Google Loses Two Foundational AI Researchers in a Week as Alphabet Stock Slides 7.2%

In the span of 72 hours in June, Google lost two of the most consequential names in modern artificial intelligence — and Wall Street did not wait to render its verdict. On June 18, Noam Shazeer, a co-author of the foundational "Attention Is All You Need" paper and a co-lead of the Gemini program, announced he was leaving for OpenAI. Two days later, John Jumper, the DeepMind scientist who shared the 2024 Nobel Prize in Chemistry for AlphaFold, said he was decamping for Anthropic. By Monday, June 22, Alphabet shares had fallen as much as 7.2% intraday, the stock's steepest single-day slide since February 2026 and its worst trading day in more than a year.

The back-to-back exits did more than dent morale inside Mountain View. They erased roughly $250 billion in market value from a company worth about $4.46 trillion, and they handed Google's two fiercest rivals a pair of marquee hires at the precise moment the competitive stakes have never been higher. Both OpenAI and Anthropic are racing toward hotly anticipated public offerings, and both have spent 2026 trying to pull ahead in a frontier-model contest that Google, until recently, looked positioned to win.

Two departures, one message

Shazeer's move is the more symbolically loaded of the two. He is arguably the single most important engineer in the lineage of the large language model — the Transformer architecture he helped invent in 2017 underpins virtually every modern AI system, including the ones built by the company he is now joining. His departure is also expensive: Google paid roughly $2.7 billion less than two years ago to bring Shazeer and part of his team back from Character.AI, a deal that was understood internally as a long-term bet on Gemini's leadership. OpenAI chief executive Sam Altman welcomed the hire publicly, calling Shazeer one of the people he had most wanted to work with since OpenAI's earliest days.

Jumper's exit cuts in a different direction. Where Shazeer represents the core of Google's language-model effort, Jumper embodies its prestige in AI-for-science — the AlphaFold work that solved protein structure prediction and earned a Nobel. "After nearly 9 years, I have decided to leave Google DeepMind and join Anthropic," Jumper wrote on X, adding that DeepMind chief executive Demis Hassabis "took a real chance letting me lead the AlphaFold team just six months after finishing my PhD." Hassabis responded graciously, writing that "what we achieved with AlphaFold changed the world, and showed the field what was possible with AI for science and medicine, lighting the way for how AI can benefit humanity."

The timing was not lost on investors. Anthropic has spent the year assembling the apparatus for serious scientific work — opening wet labs, publishing research on AI agents built for biological workflows, and announcing partnerships with the Allen Institute and the Howard Hughes Medical Institute. Hiring the field's most decorated protein-folding scientist is the keystone of that strategy. For Google, losing him reads less like ordinary attrition and more like a defection to a thesis Google itself pioneered.

A retention problem the market can price

What unsettled traders was not any single hire but the pattern. By several accounts, four senior AI researchers left Google for OpenAI and Anthropic in roughly the same window. That cadence turns an HR story into a strategic one: the question is no longer whether Google can attract elite talent — it manifestly can, and has paid billions to do so — but whether it can keep that talent once the broader market is offering richer equity, looser bureaucracy, and the gravitational pull of pre-IPO upside at OpenAI and Anthropic.

The stock reaction suggests the market now treats Google's AI personnel as a material risk factor. A 7.2% intraday move in a $4.46 trillion company is not a rounding error; it is the market repricing the durability of Google's research advantage in real time. CNBC characterized June 22 as Alphabet's worst day in over a year, explicitly tied to AI departure concerns rather than to earnings or macro news.

That anxiety is amplified by a product problem. At Google I/O on May 19, the company committed to launching Gemini 3.5 Pro into general availability in June. As of this writing, that has not happened — reporting indicates the release has slipped to July, with Google citing quality refinements after early enterprise testing. A delayed flagship is survivable on its own. A delayed flagship arriving the same week two foundational researchers walk out the door is a narrative Google cannot easily control, because it invites the inference that the delay and the exodus share a root cause: an organization straining to convert its deep research bench into shipped product at the pace its rivals are setting.

What to watch

The near-term tell is Gemini 3.5 Pro. If general availability lands in July and the model benchmarks competitively against OpenAI's and Anthropic's latest, the talent story softens into a footnote and the stock likely recovers the bulk of its losses. If it slips again, expect analysts to sharpen the question of whether Google's research-to-product pipeline is structurally slower than its competitors'.

The second tell is whether the departures stop at four. Google's bench remains extraordinarily deep, and Hassabis's DeepMind retains many of the architects of the company's most important systems. But talent moves in cascades, and the same equity-and-autonomy math that pulled Shazeer and Jumper applies to dozens of their colleagues. The third is how aggressively Google responds — whether with another nine-figure retention package, a reorganization of its AI leadership, or a louder product cadence designed to remind the market why those researchers chose Google in the first place. For now, the scoreboard reads two foundational losses, one delayed launch, and roughly $250 billion in vanished market value — all in a single week.