Alphabet’s Top AI Researchers Flee as Free Cash Flow Collapses
What You Need to Know
- Noam Shazeer and John Jumper, key Google AI researchers, departed for OpenAI and Anthropic respectively.
- Google’s talent retention struggles amid intense competition in foundation model development.
- Alphabet’s capital expenditures doubled to $35.7 billion in Q1, halving free cash flow year-over-year.
- Polymarket estimates Google has only 13% odds of having best AI model by July end.
Alphabet’s worst single-day drop in over a year was not caused by a missed earnings print or a regulatory ruling. It was caused by two researchers walking out the door.
Noam Shazeer, co-lead of Google’s Gemini models who had returned to the company in 2024 as part of a partnership arrangement, announced he was leaving for OpenAI. Days later, Nobel Prize winner John Jumper confirmed he was departing Google’s DeepMind lab for Anthropic. Together, the exits crystallized a concern that has been building for months: Google can attract talent, but it is struggling to keep it. The timing matters because the competitive window in foundation models is compressing fast, and the researchers who understand how to close that gap are the ones now leaving. Polymarket traders currently put Google’s odds of having the best AI model by end of July at just 13%, which reflects sentiment more than certainty but is still a telling number.
Personnel risk is one problem. The capital expenditure math is a separate, arguably larger one.
Alphabet has guided for $180 to $190 billion in spending during fiscal 2026, overwhelmingly on data centers and AI compute. Against that backdrop, first-quarter operating cash flow of $45.8 billion sounds comfortable until you see that capital expenditures more than doubled to $35.7 billion in the same period, cutting free cash flow from over $19 billion a year earlier to just above $10 billion. Microsoft’s Satya Nadella has publicly suggested models may become cheaper and easier to swap out, which reframes the entire capex argument: if the models commoditize, the infrastructure spend does not automatically translate into durable margin. The market is now pricing that possibility. As one analyst framed it, there is a widening gap between AI spenders and AI earners, and the hyperscalers are firmly in the first category.
The sell-off spread well beyond Alphabet. ASML fell 5%, Samsung Electronics and SK Hynix each dropped 12%, and SpaceX shares fell 16% on Monday. The correlation is significant because it suggests investors are not just reassessing Google specifically but repricing the entire infrastructure bet that has driven the AI investment cycle over the past two years. Open-source Chinese models like DeepSeek applying cost pressure from below, while ChatGPT surpasses one billion monthly active users above, leaves Google squeezed on both ends of the market it built.
0 Comments