The Google new AI boss race intensifies as top leadership restructures Gemini teams to rival generative AI leaders like OpenAI.
DeepMind’s new operational lead Koray Kavukcuoglu faces immediate pressure to close the AI performance gap with rivals OpenAI and Anthropic following a major leadership reshuffle at Google.
While recent model releases from OpenAI and Anthropic have drawn both acclaim and concern over their capabilities, Google which acquired DeepMind in 2014 has not unveiled a new frontier model since early 2026.
Koray Kavukcuoglu, previously DeepMind’s CTO and Google’s chief AI architect, steps into the role of Senior Vice President to lead the unit’s day-to-day operations and frontier model development. He succeeds co-founder Demis Hassabis, who transitions to chair of Google DeepMind to focus on long-term strategy.
“The goal will undoubtedly be to close the gap with Anthropic and OpenAI in key areas,” CCS Insight chief analyst Ben Wood told CNBC.
Wood added that Kavukcuoglu’s promotion signals a clear shift in Google’s strategy moving away from academic research to focus on sharpening frontier model performance and expanding developer tools.
After falling behind OpenAI following the launch of ChatGPT in 2022, Google shifted the dynamic in November 2025 with the release of Gemini 3.
Analysts praised the model for advancing the industry benchmark, noting its performance surpassed expectations for current-generation frontier AI.
However, 2026 saw Anthropic and OpenAI regain momentum with major releases. Anthropic broke new ground with its Mythos model, while OpenAI drew similar acclaim for GPT-5.6.
Following the launch of Gemini 3.1 Pro in February, Google’s releases this year have struggled to challenge rivals at the AI frontier. To bridge the gap, Google formed an internal team called “Code Strike” to boost coding capabilities a core pillar in the race toward artificial general intelligence (AGI).
Google fell behind Anthropic and OpenAI by prioritizing immediately monetizable areas like Search and multimodal AI, missing out on coding the first clear “killer application,” according to Morningstar senior equity analyst Malik Ahmed Khan. He added that rivals remain “miles ahead” of Google in coding capabilities.
Refuting claims that it lost focus on pure intelligence, a Google spokesperson told CNBC that the path to AGI requires concurrent breakthroughs across agents, coding, robotics, and world models. They pointed to DeepMind’s recent advances in robotics, video generation, computer-use agents, and policy research as evidence of a balanced, multi-pronged approach.
A DeepMind veteran who joined in 2012 prior to its Google acquisition, Kavukcuoglu aims to refocus the unit on competing with Anthropic and OpenAI at the AI frontier, even as the division navigates several high-profile talent departures.
Reporting directly to Google CEO Sundar Pichai, Kavukcuoglu will oversee Gemini model development, frontier AI research, and the teams leading the Gemini app and developer tools, Pichai announced in a statement.
Morningstar’s Malik Ahmed Khan noted that Kavukcuoglu’s appointment will bring clearer focus to Google DeepMind, with large language model improvements taking priority. He added that the leadership shift leaves Google better positioned to compete in the LLM race than before.
“Demis has been much more focused on achieving AGI through architectures beyond large language models,” Khan noted.
Under Hassabis’ leadership, DeepMind prioritized broader research areas including multimodal AI, world models, and computational health through projects like Isomorphic Labs, Khan added.
A source close to DeepMind told CNBC that Kavukcuoglu had already been assuming broader responsibilities over the past year including leading model development and presenting major Gemini releases alongside Hassabis.
Constellation Research principal analyst Ray Wang said that Kavukcuoglu’s promotion signals Google is prioritizing execution over deep research, paving the way for faster release cycles, bolder experimentation, and tighter alignment with a product roadmap.
Alphabet shares dipped in July following a Bloomberg report that Google delayed the launch of Gemini 3.5 Pro to refine its performance, particularly in coding.
Futurum Group AI lead Nick Patience noted that Kavukcuoglu’s immediate priority must be shipping Gemini 3.5 Pro and establishing a dependable release schedule to prove consistency. Beyond delivering a competitive frontier model, he added, Kavukcuoglu must rebuild the critical coding and pre-training expertise lost to recent high-level departures.
Founded in London, DeepMind remained headquartered there under CEO Demis Hassabis after its acquisition by Google, playing a pivotal role in catalyzing the UK’s broader AI ecosystem.
Under his previous role, Kavukcuoglu relocated from London to Google’s headquarters in Mountain View, California.
Futurum Group’s Nick Patience noted that Google is “quietly consolidating its AI leadership out of London,” though a Google spokesperson told CNBC the company remains committed to its UK presence.
Alphabet shares dipped following the AI leadership announcement, even as the stock has surged roughly 76% over the past 12 months.
On last month’s earnings call, Alphabet CEO Sundar Pichai highlighted strong enterprise momentum, noting that nearly 90% of Fortune 100 companies now use Gemini Enterprise as Google expands its cloud AI business.
“Although the model is behind and they take catching up seriously, the monetization of the model is doing extremely well,” The Infinity Machine author Sebastian Mallaby told CNBC. He highlighted Google’s widespread AI footprint across Android devices as well as integrations powering Siri and Apple Intelligence on iPhones.
Mallaby added that it is misleading to suggest a general crisis at Google over high-profile resignations, noting that with thousands of employees on the Gemini team, the market shouldn’t overreact to a few executive departures.
Forrester emerging tech VP and principal analyst Brian Hopkins noted that Kavukcuoglu’s expanded mandate overseeing Gemini model development, consumer apps, developer teams, and reporting directly to Sundar Pichai could significantly bolster Google’s competitive edge in AI.
“Unifying models, the consumer app, and developer teams under a single executive reflects how a company organizes a mature product group rather than a research lab,” Hopkins told CNBC, adding that product execution is an area where Google holds a distinct structural advantage over rivals like OpenAI and Anthropic.
