Koray Kavukcuoglu, former CTO of DeepMind and Google's chief AI architect, is now SVP of DeepMind. His goal: closing the performance gap between Google's Gemini and the latest models from OpenAI and Anthropic.
TL;DR
- Kavukcuoglu takes over DeepMind with a focus on improving Gemini's performance against OpenAI and Anthropic.
- Google aims to catch up in the AI frontier, with a renewed emphasis on coding and developer tools.
- DeepMind's shift may impact Google's broader AI strategy, including AGI research and commercial applications.
What happened
Koray Kavukcuoglu, previously DeepMind's CTO and Google's chief AI architect, has been promoted to SVP of DeepMind, replacing co-founder Demis Hassabis, who becomes chair. Kavukcuoglu will oversee Gemini model development, Frontier AI research, and the Gemini app and developer teams, reporting directly to Google CEO Sundar Pichai.
Google has lagged behind OpenAI and Anthropic in recent months, with the latter releasing advanced models like Mythos and GPT-5.6. Google's last frontier model, Gemini 3, was released in November 2025, and its subsequent releases have not challenged the leading models.
Analysts suggest that Kavukcuoglu's appointment signals a shift in focus towards improving Gemini's performance and developer tools, rather than academic projects. This move comes amid high-profile departures from DeepMind and a delay in the release of Gemini 3.5 Pro.
Why it matters
For developers, a more competitive Gemini could mean better tools and more advanced capabilities. Startups may also benefit from improved access to cutting-edge AI technologies.
Investors will be watching closely to see if Kavukcuoglu can steer DeepMind to catch up with OpenAI and Anthropic. The success of Gemini is crucial for Google's AI strategy and its ability to compete in the lucrative AI market.
However, the shift in focus towards frontier models may impact DeepMind's broader research agenda, including AGI and other long-term projects. It also raises questions about Google's commitment to its UK-based AI ecosystem, as Kavukcuoglu moves to Mountain View, California.
Key facts
- Kavukcuoglu joined DeepMind in 2012, before its acquisition by Google in 2014.
- DeepMind was based in London before its acquisition and remained there under Hassabis' leadership.
- Google's last frontier model, Gemini 3, was released in November 2025.
- OpenAI and Anthropic have released advanced models like GPT-5.6 and Mythos in recent months.
- Google delayed the release of Gemini 3.5 Pro in July to improve performance in areas like coding.
- Nearly 90% of Fortune 100 companies are using Gemini Enterprise, according to Alphabet CEO Sundar Pichai.
- Alphabet shares have risen around 76% in the last 12 months, despite a drop on Thursday following the AI reshuffle announcement.
Context
The AI industry is currently in a race to develop the most advanced models, with OpenAI and Anthropic setting the pace in recent months. Google's acquisition of DeepMind in 2014 was seen as a major move to strengthen its AI capabilities, but the company has struggled to keep up with the latest developments.
The shift in focus towards improving Gemini's performance and developer tools could have significant implications for the AI industry. It may also impact Google's broader AI strategy, including its efforts to develop AGI and commercialize its AI technologies.
The appointment of Kavukcuoglu comes at a critical time for Google, as it seeks to maintain its position as a leader in the AI industry. The success of Gemini is crucial for Google's AI strategy and its ability to compete with OpenAI and Anthropic in the lucrative AI market.
