From the inside Last year, there were at least three major AI “acquisitions” in Silicon Valley. Meta invested over $14 billion in Scale AI and hired its CEO, Alexander Wang; Google it spent a cool $2.4 billion licensing Windsurf technology and incorporating its co-founders and research teams into DeepMind; and Nvidia he bet $20 billion on Groq inference technology and hired its CEO and other employees.
Meanwhile, frontier artificial intelligence labs play a high-stakes, seemingly never-ending game of musical chairs for talent. The latest shake-up began three weeks ago when OpenAI announced it would rehire several researchers who left less than two years earlier to join Mira Murati’s startup, Thinking Machines. At the same time, Anthropic, founded by former OpenAI employees, is poaching the talents of the creator of ChatGPT. OpenAI in turn just hired former anthropic security researcher as “head of preparedness.”
Silicon Valley’s employee exodus marks a “great unbundling” of the tech startup, as Dave Munachiumiello, an investor in GV, put it. In earlier eras, technology developers and their early employees often stayed on board until the lights went out or a major liquidity event occurred. But in today’s market, where generative AI startups are growing rapidly, have plenty of capital and are valued especially for the strength of their research talent, “you invest in a startup knowing it could get broken,” Munichumiello told me.
Early founders and researchers of the most popular AI startups move to different companies for many reasons. Of course, money is a substantial incentive for many people. Last year, Meta reportedly offered top AI researchers remuneration packages in the tens or hundreds of millions of dollars, offering them not only access to cutting-edge computing resources, but also… generational wealth.
But it’s not just about getting opulent. The broader cultural shifts that have rocked the tech industry in recent years have made some workers wary of sticking with one company or institution for too long, says Sayash Kapoor, a computer science researcher at Princeton University and a senior research fellow at Mozilla. Employers used to safely assume that employees would remain with the company until at least four years, when stock options were typically scheduled to vest. During the high-flying era of the early 2000s and first decade of the 21st century, many early co-founders and employees also truly believed in their companies’ specific missions and wanted to be there to lend a hand achieve them.
Nowadays, says Kapoor, “people understand the limitations of the institutions they work in, and founders are more pragmatic.” For example, Windsurf’s founders may have calculated that their influence might be greater at a place like Google, which has a lot of resources, says Kapoor. He adds that a similar change is taking place in academia. Over the past five years, Kapoor says, more and more graduate students have dropped out of computer science PhD programs to pursue jobs in industry. He believes there are higher opportunity costs to staying in one place at a time when AI innovation is rapidly accelerating.
Investors, concerned that AI talent wars could incur additional losses, are taking steps to protect themselves. Max Gazor, founder of Striker Venture Partners, says his team is vetting founding teams “more than ever for chemistry and cohesion.” Gazor also claims that it is increasingly common for contracts to include “protective provisions requiring management approval to license material intellectual property rights or similar scenarios.”
Gazor notes that some of the largest recent acquisition deals have involved startups founded long before the current generative AI boom. For example, Scale AI was founded in 2016, when the type of deal Wang had negotiated with Meta was incomprehensible to many. Now, however, these potential outcomes can be taken into account at an early stage of implementation and “constructively managed,” Gazor explains.