Editorial illustration for DeepMind’s Hiring Momentum Slows Under New Chief
DeepMind's Hiring Slows Under New CEO Kavukcuoglu
Koray Kavukcuoglu has been running Google DeepMind for three months, and his first public appearance as CEO signaled a shift in what the lab is chasing. Speaking at The Information's AI Agenda Live Summit, Kavukcuoglu said Gemini 4 is already in early post-training, the stage where a base model gets tuned for reliable behavior. He wants it out well before year's end, faster than the usual release cadence, with fast iterations to follow once safety testing wraps up.
That's a different posture than the one DeepMind built its reputation on. Kavukcuoglu took over in August after Demis Hassabis stepped down as CEO to chair the unit and serve as Alphabet's chief scientist. Hassabis spent years framing DeepMind's mission around artificial general intelligence.
Kavukcuoglu, previously the company's CTO, has a more immediate concern: getting a competitive model out the door while OpenAI and Anthropic keep shipping. Engineers inside Google are already testing the unreleased model on the Antigravity coding tool, a sign of how much internal pressure is riding on this launch.
Why this matters
The hiring ratio tells the real story here: 12 new researchers for every departure in mid-2023, down to 2-to-1 by late 2026. That's not a lab losing talent randomly, it's a lab losing the kind of talent that joins for open-ended research and leaves when the job turns into shipping quarterly releases. For developers and founders building on Gemini, this is probably good news short-term.
A chief who calls the AGI conversation "not the right conversation" and fixates on getting Gemini 4 out the door is optimizing for the thing you actually consume: working models, on schedule, competitive with OpenAI and Anthropic. For researchers, especially anyone who joined Deepmind under Hassabis's mission framing, it's a signal to recalibrate expectations about what kind of work gets funded internally now. Google made a bet that product cadence matters more than research prestige in this market.
Worth watching whether that 2-to-1 ratio keeps sliding, and whether Gemini 4's release quality justifies the tradeoff, or whether Deepmind just becomes another feature factory with a research pedigree.
Related Reading
Browse all Google news →
Research & Benchmarks Microsoft AI Chief: "If It Isn't Safe We Shouldn't Build It
Research & Benchmarks Google's FACTS benchmark shows 70% factuality ceiling across four tests
Research & Benchmarks Nvidia's DLSS 4.5 beta adds 6x Multi Frame Generation for RTX 50 GPUs
Related Topic NVIDIA and Google Cloud let developers scale AI from prototype to production
Market Trends
Related Topic Google tests visual 'magazine-style' UI for Gemini 3 Pro users
LLMs & Generative AI
From Archives Visual AI Links City Scenes to Urban Life Experience
From Archives AI Agents Choose Papers From 114 Economics Research Articles
From Archives Google DeepMind Chief Says Gemini 4 AI Model Arriving Before 2026
LLMs & Generative AI
From Archives YouTube's Gemini AI Expands to Full Editing Tool in Creator Studio
LLMs & Generative AI
Common Questions Answered
What is Koray Kavukcuoglu's timeline for releasing Gemini 4?
Kavukcuoglu wants to release Gemini 4 well before the end of the year, which is faster than DeepMind's usual release cadence. He plans to implement fast iterations once safety testing is completed, signaling a shift toward more rapid product deployment under his leadership.
How does Kavukcuoglu's focus differ from his predecessor Demis Hassabis?
While Hassabis focused heavily on the AGI question, Kavukcuoglu has stated that the AGI conversation is 'not the right conversation' to prioritize. Instead, he is directing DeepMind's efforts toward shipping Gemini products and competing with OpenAI and Anthropic in the near term.
What does the hiring ratio reveal about DeepMind's changing priorities?
DeepMind's hiring ratio dropped from 12 new researchers for every departure in mid-2023 to 2-to-1 by late 2026, indicating the lab is losing talent that was attracted to open-ended research. This shift suggests that researchers are departing when the focus turns to shipping quarterly product releases rather than pursuing fundamental AI research.
How has Google DeepMind's mission changed under new leadership?
Under Kavukcuoglu, Google's former research lab has transformed into what the article describes as 'a Gemini factory,' prioritizing the development and release of Gemini products over long-term AGI research. This represents a strategic pivot driven by competitive pressure from companies like OpenAI and Anthropic.
Further Reading
- Papers with Code Benchmarks - Papers with Code
- Chatbot Arena Leaderboard - LMSYS
