Google Just Paid $2.7B for an Engineer Who Just Quit for OpenAI
In a move that defines 'expensive hobby,' Google shelled out billions to bring Noam Shazeer home, only to watch him sprint straight to OpenAI. It’s a masterclass in how to burn cash while simultaneously handing your crown jewels to the competition.
Noam Shazeer, the tech world’s favorite wanderer, has officially joined OpenAI. The timing is comedic: less than two years ago, Google spent a staggering $2.7 billion to acquire his startup, Character.AI, essentially buying him back to lead the development of Gemini.
Shazeer isn't just another engineer; he is one of the eight architects behind the 2017 paper Attention Is All You Need. He didn't just contribute; he physically engineered the multi-head attention mechanisms that make today's LLMs possible. His resume is basically a roadmap of modern AI, including inventions like Mixture-of-Experts and multi-query attention, the latter of which makes your AI chatbot responses actually affordable to run.
His history with Google is a rollercoaster. He originally left in 2021 after the company refused to release Meena, a chatbot he built, citing safety concerns. Google eventually realized that while they were busy playing it safe, OpenAI was busy building ChatGPT on the very architecture Shazeer pioneered. By the time they bought him back to fix Gemini, they were effectively paying billions for the privilege of keeping their own original ideas from totally disappearing into the competition's pocket.
This isn't just a simple resignation; it is a billion-dollar slap to the face of corporate caution. While Google sits on a mountain of R&D, they seem incapable of keeping the actual builders from defecting to the company currently winning the race. The talent war is no longer about salaries; it's about who gets to hold the leash of the people who actually invented the future.
Source: CNBC
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.