Alphabet is restructuring DeepMind's leadership, moving Nobel laureate Demis Hassabis from day-to-day CEO duties to a strategic role as chairman and Alphabet's chief scientific officer. The shift separates DeepMind into tracks for long-term research and product commercialization, with CTO Koray Kavukcuoglu taking over Gemini development and operations. The reorganization comes as Alphabet raises its capital expenditure outlook to as much as $205 billion and reports a $5.9 billion free cash flow deficit for the second quarter.
The move signals a deliberate recalibration: Alphabet wants to keep pursuing long-term research like artificial general intelligence while accelerating Gemini's path to market. Hassabis will focus on AGI strategy and continue leading Isomorphic Labs, the AI drug discovery subsidiary he founded. Kavukcuoglu, who relocated from London to Google's Mountain View headquarters last year, will oversee Gemini model development and the developer organization.
Research culture meets commercial pressure
The structural change reflects growing tension between DeepMind's research-first identity and the financial demands of AI development. The Financial Times reported that Google is centralizing AI decision-making at headquarters to speed Gemini's commercialization and close the gap with OpenAI and Anthropic.
Hassabis framed the move internally as a shift toward the "big picture" rather than a retreat from research. He has said on X that "we are only a few years away from AGI."
The reorganization also comes with visible talent movement. Jeff Dean, Google's chief scientist for 27 years, is leaving alongside other AI researchers to found "Discovery Loop," a public-interest startup. His departure follows internal concerns that DeepMind's research focus could be diluted under commercial pressure.
For executives tracking AI for Executives & Strategy, the key signal is how Alphabet is balancing these competing priorities.
The agent shift and business metrics
The competitive context matters here. AI competition is moving beyond frontier model performance into agents that execute tasks autonomously - coding, scientific research, and workflow automation. That shift favors organizations that can ship products quickly, not just publish papers.
Alphabet is emphasizing that Gemini already connects to business results. The Gemini app reached 950 million monthly active users in the second quarter. More than 9 million developers use Google's AI models monthly, and the company's model API processes roughly 22 billion tokens per minute, up 37.5% from the previous quarter's 16 billion. Cloud revenue grew 82% year over year to $24.8 billion.
The company is also investing heavily to sustain that growth. Alphabet raised its capex outlook to $195-$205 billion for the year, up as much as $30 billion from the earlier $175-$185 billion range. The spending funds data centers and servers for AI and cloud demand, but it pushed free cash flow to a $5.9 billion deficit in Q2.
This is where the strategy around AI Agents & Automation becomes relevant: the competition is shifting from research milestones to real-world deployment.
What the reshuffle means
Some reports suggest Hassabis considered leaving alongside Jeff Dean before being persuaded to stay in an adjusted role. Others note that DeepMind now functions more like a standard business unit within Alphabet than an independent research lab. These accounts remain unconfirmed, but the official changes are clear: Hassabis moves to strategy, Kavukcuoglu runs operations, and key research talent is exiting.
Search revenue still grew 17% in the quarter, undercutting earlier fears that generative AI would cannibalize Google's core business. That result, combined with cloud growth, gives Alphabet a defensible case that its AI investments are paying off - even as the spending creates short-term financial strain.
Why this matters for executives and strategy
The DeepMind reorganization is a case study in how AI leaders are restructuring for the next phase of competition. Research prestige alone no longer justifies massive capital outlays; investors want to see models converted into products, revenue, and market share. Alphabet's structure now separates the two mandates explicitly: Hassabis owns the long horizon, Kavukcuoglu owns the quarterly numbers.
For executives evaluating AI investments, the lesson is about governance. Alphabet is separating research ambition from commercial execution, tying the latter to clear business metrics, and betting that the agent shift rewards companies with distribution and product velocity. The open question is whether that structure can retain the research talent that made DeepMind valuable in the first place.
Your membership also unlocks:
