What do you give the AI engineer who has everything? Maybe the freedom to do whatever they want.
While the AI talent wars have been defined by multimillion-dollar pay packages and multibillion-dollar acquihires, Jason Lemkin, a venture capitalist sometimes known as the “Godfather of Saas,” says money isn’t everything.
Lemkin said on a recent episode of the “20VC” podcast that the companies trying to win over top researchers should offer something just as valuable: The ability to work on the problems they care about with fewer constraints.
“When I talk to folks at the bleeding edge of AI, that’s just so appealing,” Lemkin said.
Google, for example, nurtured a research lab that turned it into a gravitational center for elite AI talent, well before the generative AI boom. It acquired DeepMind in 2014, allowing the people running it to stay in London and continue building their own thing. It went on to become one of the most important AI labs in the world.
“Google, back in the day, created an environment where the best researchers in the world wanted to be there,” Lemkin said.
This month, however, two high-profile AI researchers left the company.
Noam Shazeer, a co-lead of Gemini, Google’s flagship family of AI models, announced earlier this month that he was leaving Google to join OpenAI. Shazeer is a co-inventor of the Transformer architecture that underlies most of the major large language models.
Days later, Google DeepMind’s John Jumper, who won the Nobel Prize alongside CEO Demis Hassabis for his work on AlphaFold, announced he, too, was leaving for Anthropic.
Lemkin said the departures may reflect “the realities of having to try to be No. 1 in AI.” As new competitors enter the race, the freewheeling research environment that once made Google DeepMind so attractive could be under threat as it faces pressure to ship new products and integrate into Google’s larger product ecosystem. Anthropic and OpenAI, meanwhile, may be better able these days to offer the freedom to focus on the biggest AI questions of the day, Lemkin said.
“When you talk to some of the smartest engineers and developers in AI, they are really looking for a very specific kind of environment where they get to do what they want to do,” he said.
