San Francisco – During the summer, Matt Deitke received a call from Mark Zuckerberg, CEO of the goal.
Zuckerberg wanted Deitke, a 24 -year -old artificial intelligence researcher who had recently helped to found a startup, joining the research effort of the goal dedicated to “superintelligence,” a technology that could hypothetically overcome the human brain. The company has promised it about $ 125 million in stock and money over four years if it accepted.
The offer was not enough to convince Deitke, who wanted to continue with his startup, two people were aware of the negotiations. He refused Zuckerberg.
Take your business to the next level with the help of the country’s leading entrepreneurs!
Then Zuckerberg met in person with Deitke. Then the goal came back with a revised offer of about $ 250 million in four years, with potential up to $ 100 million paid in the first year, the sources said. The increase in compensation was so surprising that Deitke consulted his colleagues about what to do. After many discussions, some encouraged him to accept the deal – what he did.
The war for talents in the Silicon Valley has become so frantic-and so absurd-that it increasingly resembles the stratospheric market of the NBA stars.
Young AI researchers are being recruited as if they were Steph Curry or LeBron James, with nine -digit -structured compensation packages to be paid over several years. To navigate this scenario, many of these young people in their 20s have resorted to agents and unofficial teams to trace strategies. And they are negotiating hard with companies to get the best value, just as basketball players look for the best contracts.
Continues after advertising
The difference is that, unlike NBA teams, AI companies with great features such as Goal, OpenAi and Google have no salary ceiling. (The latest contract of four years of curry with the Golden State Warriors was $ 35 million lower than Deitke’s agreement with the goal.) This made the battles for even more intense AI talents.
In recent weeks, the recruitment of free agents in IA has become a show on social networks, similar to the period before the deadline of exchanges in the sport. As Microsoft, Google, and OpenAi have played employees with each other, hiring ads have been posted online with graphics that resemble major sports exchanges made by the TBPN streaming channel, which features a ESPN style program on technology and business.
“Urgent: Microsoft has recruited more than 20 Deepmind employees in the last six months,” said a recent TBPN post about Microsoft’s signings in Google’s Deepmind Laboratory.
Continues after advertising
On Wednesday, Zuckerberg said the goal plans to continue investing heavily in AI talents, “because we are convinced that superintelligence will improve all aspects of what we do.” The superintelligent AI would not only improve the company’s business, he said, but would also become a personal tool that “has the potential to start an exciting new era of individual empowerment.”
A goal spokesman preferred not to comment. Deitke did not respond to the request for comment.
The job market for AI researchers for a long time has parallels with professional sport. In 2012, after three scholars of the University of Toronto published an article describing an AI seminal system capable of recognizing objects such as flowers and cars, they auctioned themselves to the largest corporate bid – Google – for $ 44 million.
Continues after advertising
This began a talent race in the technology industry. In 2014, Peter Lee, Microsoft’s head of research, compared the market to the rising football players, many of whom earned about $ 1 million a year.
“Last year, the cost of a world-class Deep Learning expert was approximately the same as NFL’s quarterback prospect,” Lee told Bloomberg Businessweek at the time, referring to a type of AI expert. “The cost of this talent is really remarkable.”
The negotiation power of AI researchers has only increased since OpenAi launched Chatbot Chatgpt in 2022, triggering a race to lead technology. They are helped by scarcity: Only a small group of people have technical knowledge and experience to work on advanced artificial intelligence systems.
Continues after advertising
This is because AI is constructed differently from traditional software. These systems learn by analyzing huge amounts of digital data. Few researchers have experience with the most advanced systems, which require large volumes of computational power available for few companies.
The result has been a new war for talent, with compensations that jumped to hundreds of millions of dollars a year, leaving millions a year.
In April, Zuckerberg – whose company faced difficulties to advance his AI research – came into action by sending personal messages to potential recruits, offering increasing sums.
Continues after advertising
His approach was compared to that of sports franchise owners, two staff said. Even though offers seemed absurd, if new contractors could help increase revenue at half a percentage point – especially for a company that approaches a $ 2 trillion market capitalization – it would be worth it, the sources said.
“If I’m Zuck and spent $ 80 billion in a year only on capital expenses, is it worth investing another $ 5 billion or more to set up a truly world -class team to bring the company to the next level?” said Hays. “The answer is obviously yes.”
The initial offers of the goal for engineers varied, but revolved around tens of millions of dollars, three people said with the process.
The company also offered something that may be more attractive than money: computational power. Some potential hired were informed that they would receive 30,000 graphic processing units, or GPUS, for their IA research, one of the sources said. GPUS, powerful chips ideal for running the AI -feeding calculations, are highly coveted.
Talent wars began to impact. OpenAi has changed its compensation structure to adapt to the market, said company employees, and is asking employees approached by competitors to consult executives before accepting offers immediately.
“Are we counterattacking? Yes,” said Mark Chen, OpenAi’s research director at a company meeting this month, according to a recording revised by The New York Times. But he added that OpenAi did not equalize the goal offers because “I personally think to work here, you have to believe the potential of OpenAi.”
OpenAi declined to comment. (Times sued OpenAi and Microsoft, alleging copyright violation of news content related to AI systems. The two companies denied the accusations.)
Not all goal attacks were successful. The company was rejected by some researchers, two people said, in part because Zuckerberg’s view for artificial intelligence was less clear compared to other companies.
Still, euphoria allowed researchers little known as Deitke to trace their own destinations.
Deitke, who recently abandoned a doctoral program in computer science at the University of Washington, worked in parallel at an AI laboratory in Seattle called Allen Institute for Artificial Intelligence. There, he led the development of a project called Molmo, an AI chatbot that combines images, sounds and text – the type of system that the goal tries to build.
In November, Deitke and several colleagues from the Allen Institute founded Vercept, a startup that tries to build AI agents capable of using other software on the Internet to perform tasks autonomously. With about 10 employees, Vercept raised $ 16.5 million of investors like former Google Eric Schmidt.
Then came Deitke’s negotiation with Zuckerberg. After Deitke accepted the offer of about $ 250 million from goal to four years, Vercept CEO posted on social networks: “We look forward to joining Matt on its private island next year.”
c.2025 The New York Times Company
