Google Reshuffles DeepMind Leadership as Gemini Falls Behind Rivals
Co-founder Sergey Brin pushed for faster development while structural changes shift power from London to Mountain View.
Google's AI Lab Undergoes Major Restructuring
Google announced a sweeping leadership change at its DeepMind AI division on August 5, 2026, with co-founder Demis Hassabis stepping back from day-to-day management to become chair. His deputy, Koray Kavukcuoglu, took over as chief, gaining final authority over major decisions at the lab, according to details first reported by Reuters.
The reorganization comes as Google's Gemini model has struggled to maintain competitive positioning. While Gemini briefly led the field in November 2025, subsequent releases from Anthropic and OpenAI pushed it back into catch-up mode. By August 2026, Google had delayed its new flagship Gemini version by two months after internal testing revealed continued performance gaps in areas like coding, according to two sources familiar with the matter.
Simultaneously, the two original technical co-leads of Gemini, Jeff Dean and Oriol Vinyals, departed to co-found a startup, marking a significant talent exodus from the project.
Why It Matters
This restructuring represents more than personnel shuffling—it signals a fundamental shift in how Alphabet approaches AI development. The move consolidates decision-making authority in Mountain View and tightens the connection between DeepMind's research and Google's commercial products. For enterprise technology leaders evaluating AI partnerships, Google's organizational turbulence and acknowledged performance gaps raise questions about execution velocity at a critical moment in the generative AI race.
Brin's Push for Speed and Focus
Google co-founder Sergey Brin, who holds no formal executive role, has been actively pressuring AI staff to accelerate development, two sources told Reuters. In an April 2026 town hall addressing hundreds of employees, Brin urged DeepMind to move faster as Anthropic previewed its Claude Mythos model.
Brin has used his influence as co-founder to direct resources toward specific technical areas, including recursive self-improvement—the point at which AI systems can enhance themselves without human intervention. His informal involvement echoes his response to OpenAI's ChatGPT launch in 2022, when he similarly rallied Google's AI teams.
DeepMind's Eroding Autonomy
The changes further reduce DeepMind's independence, which Google has gradually diminished since acquiring the London lab in 2014. During an all-hands meeting on August 6, leadership revealed that several non-technical teams would move from DeepMind into Google's corporate structure, according to two sources present.
Kavukcuoglu's rise reflects this shift. After moving from London to Mountain View in 2025 for a promotion to chief AI architect reporting directly to CEO Sundar Pichai, he became the central figure directing Gemini's development. Five sources confirmed he had assumed oversight of model development more than a year before his formal appointment as DeepMind chief.
Some London-based leaders found their influence diminished as Kavukcuoglu built relationships across Google's commercial divisions. He served as the primary contact between DeepMind and Google Cloud, the company's main AI revenue engine, even before relocating to California.
Structural Challenges Persist
Four sources characterized Google's lagging position as stemming from persistent structural issues. Computing capacity constraints and disagreements among Gemini's multiple project leaders resulted in insufficient resources for critical areas like coding. Google's bureaucratic culture also slowed model releases compared to more agile competitors.
Senior cloud division leaders expect Kavukcuoglu's appointment to ease tensions over allocation of Google's limited supply of Tensor Processing Units (TPUs), the specialized chips required for AI model training.
Google's stock dropped 4% on the day of the announcement. The company declined to comment on specific claims about internal challenges but stated it would continue scaling infrastructure to address TPU constraints.
Reuters reported these exclusive details based on interviews with seven people knowledgeable about Gemini's development.
This is an original analysis by the Omega editorial team. Source reporting: AI Watch.
Want systems like this working for your business?
Book a Call