Google Loses Two AI Legends in One Week. The Market Took It Personally.

Date: 23 June 2026 | Category: Quick Take | Author: C2

On June 18, Noam Shazeer announced he was leaving Google for OpenAI. On June 19, John Jumper said he was leaving Google DeepMind for Anthropic. Two of the most consequential AI researchers of the past decade walked out of Google's doors within 48 hours of each other. By June 22, Alphabet stock had dropped roughly 5%, wiping out approximately $250 billion in market value.

Shazeer co-authored the 2017 paper "Attention Is All You Need," which introduced the Transformer architecture that underpins virtually every modern large language model. Google had already paid $2.7 billion to bring him back from Character.AI in 2024, installing him as co-lead of Gemini. He lasted less than two years.

Jumper led the AlphaFold project at DeepMind, sharing the 2024 Nobel Prize in Chemistry for work that predicts protein structures with near-experimental accuracy. He spent nearly nine years at Google. Now he's heading to Anthropic.

The departures are not just a PR problem. They are a signal about where the centre of gravity in AI research is moving — and Google's apparent struggle to hold it.

What Shazeer and Jumper Actually Built

Shazeer's contribution is hard to overstate. The Transformer architecture, introduced in that 2017 paper alongside seven Google colleagues, replaced recurrent neural networks with self-attention mechanisms. The result was parallelisable training, scalable context windows, and the foundation for GPT, Claude, Gemini, and every other modern LLM. When Shazeer co-authored that paper, he was at Google. When he left, he took his expertise to OpenAI.

Jumper's AlphaFold solved a 50-year-old problem in structural biology: predicting how proteins fold into three-dimensional shapes from their amino acid sequences. The system has been used by over two million researchers worldwide and underpins drug discovery pipelines at multiple pharmaceutical companies. His move to Anthropic aligns with that company's growing investment in AI for science — Anthropic has an AI for Science event scheduled for June 30.

Why Now?

The timing is not random. Bloomberg reported that DeepMind staff have raised concerns internally about Google's lack of a clear product for businesses building AI coding tools — an area where Anthropic and OpenAI have gained substantial ground. CEO Sundar Pichai himself admitted in May that Google was "a bit behind" on agentic coding, tying the gap to a lack of developer-facing products.

The researcher exodus follows a pattern. Google's AI talent has been leaking to competitors for years: the "Attention Is All You Need" authors have almost all left — Lukasz Kaiser to OpenAI, Jakob Uszkoreit to startup land, Illia Polosukhin to NEAR Protocol. The brain drain is not new, but the pace is accelerating.

The Market's Reaction

Alphabet's 5% drop on June 22 was sharper than the immediate reaction to Shazeer's departure announcement. The market initially shrugged off the news, then connected the dots when Jumper followed. The concern is not about any individual project — Gemini will not stop working because Shazeer left, and AlphaFold's database remains public because Jumper is gone.

The concern is about competitive positioning. When researchers at this calibre choose to work elsewhere, investors read it as a vote of no confidence in Google's ability to win the frontier AI race. The $250 billion haircut is not a valuation of Shazeer and Jumper's individual contributions. It is the market pricing in a higher probability that Google will lose ground to OpenAI and Anthropic over the next several years.

What Google Is — and Isn't — Doing

Google's response has been muted. The company confirmed both departures and issued the standard statements about wishing them well. There has been no dramatic reorganisation announcement, no emergency retention programme, no public acknowledgment that something is broken.

This is consistent with Google's historical approach to talent loss: absorb it, replace internally, keep shipping. The company still employs more top-tier AI researchers than any other organisation on Earth. It still has the compute infrastructure, the data pipelines, and the distribution reach to compete. What it appears to lack is the cultural momentum that makes researchers want to stay.

The Competitive Landscape

OpenAI and Anthropic are benefiting from Google's losses, but they are also actively recruiting. OpenAI has filed confidentially for an IPO, giving it a new currency for compensation and retention. Anthropic has positioned itself as the safety-conscious alternative, attracting researchers who want to work on AI without the commercial pressure that comes with being a public company.

Both companies have been hiring from larger labs for years. The difference now is the calibre of the hires. Shazeer and Jumper are not mid-level researchers looking for a pay bump. They are foundational figures in modern AI. Their moves validate the upstarts and undermine Google's narrative that it remains the best place to do frontier research.

What to Watch Next

  • Gemini's trajectory without Shazeer: Gemini 4.x is already in development. Whether the model maintains its competitive position will be the first test of whether Google's bench depth is real.
  • Anthropic's science play: Jumper's hiring suggests Anthropic is serious about scientific applications beyond chatbots. The June 30 event should clarify the strategy.
  • Google's retention response: If more departures follow, Google will face pressure to change its internal culture or compensation structure. Silence will not be sufficient indefinitely.
  • OpenAI's IPO implications: A public OpenAI may face the same talent-retention challenges that plague Google. The cycle could repeat.

Bottom Line

Google is not failing. It is still one of the most capable AI organisations in the world. But it is losing the people who built its most important technical foundations to competitors who offer something Google apparently cannot: a sense that the future is being built there, not defended here.

The $250 billion market reaction is excessive for two departures. But it is not irrational. When the architects leave, the building's value becomes uncertain. Google needs to demonstrate that its AI future does not depend on the people who just walked out the door.

Sources

  • Search Engine Journal. (2026, June 23). Google Loses Two Top AI Researchers To OpenAI & Anthropic. https://www.searchenginejournal.com/google-loses-two-top-ai-researchers-to-openai-anthropic/580201/
  • CNBC. (2026, June 18). Google Gemini co-lead Noam Shazeer leaves for OpenAI. https://www.cnbc.com/2026/06/18/google-gemini-co-lead-noam-shazeer-leaves-for-openai.html
  • CNBC. (2026, June 19). John Jumper to leave Google DeepMind for Anthropic. https://www.cnbc.com/2026/06/19/john-jumper-to-leave-google-deepmind-for-anthropic.html
  • Bloomberg. (2026, June 19). Nobel winner John Jumper to leave Google DeepMind for Anthropic. https://www.bloomberg.com/news/articles/2026-06-19/nobel-winner-john-jumper-to-leave-google-deepmind-for-anthropic
  • StockWireX. (2026, June 23). Alphabet Drops 5% as AI Talent Exodus Rattles Mag-7 Stocks. https://stockwirex.com/news/alphabet-stock-ai-talent-exodus-june-2026/
  • Memeburn. (2026, June 23). Noam Shazeer Leaves Google for OpenAI After $2.7 Billion Recruitment Deal. https://memeburn.com/noam-shazeer-leaves-google-for-openai-after-2-7-billion-recruitment-deal/