Google Lost the Transformer Inventor and a Nobel Laureate in One Week

📖 5 min read

Google just had its worst trading day in over a year. Shares of Alphabet fell 7.2% on Monday, June 22 – wiping out tens of billions in market cap in a single session. The cause wasn’t a bad earnings report or a regulatory ruling. It was two resignation letters.

In the span of four days, Google lost two of the most important AI researchers in the world: Noam Shazeer, co-inventor of the Transformer architecture that powers virtually every modern AI system, and John Jumper, the Nobel Prize-winning scientist behind AlphaFold. Shazeer is going to OpenAI. Jumper is going to Anthropic. The AI talent war just escalated to a level nobody saw coming.

The Numbers That Tell the Story

Event	Date	Impact
Google acquires Character.AI to bring Shazeer back	September 2024	$2.7 billion paid
Shazeer announces departure to OpenAI	June 18, 2026	20 months later
Jumper announces departure to Anthropic	June 19, 2026	9 years at DeepMind
Alphabet stock drops	June 22, 2026	-7.2% (worst day in ~1 year)
Google 2026 AI capex plans	Ongoing	$180-190 billion projected

To understand why $2.7 billion wasn’t enough to keep Shazeer, you need to understand who he is. In 2017, Shazeer was one of eight co-authors on a paper called “Attention is All You Need.” That paper introduced the Transformer – the architecture behind GPT-4, Gemini, Claude, and almost every other major AI model in existence today. He’s not just important to Google. He’s one of the people who built the foundation the entire AI industry runs on.

After leaving Google in 2021 to co-found Character.AI (the AI chatbot company), Shazeer and his team were brought back in September 2024 when Google acquired Character.AI’s technology and key personnel for $2.7 billion. He became VP of Engineering and co-lead of Google’s Gemini AI models. Less than two years later, he’s heading to OpenAI – the company preparing for its own IPO.

OpenAI CEO Sam Altman didn’t try to be subtle about the significance of the hire. He publicly called Shazeer “one of the people I’ve most wanted to work with.”

The Nobel Prize Winner Who Chose a Startup

If Shazeer’s departure was a punch, John Jumper’s was the follow-up. Jumper spent nearly nine years at Google DeepMind, leading the team behind AlphaFold2 – the AI system that cracked one of biology’s most stubborn problems by accurately predicting the 3D structure of proteins. That work earned him the 2024 Nobel Prize in Chemistry.

He’s now going to Anthropic.

Anthropic is no longer the scrappy safety-focused spinoff it was when Dario Amodei and others left OpenAI in 2021. Recent funding discussions have reportedly valued the company near $1 trillion. D.A. Davidson analyst Gil Luria put it bluntly: “The race at the frontier right now appears between Anthropic and OpenAI.”

That framing is striking given that Google has poured more money into AI infrastructure than either competitor. Alphabet announced plans to raise $80-85 billion through equity offerings to fund AI capital expenditures, with total 2026 AI spending projected at $180 billion to $190 billion. For comparison, that’s more than the GDP of most countries.

What Money Can’t Buy

The obvious question: why does any of this happen if Google is paying billions?

The answer isn’t just about compensation – frontier AI researchers at this level are compensated extraordinarily well everywhere. It’s about perceived momentum, equity upside, and where you think the most interesting work is being done. Shazeer co-invented the Transformer at Google, then left to build his own company. When Google brought him back for $2.7 billion, it looked like a lock-in. But researchers at the frontier care about being at the cutting edge – and right now, the cutting edge increasingly looks like it’s at OpenAI and Anthropic, not Google.

Alphabet still has zero analyst sell ratings and reported Q1 2026 earnings growth of 82%. Google Cloud revenue surged 63% to $20 billion in the same quarter. From a business perspective, Google is not failing. But in the specific, narrow competition for the top tier of AI research talent, the departure of two people in four days – one for $2.7 billion who still left – sends a signal that’s hard to spin.

What This Means for the AI Race

The competitive dynamics have shifted. Six months ago, the conventional wisdom was that Google’s massive compute infrastructure and internal talent gave it an insurmountable advantage in training frontier models. That assumption is now under scrutiny.

OpenAI is reportedly weeks away from releasing GPT-5.6, and Anthropic is gearing up to launch Claude Sonnet 5. Both companies have just added serious firepower to their research teams. Shazeer’s architectural expertise and Jumper’s scientific credibility are exactly the kinds of additions that accelerate model development and attract further talent.

For Google, the challenge isn’t just replacing two individuals – it’s addressing whatever is making top researchers choose to leave despite being very well compensated to stay. That’s a cultural and organizational question, not a financial one, and those are harder to fix.

The Talent War in Numbers

Researcher	From	To	Why They Matter
Noam Shazeer	Google (Gemini co-lead)	OpenAI	Co-invented the Transformer (2017)
John Jumper	Google DeepMind	Anthropic	2024 Nobel Prize, AlphaFold

Both moves happened within 48 hours of each other. Alphabet stock fell 7.2% the next trading day. Whether that market reaction is proportionate or overblown is debatable. What’s not debatable: losing the co-inventor of the Transformer and a Nobel laureate in the same week is a signal the AI market is paying close attention to.

BetOnAI Verdict

Google still has the infrastructure, the compute, and the talent pipeline to compete at the frontier. An 82% earnings growth quarter and $20 billion in Cloud revenue don’t disappear because two researchers left. The 7.2% stock drop likely overestimates the short-term impact.

That said, this isn’t just a one-week story. It’s evidence of a structural issue: Google’s culture and incentive structure may not be optimized to retain the highest-tier AI researchers long-term, especially when competitors offer equity upside tied to companies that feel more like rocketships right now than a $2 trillion tech conglomerate.

If you’re watching the AI race, watch what happens at OpenAI and Anthropic over the next 6-12 months. Shazeer’s architectural knowledge and Jumper’s scientific credibility are the kinds of additions that compound. The models those teams ship in the next year will tell us whether this talent shift actually changes the frontier – or whether Google’s scale absorbs the loss.

Right now, the honest bet is that the gap between Google and its top two competitors just got a little smaller.

---

Sources:

• CNBC – Google Gemini co-lead Noam Shazeer leaves for OpenAI
• Reuters – Google’s Gemini co-lead Noam Shazeer to join IPO-bound OpenAI
• Bloomberg – Alphabet Shares Drop After Second AI Star Departs for a Rival
• Eciks – Google stock falls 7% as AI talent exits to OpenAI, Anthropic