Why Meta Still Can't Recruit This 50-Person AI Startup Despite Offering Over $10 Billion

Mark Zuckerberg is targeting Murati's employees

Phoenix Technology News, Beijing time July 30, according to Line magazine, Meta CEO Mark Zuckerberg is actively recruiting top AI talent for his new superintelligent laboratory. After successfully luring away many OpenAI top researchers, he seems to have set his sights on the next target.

This target is Thinking Machines Lab (TML), a 50-person AI company founded by Mira Murati, former chief technology officer of OpenAI. TML has already had over 12 people contacted or offered high-end compensation packages by Meta.

Insiders reveal that one such package is worth over $10 billion for a period of several years. Other sources confirm that the remaining packages are between $2 billion and $5 billion, paid out over four years. Meta assures that these employees will receive compensation of $50 million to $1 billion in their first year alone.

However, as of now, TML has not accepted any invitations from Meta.

Meta Communications Director Andy Stone issued a statement denying the report. "We only extended invitations to a small number of TML employees," he said. "Although there was one very attractive offer, the details of the report are inaccurate. This raises questions about who is manipulating this narrative and what their motives are." He added.

Zuckerberg's Recruitment Strategy

According to Line magazine, Zuckerberg's initial approach was low-key. In some cases, he would send private messages to potential recruits on WhatsApp, inviting them for a chat. The interview process then progressed quickly, with the recruiter having a long conversation with the candidate, followed by meetings with Meta's chief technology officer Andrew "Boz" Bosworth and other high-ranking executives.

Here is the information sent to a potential recruit by Zuckerberg (in a similar tone today):

"Over the years, we've been following your work in driving technological advancements and making AI beneficial to humanity. To create valuable AI products and services for people, we're investing significantly in research, product development, and infrastructure. We believe that every user will have a world-class AI assistant to help them complete various tasks; every creator will have an AI that can interact with their community; every business will have an AI that can provide shopping and support services; and every developer will be able to use the most advanced open-source models to build AI. We hope to bring in the best talent at Meta and are happy to share more about our work."

During the recruitment process, Bosworth candidly expressed his thoughts on how Meta competes with OpenAI. According to insiders, despite Meta falling behind smaller competitors in building leading-edge models, it plans to use open-source strategies to undermine OpenAI. Meta's approach is that it can release open-source models directly competing with OpenAI, commercializing AI technology.

A message from Meta claims that "since the beginning of this year, the company has been under pressure. I think this pressure reached its peak when Llama 4 was hastily released." Meta's latest generation series models had previously struggled with performance enhancements and were delayed. After their official release, Meta seems to have manipulated evaluation benchmarks to make its models appear better than they actually are, sparking controversy.

High Pay Doesn't Always Guarantee Talent

So why can't Meta's high-profile recruitment strategies win over TML's top talent?

Since Zuckerberg appointed Scale AI co-founder Alexandr Wang and former GitHub CEO Nat Friedman to jointly lead Meta's superintelligent laboratory, insiders have been telling Line magazine that Wang's leadership style is controversial. Moreover, his experience is relatively lacking, sparking concerns. Although this did not prevent Zuckerberg from recruiting nearly 24 people for his laboratory in a short period of time, it was not everyone who wanted to work with Wang. Ottermann once said, "Meta hasn't lured away our top talent," and they had to choose second-tier employees.

Other insiders claim that Meta's product roadmap is not very inspiring. For some people, money can be made anywhere for Reels or Facebook, but building a substandard AI doesn't have much significance. In OpenAI or Anthropic, they can earn just as high compensation while participating in more noble missions, such as "building a universal artificial intelligence beneficial to humanity." Some interviewees who accepted Meta's offer say that the process now resembles a test of one's value in the AI market.

Moreover, TML does not lack funds. This startup company has just completed its largest seed round funding in history, showing that the researchers they select do not need to choose between idealism and financial incentives. The company, which was established only a year ago, is already valued at $120 billion, with no products released yet. As the saying goes, "Why can't we have both?" (Author/Arax)