Alphabet Inc. (GOOGL) is racing to close the gap with rivals in the artificial intelligence arena, and co-founder Sergey Brin has stepped into the fray. Over the past several months, Brin has been urging key AI staff to commit fully to the company’s Gemini model, according to people familiar with his remarks. The push comes as Google’s flagship AI model has slipped behind competing systems from Anthropic and OpenAI, forcing a sweeping leadership overhaul at Google DeepMind.

At an internal town hall in April attended by hundreds of employees, Brin addressed the DeepMind AI lab directly, demanding faster progress. The meeting took place shortly after Anthropic released a preview of Claude Mythos, a powerful model that quickly gained recognition as a front-runner in the field. Brin’s message was unambiguous: the lab needed to move faster and narrow the competitive gap.

Google’s Gemini model briefly surpassed rivals in November of last year with the release of Gemini 3, but the lead proved short-lived. Anthropic and OpenAI responded with new model updates that once again placed Google in catch-up mode. By August, internal testing revealed that the next flagship version of Gemini continued to trail competitors in key areas such as coding, prompting Google to delay its release by approximately two months, according to two people familiar with the matter.

Leadership overhaul at DeepMind

On August 5, Alphabet announced the most significant restructuring of its Google DeepMind division in years. Demis Hassabis, who had led the AI lab since Google acquired it in 2014, stepped down from day-to-day management to become chairman of the unit. He also assumed a newly created role as Alphabet’s first Chief Scientist. Koray Kavukcuoglu, previously DeepMind’s chief technology officer, took over as senior vice president of Google DeepMind, reporting directly to Alphabet CEO Sundar Pichai.

Kavukcuoglu’s responsibilities will include Gemini model development, frontier AI research, and the Gemini app and developer teams. A Google spokesperson confirmed that he will have final say on major decisions for DeepMind.

The restructuring extended beyond the executive suite. Jeff Dean, a 27-year Google veteran who served as DeepMind’s chief scientist, left the company to co-found a new startup called Discovery Loop, a public-benefit corporation focused on using AI to automate scientific and engineering research. Google will invest in the venture. Oriol Vinyals, one of Gemini’s original technical co-leads, also departed to join the startup, along with researchers Quoc Le and Sanjay Ghemawat.

DeepMind’s autonomy erodes

The changes mark a further erosion of DeepMind’s autonomy, which Google has gradually chipped away at since acquiring the London-based lab. For years, DeepMind operated with significant independence, prioritizing long-term foundational research over immediate commercial returns. That approach has now shifted decisively toward productization and revenue generation.

During an all-hands meeting on August 6, the day after the public announcement, Hassabis and Kavukcuoglu informed staff that several non-technical support teams would be moved out of DeepMind and integrated into Google’s corporate reporting structure. The executives sought to reassure employees that day-to-day work would remain unchanged, but the message landed uneasily among staff who had long valued the lab’s research-driven culture.

Some London-based DeepMind leaders have felt their influence on Gemini’s technical direction diminish since Kavukcuoglu relocated to Google’s Mountain View headquarters in 2025. Kavukcuoglu had been appointed Google’s chief AI architect that year, tasked with embedding Gemini across the company’s products, including search, cloud, and enterprise software. Four sources said that in the months following that appointment, Kavukcuoglu became the central figure shaping Gemini’s direction, with Brin’s backing.

Brin’s informal influence

Although Brin no longer holds an executive title, the 52-year-old co-founder has been informally influencing model training efforts for months. He has repeatedly told key staff that Google must catch up to the AI frontier, and has used his implicit authority to push resource allocation toward specific areas, including recursive self-improvement — the point at which AI systems can enhance their own capabilities without human intervention.

Brin’s re-engagement echoes his response to OpenAI’s ChatGPT launch in 2022, which jolted Google into action. The current push reflects similar urgency as Anthropic and OpenAI continue to ship increasingly capable models at a rapid pace.

