Джефф Дин покинул Google: что стоит за перестановками в AI-команде DeepMind

In early August 2026, Google DeepMind experienced one of the most sweeping personnel reshuffles in the AI division's entire history: several top executives received new roles simultaneously. Among those who left Google is Jeff Dean, a legendary engineer with more than twenty-five years at the company.

Who Left Google and Why It Matters

Jeff Dean is one of the most renowned technical architects in Silicon Valley. During his years at Google he participated in building MapReduce and Bigtable — foundational distributed-computing technologies that still underpin most AI systems today — as well as TensorFlow, the most popular neural-network training framework of the 2010s. Dean later led Google Brain, the lab that trained many of the researchers now working at OpenAI, Anthropic, and other leading AI companies around the world.

According to The Verge, Dean's new role is outside Google. A number of other top AI-division executives left alongside him — the full list had not been disclosed at the time of publication.

• Jeff Dean — co-creator of MapReduce, Bigtable, TensorFlow, and Google Brain
• Several key positions in the AI team changed simultaneously
• New roles for some of the executives are outside Google
• Demis Hassabis, CEO of Google DeepMind, retains his position

For the industry, the scale of the reshuffle came as a surprise: engineers of Dean's caliber leave one of the world's largest technology companies once in a decade.

Is Google Falling Behind in the AI Race?

The changes are happening under maximum competitive pressure. The Verge's editors state plainly that Google's models "appear to be lagging behind the best products from Anthropic and OpenAI." In 2025–2026, the flagship models from Anthropic (Claude) and OpenAI (GPT-5) repeatedly outperformed Gemini on independent benchmarks — especially in reasoning and code-writing tasks, which have become the primary measure of AI progress.

"Is this a sign of chaos inside Google?

Or does Demis Hassabis simply want to work on something more interesting than virtual assistants?" — Verge editors Nilay Patel and David Pierce frame the key question on the Vergecast podcast.

Both scenarios are plausible and not mutually exclusive. Hassabis built DeepMind as a fundamental-research laboratory focused on AGI: AlphaFold made a breakthrough in biology, AlphaGo changed the world of board games. Consumer AI assistants are a completely different kind of work, with different KPIs and a different development logic. If the CEO wants to return to science, the team inevitably reorganises around new priorities.

On the other hand, mass simultaneous departures of experienced AI engineers are a classic symptom of a platform losing its lead. Competitors at Anthropic and OpenAI offer not only money, but also the feeling of being at the cutting edge of the race.

What This Means

The changes at Google DeepMind are happening at a pivotal moment for the entire AI industry. If this is a deliberate strategic reorganisation, the company is regrouping and may emerge from the restructuring stronger. If it is a symptom of a systemic crisis, Google risks falling out of the front rank precisely when Anthropic and OpenAI are accelerating their release pace. The answer will become clear with the next versions of Gemini: a gap in model quality cannot be hidden behind personnel reshuffles.