"Blood transfusion": how and why Meta "sucks" talent from the market

Why are top soccer players bought up by the richest clubs for millions of dollars, while outstanding doctors and scientists settle for modest salaries of tens of thousands of dollars a year? The staggering fees of soccer players are explained by the high competition in world soccer, but shouldn't we compete in science in the same way, attracting the best talents from all over the world? These were the questions asked back in 2014 by then British parliamentary candidate Daniel Zeichner (and, let's be honest, he was far from alone). And so a new frenzy seems to have begun in Silicon Valley. Tech companies are poaching each other's best AI experts for tens, and according to some reports, hundreds of millions of dollars in a frantic race to create artificial "superintelligence." The funny thing is that no one seems to know exactly what it is. But the money being offered is already quite real.

As I wrote in mid-June, Meta co-founder Mark Zuckerberg is concerned about his company's lagging behind competitors in AI and has set his sights on creating a "superintelligence" lab. To do this, he began actively recruiting specialists, for whom there is already a huge demand. In particular, for $14.3 billion he acquired 49% of the startup Scale AI, it seems, only to get its co-innovator Alexander Wang on his team, who wrote on the X network that the main condition of the deal was his transfer to Meta. Zuckerberg also managed to bring former GitHub CEO director Nat Friedman to the lab.

But that was just the beginning. One of the recruits who spoke to Zuckerberg - and the head of Meta is personally seeking potential candidates for his dream team - describes the company's goal as nothing less than "blood transfusion" at Meta, one of the country's leading AI labs, the WSJ reports. People are being lured away by entire departments: for example, in June it became known that three OpenAI employees who were involved in organizing the Zurich office were leaving for Zuckerberg: Lucas Bayer, Alexander Kolesnikov and Xiaohua Zhai;

The "bigger fish" are being targeted. On Friday, July 4, Ilya Sutskever, founder of the AI lab Safe Superintelligence (incidentally, former chief scientist of OpenAI), announced that he is taking over the functions of CEO of his company. Why? His CEO Daniel Gross was also snatched by Meta, writes Reuters. Zuckerberg was not against getting Sutzkever himself and offered to buy his company in its entirety, but was refused. He had to settle for the CEO.

The whole of Silicon Valley is buzzing, discussing a certain Meta "list" (The List) for hiring, allegedly including both top AI specialists and promising newcomers. The main topic of conversation is who is on The List, who isn't, and why. "I hear a lot of discussion in the Valley. There's a sense of jealousy, envy and helplessness, with everyone saying, 'I thought I was doing well. What am I doing wrong?' I'd love to get these offers,'" quoted by Business Insider as Didi Das, a tech investor at Menlo Ventures.

They have a lot to worry about.

According to Wired magazine, top AI specialists at Meta  are being offered compensation packages of up to $100 million per year, and up to $300 million in the first four years. 

By comparison, in 2017, Barcelona player Neymar's move to Paris Saint-Germain cost a record 222 million euros cost. He himself, under a five-year contract  was to receive 45 million euros (about $54 million) a year - totaling 225 million euros (about $270 million).

Again by comparison, Microsoft CEO Satya Nadella, according to the company's financial statements, received $79.1 million in total compensation in 2024, most of which was stock, Wired notes. At the end of 2024, the company's market capitalization was about $3 trillion.

There even seems to be a new profession in the Valley. "It's funny - now there are people who help researchers negotiate their compensation packages and charge for it, like athletes' agents," wrote Sarah Goh, a prominent venture capitalist and founder of the Conviction fund, on Web X.

"These claims are not true - the size and structure of these compensation packages have been significantly misrepresented," Meta spokesman Andy Stone told Wired. - Some people decided to greatly exaggerate what was going on for their own purposes".

Be that as it may, the situation is serious enough for the CEO of leading AI lab OpenAI, Sam Altman, to issue a special statement in corporate Slack. 

"We've gone from a bunch of nerds in the corner to the most interesting people in the tech industry (at least). AI-Twitter is toxic (apparently referring to Ilon Musk's AI startup xAI, which bought Twitter and renamed it X - ed. note); Meta is behaving in a way that makes it look pretty obnoxious; I think it's going to get even crazier from here on out," he wrote;

He expressed confidence that the "missionaries" (apparently, the faithful ideological employees of OpenAI) will always defeat the "mercenaries" (apparently, meaning those who went to Meta for more money). However, not relying only on ideology, he hinted that OpenAI is preparing a new compensation program for its employees and that his company's shares have "much, much more room for growth than Meta's shares".

However, what exactly does Zuckerberg intend to occupy his new team with, acquired at such an expensive, literally, price?  

On Monday, June 30, CNBC got a look at Meta's internal memorandum, in which Zuckerberg lists his main objectives.

"We are going to name our overall organization Meta Superintelligence Labs (MSL). It includes all of our FAIR (Meta's former AI development) funds, products and teams, as well as a new lab focused on developing the next generation of our models," the memo says.

The main goal of her work is proclaimed as "personalized super intelligence for everyone", but what this is, Zuckerberg does not explain.

It seems that the term "super-intelligence" he did not hesitate to "cut" from the manifesto of his competitor Sam Altman published a year earlier. The latter, however, didn't give a precise definition either, simply promising some new set of technologies that would lead to unprecedented prosperity for humanity. From the point of view of AI visionaries, the so-called general AI (AGI) is a system capable of performing all tasks as well as a human, and such systems are about to appear. And "superintelligence" is a certain next stage of its development, at which the AI will be able to do what humans cannot.

"Superintelligent tools could greatly accelerate scientific discovery and innovation, far beyond what we are capable of doing on our own, and in turn greatly increase abundance and prosperity," stated Altman in his January 2025 blog post.

However, there is considerable reasonable doubt about the imminent emergence of AGI, wrote The New York Times in its piece "Why We're Unlikely to Get AGI in the Foreseeable Future." Scientists don't even agree on how to measure human intelligence, let alone machine intelligence, which is fundamentally different. Not to mention the subsequent correct comparison of the results of these measurements.

Judging by Zuckerberg's memo, it appears that for now it's about accelerating the development of the already existing Llama family of AI models, which will form the basis of the Meta AI chatbot and other AI tools the company is developing, particularly "in the category of AI-enabled glasses and wearable devices, which is growing very rapidly."

Smart glasses? Honestly, not even AGI, let alone "super smart".

"These (AI) are great tools," the NYT quoted Subbarao Kambhampati, a professor and AI researcher at the University of Arizona (US). - But I don't believe we're anywhere close to AGI or superintelligence. With these tools, you always need a human in the toolkit."

It takes more than money to attract top talent, argues the NYT.  Top researchers are already making millions of dollars a year and want to work on the most ambitious projects. That's why Meta uses the term "superintelligence."

"Where these terms come from, one can only speculate," believes Kambhampati. - It is probably commerce, branding, advertising. But they have nothing to do with technology.

This article was AI-translated and verified by a human editor