Sergey Brin leads Google’s AI revival from a California microkitchen

On the second floor of Gradient Canopy, a tent-like structure near Google’s Mountain View, California headquarters, a microkitchen has become an unlikely nerve center for the company’s efforts to stay ahead in the accelerating AI arms race. Here, among coffee grinders and an espresso machine, is a U-shaped formation of desks occupied by Google cofounder Sergey Brin and a handful of select leaders now helping steer the direction of Gemini, the company’s flagship AI model. According to current and former employees interviewed by Business Insider, this kitchen-turned-strategy war room is gaining prominence as Google faces intense pressure from competitors and navigates ongoing internal upheaval.

The growing influence of the microkitchen comes as Google restructures its AI leadership following industry disruptions and talent losses. In recent weeks, DeepMind CEO Demis Hassabis relinquished his role to become Alphabet’s chief scientist, while Koray Kavukcuoglu moved from London to head Google DeepMind as senior vice president. Jeff Dean, the company’s longtime chief scientist, departed after 27 years. While Brin’s official position remains undefined, insiders say his role with Gemini has grown more decisive, especially as he wields unique authority to marshal resources and bypass bureaucratic obstacles that others cannot.

After the 2023 merger of Google’s two central AI labs, DeepMind and Brain, tensions emerged over Gemini's direction, leading to friction between teams and the departure of top talent, including star scientists John Jumper and Noam Shazeer. Some leaders in Mountain View reportedly grew frustrated with Hassabis and London-based researchers, whom they viewed as less pragmatic about making Gemini more competitive. The microkitchen’s informal, startup-like dynamic enables Brin and his close allies, such as Kavukcuoglu and research engineer Sebastian Borgeaud, to make decisions more rapidly. Employees say this environment allows Brin to clear red tape and protect high-priority projects, with some noting that being present in the microkitchen is essential for those hoping to shape Gemini’s future.

Brin, who retains significant stock-based influence over Google’s board despite stepping down from executive duties in 2019, became deeply involved with Gemini after the launch of OpenAI’s ChatGPT in late 2022. “As a computer scientist I've never seen anything as exciting as all of the AI progress that's happened in the last few years,” Brin said at the All-In Summit in Los Angeles in 2024. His involvement ranges from weighing in on technical decisions like model size and development timelines to advocating controversial strategies, such as monitoring Google engineers as they code to improve Gemini’s reinforcement learning for coding abilities.

His management style, however, has not been without controversy. Brin’s decision to cut Project Frozen, an effort led by Jeff Dean to embed Gemini’s model architecture directly into hardware, frustrated some senior staff and preceded Dean’s eventual exit. Brin also faced internal pushback after advocating for longer working hours, suggesting that 60 hours per week was the productivity “sweet spot.” Some managers instructed their teams to disregard this guidance.

Tensions between DeepMind’s research-driven culture and Google Brain’s focus on product-centric results remain, insiders report. After the merger, Brin’s presence grew as a quiet but forceful influence, with leadership increasingly viewing his intervention as crucial to breaking internal deadlocks and keeping development on track. For now, as Google aims to deliver a breakthrough with Gemini 4, Brin’s startup-style leadership in the improvised heart of Gradient Canopy may be central to Google’s efforts to reclaim its position at the forefront of artificial intelligence.

Google and the identified leaders declined to comment for this story.