Alphabet shares fell as much as 7.2% intraday Monday, closing at $346.30 and down roughly 5.9% on the session, after John Jumper, a vice president at Google DeepMind and co-recipient of the 2024 Nobel Prize in Chemistry, told colleagues over the weekend that he was leaving for Anthropic. It was Alphabet’s worst intraday session since February, and it landed five days after Gemini co-lead Noam Shazeer told colleagues he was heading to OpenAI. Two résumés, five days, one stock chart.

The market is reading this as a referendum on something larger than two hires.

Jumper isn’t a generic AI executive. He’s the co-architect of AlphaFold, the system that has predicted more than 200 million protein structures and shared the 2024 Nobel in Chemistry with DeepMind chief Demis Hassabis. Shazeer is one of the original authors of the transformer architecture that essentially defines the current AI era. Losing either would sting. Losing both inside a week, to the two labs Alphabet is most directly trying to outspend, is a different category of event.

“Losing John is a big loss for Google and there is no way to sugarcoat it,” Wedbush analyst Dan Ives said. D.A. Davidson’s Gil Luria went further, “raising the concern that Google is losing the war for talent at the frontier of AI.”

That concern has a price tag attached. Alphabet has raised roughly $141 billion through debt and equity financing since October to fund its AI ambitions, and just unveiled Gemini 3.5 Flash and the Gemini Spark agent at I/O. Capex is committed. The humans executing it are walking out.

There’s also a recursive quality to Shazeer’s exit. He and Daniel De Freitas left Google in 2021 to found Character.AI, then were brought back in August 2024 through a partnership deal Google explicitly framed as a talent reacquisition. Less than two years later, Shazeer is gone again, this time to the competitor.

Alphabet’s official line, given to Barron’s: “We are grateful for John’s significant contributions to Google DeepMind’s work in advancing science and AI. We wish him well in his next chapter.” The stock is down roughly 9% on the month, though still 11% year to date, and Tuesday’s 2.21% Nasdaq slide, amplified by an SK Hynix and Samsung-led memory selloff, gave the narrative more room to run.

The unresolved question isn’t whether Google can build frontier models. It’s whether the people who know how to build them want to do it there.

Sources