frustrated talent on both sides.
- The urgency of the reorganization reflects Google’s deteriorating competitive position: the company is months behind schedule on the most powerful version of its flagship Gemini model; its products are widely seen as an afterthought in the coding automation market — one of AI’s most lucrative early commercial applications — where Anthropic’s Claude and OpenAI’s models dominate; Google has pledged up to $205 billion in capital expenditures for AI in 2026 alone, which creates enormous pressure on Kavukcuoglu to deliver results quickly; Kavukcuoglu is now the only remaining Gemini co-lead at the company, following an exodus of senior AI talent that includes legendary Google engineer Jeff Dean, who left to found a startup called Ineffable Intelligence that raised $1.1 billion in seed capital, and Simon Bouton, DeepMind’s Chief Experience Officer and 13-year veteran, who is departing for London-based startup CuspAI.
- Kavukcuoglu’s background is the essential context for understanding why Google chose him for this moment: he founded DeepMind’s deep learning team, pioneered major research breakthroughs including work that convinced the field that AI could fundamentally change how machines process language, and has been quietly rising through DeepMind’s leadership for years — promoted to Chief AI Architect last summer; he kept a lower public profile than Hassabis but has been deeply trusted by Google leadership; Yann LeCun, the deep learning pioneer who mentored Kavukcuoglu during his PhD at NYU, describes him as “a great engineer” who “developed excellent research management skills”; critically, Sergey Brin — who has been deeply involved in Gemini — has been working closely with Kavukcuoglu and described him as the one who “organizes the groups and has them deliver stuff.”
- The London-to-California shift represents more than an org chart change — it signals Google’s acknowledgment that the distributed, multi-continent lab structure that was designed to integrate DeepMind’s research culture with Google Brain’s engineering culture has not worked competitively; the departures of senior London-based DeepMind researchers to startups (David Silver founding Ineffable Intelligence, Bouton heading to CuspAI) suggest the integration friction is genuinely driving attrition rather than just restructuring optics; Google says London remains an important hub and has opened a new office there where Hassabis will focus on AI vision and strategy, but the operational center of gravity for the Gemini model — the product that determines whether Google wins or loses the AI race — has moved firmly to California.
What Happened?
Google appointed Koray Kavukcuoglu to run its AI research and operations as Demis Hassabis steps back to a chairman/Chief Scientist role. The move consolidates leadership in Mountain View, reversing years of London-California friction since the 2023 Google Brain-DeepMind merger. Kavukcuoglu is now the only remaining Gemini co-lead. The backdrop: Google is months behind on Gemini’s most powerful version, researchers are defecting to Anthropic and OpenAI, Jeff Dean left to raise $1.1B for a new startup, and Google has committed up to $205B in 2026 capex to back its AI ambitions.
Why It Matters?
Google has the most capital, the most data, and some of the best AI researchers in the world — and it’s still losing. The distributed lab structure that seemed like a strength (bringing in DeepMind’s research culture) has become a liability (slow decisions, frustrated talent, researcher flight). Kavukcuoglu’s appointment is a bet that operational focus and California consolidation can fix what organizational complexity broke. Whether he can deliver a competitive Gemini model before Anthropic and OpenAI extend their leads in coding and enterprise is the question that determines whether $205B in capex generates returns.
What’s Next?
Watch the Gemini model launch timeline for any acceleration or further delays — this is the first hard test of whether the reorganization improves execution velocity; watch the talent attrition rate, particularly from the London DeepMind office, for signals on whether consolidation is stabilizing or accelerating departures; watch Google’s coding AI product (competing with Claude Code and Cursor) for competitive benchmarks against Anthropic and OpenAI; and watch Sergey Brin’s continued involvement with the Gemini team for signals on whether Google’s co-founder is treating this as a mission-critical personal priority.
Source: Bloomberg
