Article
AI Ethics AI News

Big Tech's AI stars are leaving for a payday only an IPO can offer

by TechDefused Newsroom
The image features a large illuminated star sculpture against a dark night background. The star is composed of glowing neon lights, creating a striking visual effect. — Credit: Photo by B R on Unsplash c Photo by B R on Unsplash

Talent moves between technology companies all the time. This past week was different in kind. The departures cluster around a single financial event that Google and Apple cannot replicate. Both OpenAI and Anthropic are approaching public listings. For a senior researcher, joining before the bell offers equity upside that no mature, already-public employer can match.

Google lost four names in days. Noam Shazeer, co-author of the Transformer paper that launched the current AI boom, left for OpenAI. John Jumper, a 2024 Nobel laureate who led AlphaFold, went to Anthropic. Two more Gemini contributors, Jonas Adler and Alexander Pritzel, followed him there. At Apple, Paul Meade, the vice president who built the Vision Pro and ran its smart glasses effort, starts at OpenAI's hardware unit next week.

These are not the cash raids of a year ago. Meta pulled talent then with pay alone. The new lever is ownership in a company about to be repriced by public markets. Cash cannot compete with a claim that reprices overnight.

Equity a public company cannot print

The common thread is timing. Anthropic recently raised at a reported $965bn valuation and is weighing a listing as soon as the autumn. OpenAI is on a similar path. A tenured engineer at Alphabet already holds liquid stock. The same person at a pre-listing lab holds a claim that could multiply the day it floats. That gap is the entire story, and it runs on a clock.

Two firms gave their own reasons to leave

The pull is half of it. Both Google and Apple supplied a push. At Google, insiders describe a culture grown slow and risk-averse, a charge Shazeer made years ago. Compute has become a flashpoint. Before his exit, capacity tied to one of his projects was reassigned to a London team. At Apple, the trigger was structural. John Ternus became chief executive in September. Johny Srouji took over hardware and reordered the engineering ranks. Several vice presidents, Meade among them, dropped a level, and some felt demoted. At OpenAI, Meade joins former Apple colleagues Jony Ive, Tang Tan and Evans Hankey, whose hardware startup the company bought last year for $6.5bn.

Why the moves will keep coming

Expect more. Three forces point the same way. The equity window stays open for months, so the incentive holds until these firms list. Defections feed on themselves. Each one lowers the cost of the next and signals internal doubt to those still inside. And Apple's reorganisation is fresh, its fallout still moving through the ranks.

The pattern reaches beyond Google and Apple. Some researchers are not switching labs but founding their own. David Silver, one of DeepMind's earliest hires, left to start a company. Nvidia absorbed the team behind Essential AI.

There are brakes. Many DeepMind researchers sit in the United Kingdom, where non-compete clauses bind and can delay a move for months. Google keeps the deepest research bench in the field, and chief Demis Hassabis says the lab wins its share of top talent. One 2025 study found DeepMind staff were almost 11 times more likely to leave for Anthropic than the reverse. That ratio now looks kind to Google.

The bill arrives in public

Investors have noticed. Alphabet shares fell as much as 6% when the Shazeer and Jumper moves landed, before recovering. The market reads each exit as a verdict on whether Google can hold the people who built its lead.

The window shuts once OpenAI and Anthropic list. Until then, every researcher at a public AI employer faces the same sum. The maths favours leaving.

by TechDefused Newsroom