SAN FRANCISCO (Reuters) — Google co-founder Sergey Brin has been urging key artificial intelligence staff to focus intensely on the company’s Gemini model as Alphabet works to regain momentum against rivals, according to people familiar with his remarks.
At an April town hall, Brin told hundreds of employees that Google DeepMind needed to move faster as competitors such as Anthropic and OpenAI advanced their AI models. Gemini briefly surpassed rival systems last year, but newer releases have again put Google under pressure. The company has also delayed its next flagship Gemini model by about two months after internal testing showed weaknesses, particularly in coding, according to people familiar with the matter.
The pressure has coincided with a major leadership overhaul. On Aug. 5, Google announced that Demis Hassabis would step aside as head of Google DeepMind and become chairman, while his deputy, Koray Kavukcuoglu, would take over day-to-day leadership. At the same time, the two original technical co-leads of Gemini left Google to launch a startup.
The restructuring reflects Alphabet’s growing emphasis on turning Gemini into a commercial engine across its products. During an Aug. 6 all-hands meeting, employees were told that some teams would move from DeepMind into Google’s corporate organization, according to sources.
The changes also represent a further reduction in DeepMind’s independence since Google acquired the London-based AI lab in 2014. Under Hassabis, DeepMind pursued broad AI research while developing Gemini, but some of his decisions had previously clashed with efforts focused on faster commercialization.
Kavukcuoglu’s influence had already expanded after CEO Sundar Pichai appointed him Google’s chief AI architect in 2025, giving him responsibility for integrating Gemini across the company’s products. Several people familiar with the organization said his growing role reduced the influence of some DeepMind leaders and made him increasingly central to Gemini’s strategy.