The contest in Silicon Valley to dominate artificial intelligence is playing out on a new court: superstar researchers.

While the scramble to attract top talent and keep them happy has always been a hallmark of the tech industry, since ChatGPT launched in late 2022, recruiting has escalated to professional athlete levels, a dozen people who have been involved in recruiting AI researchers told Reuters.

“The AI labs approach hiring like a game of chess,” said Ariel Herbert-Voss, CEO of cybersecurity startup RunSybil and a former OpenAI researcher who entered the talent fight after launching his own company. “They want to move as fast as possible, so they are willing to pay a lot for candidates with specialized and complementary expertise, much like the game pieces. They are like, do I have enough rooks? Enough knights?”

Companies including OpenAI and Google, eager to get or stay ahead in the race to create the best AI models, court these so-called “ICs” – the individual contributors whose work can make or break companies.

Noam Brown, one of the researchers behind OpenAI’s recent AI breakthroughs in complex math and science reasoning, said when he explored job opportunities in 2023, he found himself being courted by tech’s elite: lunch with Google founder Sergey Brin, poker at Sam Altman’s, and a private jet visit from an eager investor. Elon Musk will also make calls to close candidates for xAI, his AI company, said two people who have spoken to him. xAI did not respond to a request for comment.

Ultimately, Brown said, he chose OpenAI because OpenAI was willing to put resources – both people and compute – behind the work he was excited about.

“It was actually financially not the best option that I had,” he said, explaining that compensation is not the most important thing for many researchers. That hasn’t stopped companies from throwing millions of dollars in bonuses and pay packages at star researchers, according to seven sources familiar with the matter.

A few top OpenAI researchers who have indicated interest in joining former chief scientist Ilya Sutskever’s new company, SSI, were offered retention bonuses of $2 million, in addition to equity increases of $20 million or more, if they stayed, two sources told Reuters. Some have only been required to stay for a year in order to get the entire bonus. SSI and OpenAI declined to comment.

Other OpenAI researchers who have fielded offers from Eleven Labs have received bonuses of at least $1 million to stay at OpenAI, two sources told Reuters. Top OpenAI researchers regularly receive compensation packages of over $10 million a year, sources said.

Google DeepMind has offered top researchers $20 million per year compensation packages, awarded off-cycle equity grants specifically to AI researchers, and has also reduced vesting on some stock packages to 3 years, instead of the normal 4 years, sources said. Google declined to comment.

In contrast, top engineers at big tech companies receive an average yearly compensation of $281,000 in salary and $261,000 in equity, according to Comprehensive.io, a company that tracks tech industry compensation.

10,000x TALENT

While talent has always been important in Silicon Valley, the difference with the AI boom is how few people are in this elite group – depending on who you ask, the number could range from a few dozen to around a thousand, eight sources told Reuters.

That is based on the belief that this very small number of ‘ICs’ have made outsized contributions to the development of large language models, the technology today’s AI boom is based on, and therefore could make or break the success of an AI model.

“sure 10x engineers are cool but damn those 10,000x engineer/researchers…,” OpenAI CEO Sam Altman tweeted in late 2023, alluding to a long held maxim that the best software engineers were 10 times as good as the average (10x), but now in the AI industry, the best researchers are 10,000 times (10,000x) as effective as the average.

The September departure of OpenAI’s chief technology officer, Mira Murati, who then founded a rival AI startup, has intensified the AI talent war.

Murati, who was known at OpenAI for her management skills and execution prowess, recruited 20 OpenAI employees before announcing her company in February. She has now lured even more researchers from OpenAI and other labs, and the team is now around 60 people, two sources told Reuters. Though the company has no product in the market, Murati is in the middle of closing record-breaking seed round that is, based on the team’s strength. A representative for Murati declined to comment.

The scarcity of talent has forced companies to approach hiring creatively. Zeki Data, a data firm focused on identifying top AI talent, said it is employing sports industry data analysis techniques like the one popularized by the movie "Moneyball" to identify promising but undiscovered talent. For instance, Zeki Data discovered Anthropic has been hiring researchers with theoretical physics backgrounds, and other AI companies have hired individuals with quantum computing backgrounds.

Anthropic did not reply to a request for comment.

“On my team, I have extraordinarily talented mathematicians who wouldn't have come to this field if it weren't for the fast progress that we're seeing now,” said Sébastien Bubeck, who left, opens new tab his role as vice president of GenAI research at Microsoft last year to join OpenAI. “We're seeing an influx of talent from all fields going into AI now. And some of these people are very, very clever, and they make a difference.”