Facing a cascade of internal crises and competitive setbacks, Meta has radically pivoted its artificial intelligence strategy, abandoning a broad campaign to acquire major AI startups in favor of a multi-billion-dollar spending spree to hire their top executives. After discussing takeovers with AI-native search engine Perplexity, ex-OpenAI CTO Mira Murati’s Thinking Machines Lab, and Ilya Sutskever’s Safe Superintelligence (SSI), Meta has instead committed over $14 billion to install Scale AI’s founder as its new superintelligence chief and is bringing on SSI’s CEO to help lead its AI assistant.
The aggressive and costly “acqui-hire” gambit reveals the depth of the pressure on CEO Mark Zuckerberg to reboot the company’s AI efforts. Stifled by the departure of key talent, the significant delay of its flagship Llama 4 “Behemoth” model, and mounting legal challenges over its training data, Meta has turned to its vast financial resources in a desperate bid to buy its way back to the forefront of the escalating AI talent war. A Meta spokesperson confirmed the company “will share more about our superintelligence effort and the great people joining this team in the coming weeks.”
This strategic shift, however, is already creating significant new problems. The massive investment in data-labeling firm Scale AI has shattered its image of neutrality, a critical asset in the industry. The move has triggered a customer exodus, creating a seismic crisis for Meta’s new partner and reshaping competitive alliances across Silicon Valley.
A House on Fire: The Internal Crisis Forcing Meta’s Hand
Meta’s audacious spending is a direct response to a firestorm of internal challenges that have left its AI division in a precarious position. The company has been hemorrhaging the talent behind its foundational AI work, having lost 11 of the 14 original authors of its Llama research paper, compounded by the high-profile exit of its Fundamental AI Research (FAIR) group head, Joëlle Pineau. In response to these struggles, the company recently restructured its AI operations into two new units, “AI Products” and “AGI Foundations”.
This turmoil appears to be a symptom of deeper issues. In January 2025, anonymous employee posts on the platform Blind described a “panic mode” inside the AI division, with one engineer stating, “Engineers are moving frantically to dissect DeepSeek and copy anything and everything we can from it,” Alongside the restructuring, Meta also tightened performance standards, requiring managers to classify more staff as “below expectations” in what an internal memo called an “opportunity to make exit decisions.”
These personnel issues are exacerbated by significant legal headwinds, including a study showing Llama 3.1 had memorized 42% of a Harry Potter book and allegations from court filings that the company used pirated “shadow libraries” for training.
If You Can’t Buy It, Poach Its CEO
Unable to acquire entire companies, Zuckerberg settled for hiring their leadership. The centerpiece of this new strategy is the deal for Scale AI. While initial reports suggested a potential $10 billion investment, the finalized agreement saw Meta take a massive 49% stake for $14.3 billion. Scale AI’s founder, Alexandr Wang, has already started at Meta, where he will lead the new superintelligence lab.
Following a failed bid to acquire the entire $32 billion Safe Superintelligence startup, Meta pivoted to hiring its co-founder and CEO, Daniel Gross, along with his partner, former GitHub CEO Nat Friedman. The pair are said to co-lead the Meta AI assistant, reporting directly to Wang. As part of the alleged deal, Meta is also acquiring a stake in their venture capital firm, NFDG, which holds investments in other prominent AI startups, according to a recent report by The Information.
The $100 Million Signing Bonus
Meta’s actions are the latest and most dramatic escalation in an industry-wide war for a handful of elite AI minds. The competition is so fierce that OpenAI CEO Sam Altman recently accused Meta of attempting to poach his developers with signing bonuses as high as $100 million. Speaking on the “Uncapped” podcast, Altman remarked that he’s “heard that Meta thinks of us as their biggest competitor” because their “current AI efforts have not worked as well as they have hoped.” He later added that while the compensation offers are massive, he believes his top people are staying for the mission.
This level of spending is becoming the norm. In 2025, Meta reportedly crossed the $2 million mark in some compensation offers and was still losing candidates, prompting Zuckerberg to make personal outreach with offers of $10 million or more. As The SignalFire State of Talent Report 2025 shows, Meta’s AI division had a higher attrition rate and lower two-year retention rate than competitors like OpenAI and Anthropic, highlighting a potential culture gap that money alone may not fix.
When Neutrality Becomes Toxic
Meta’s attempt to solve its internal crisis has paradoxically destabilized a critical part of the AI supply chain. The investment in Scale AI immediately compromised the very neutrality that made it a go-to partner for Meta’s biggest rivals. The fallout was swift and severe, as Google, Scale’s largest customer, began planning to sever a contract worth up to $200 million.
The damage, however, did not stop with Google. According to reporting from AInvest, the deal also prompted a quiet but significant client exodus, with Microsoft, OpenAI, and Elon Musk’s xAI all reportedly reconsidering or seeking to exit their partnerships over data confidentiality concerns.
The situation has created a massive opening for Scale’s competitors. Brendan Foody, CEO of rival data firm Mercor, told Forbes, “We’re already seeing a huge influx of demand from customers that are phasing out of Scale AI.” The episode underscores a new industry reality, with Turing CEO Jonathan Siddharth stating that for leading AI labs, neutrality is “no longer optional, it’s essential.”
This high-risk, high-reward strategy of buying talent and infrastructure has solved one of Meta’s problems—a lack of proven leadership—by creating a significant new one for its partners and the broader ecosystem. The coming months will reveal whether this costly gamble can truly fix the deep-seated developmental and cultural issues plaguing its AI ambitions or if it has simply traded one crisis for another.