Competitive pressures mount

Google’s struggles in the model race have been well documented. According to Artificial Analysis, an AI performance analytics firm, Google’s top-ranked model, Gemini 3.6 Flash, tied for 11th place with OpenAI’s GPT-5.6 Luna and China’s DeepSeek V4 Flash 0731. Gemini 3.5 Pro, which Pichai had announced at the I/O developer conference in May for a June release, has still not launched.

Analysts point to structural challenges that have hindered Google’s progress. Computing capacity constraints, disagreements among Gemini’s multiple project leaders, and Google’s bureaucratic decision-making processes have all contributed to slower model releases compared with nimbler rivals. The company formed an internal team called CodeStrike to improve coding capabilities, recognizing that coding has emerged as the first clear killer application for large language models — an area where Anthropic and OpenAI have taken a substantial lead.

Morningstar senior equity analyst Malik Ahmed Khan noted that Google’s focus on monetizable areas like search and multimodal applications caused it to miss the coding opportunity. He said the appointment of Kavukcuoglu signals a sharper focus on large language model improvements and positions Google more favorably in the LLM race.

Cloud tensions and commercial focus

Kavukcuoglu’s appointment is also expected to ease tensions between DeepMind and Google Cloud over allocation of Tensor Processing Units (TPUs), the company’s custom AI chips. Senior leaders in the cloud division anticipate that Kavukcuoglu, who served as the primary point of contact between DeepMind and Google Cloud even before his relocation, will help resolve resource allocation disputes that have hampered both teams.

The commercial stakes are substantial. Alphabet reported second-quarter revenue of $119.8 billion, up 24% year over year, with Google Cloud revenue surging 82% to $24.8 billion. Cloud backlog reached $514 billion. Gemini App now counts 950 million monthly active users, and nearly 90% of Fortune 100 companies have adopted Gemini Enterprise. Yet Pichai acknowledged that demand for AI products exceeds the company’s current supply capacity, particularly around TPUs.

MetricQ2 2026Year-over-Year ChangeTotal revenue$119.8B+24%Google Services revenue$94.5B+15%Google Cloud revenue$24.8B+82%Cloud backlog$514BN/AGemini App monthly active users950MN/A

Note: Figures reflect Alphabet’s second-quarter 2026 results as reported in earnings materials.

Market reaction and outlook

Alphabet shares fell approximately 4% on the day of the restructuring announcement, reflecting investor uncertainty about the leadership transition and the departure of key technical talent. The stock has gained roughly 76% over the past 12 months despite the dip.

Analysts offered mixed views on the reorganization. CCS Insight chief analyst Ben Wood said the goal is clearly to narrow the gap with Anthropic and OpenAI in specific areas, and that Kavukcuoglu’s promotion reflects a shift from academic projects toward improving frontier model performance and developer tools. Constellation Research chief analyst Ray Wang said the move prioritizes execution over deep research, predicting faster product releases and more experimentation.

Futurum Group AI director Nick Patience said Kavukcuoglu’s first priority will be shipping Gemini 3.5 Pro as soon as possible and proving the release is not a one-off by maintaining a steady cadence. He also noted that Kavukcuoglu will need to rebuild coding and pre-training expertise lost through recent departures.

Hassabis, for his part, framed the change as an opportunity to focus on long-term AGI strategy and scientific applications of AI. He will continue leading Isomorphic Labs, Alphabet’s AI drug-discovery spinoff. In the all-hands meeting, he spoke about how AI could advance science and alter society — themes that will define his new role as chairman.

Google declined to make Brin available for an interview and did not answer questions about his recent involvement, DeepMind’s structural changes, or staff concerns. A company spokesperson said Google remains committed to its presence in London, even as AI leadership consolidates in Mountain View.

The restructuring represents a pivotal moment for Alphabet’s AI ambitions. Whether Kavukcuoglu can shorten Gemini’s product iteration cycles, resolve computing resource constraints, and close the gap with Anthropic and OpenAI will determine whether this organizational bet pays off.