Why Is Meta Fueling the Frenzy for AI Talent and Acqui-Hires

In a sign of just how heated the artificial intelligence talent war has become, Meta CEO Mark Zuckerberg has reportedly taken to personally emailing elite AI researchers with job offers worth at least $10 million a year. The move is part of an aggressive push to build a new “superintelligence” group, a strategy that also saw Meta agree to invest nearly $15 billion for a 49% stake in data-labeling firm Scale AI, while simultaneously hiring away its CEO, Alexandr Wang, to lead the new team.

The extraordinary compensation packages and strategic “acqui-hires” highlight a new reality in the AI boom: The race for dominance is not just about building the biggest data centers or owning the most powerful chips. It’s an increasingly desperate and astronomically expensive global contest to recruit and retain a tiny cohort of human minds who know how to build the frontier models that are reshaping the tech landscape.

Why is the talent pool so small and valuable?

Building cutting-edge foundation models is less a straightforward engineering task and more, as one AI expert put it, like “alchemy.” The number of researchers with hands-on experience in successfully training these massive, complex systems is estimated to be in the low thousands, or perhaps even just a few hundred, globally. OpenAI CEO Sam Altman alluded to this phenomenon in late 2023, tweeting “sure 10x engineers are cool but damn those 10,000x engineer/researchers,” suggesting the most effective AI talent is exponentially more productive than the average.

This scarcity has created a seller’s market for top researchers. According to a report from SignalFire, the attrition rate for AI talent at major tech firms is high, with Meta and Google losing 4.3% and 5.4% of their AI staff to rivals in 2024, respectively. This brain drain forces companies to pay a steep premium to both attract new talent and prevent their existing stars from leaving.

How intense is the competition?

The battle for talent has escalated to levels typically reserved for professional athletes. Besides Meta’s eight-figure offers, Google’s DeepMind has reportedly offered top researchers compensation packages of $20 million per year. OpenAI, not wanting to lose its top minds to rival startups, has offered retention bonuses of $2 million on top of equity packages worth $20 million or more to key researchers considering leaving.

The courting process has also become intensely personal. Here's Reuters reporting on the recruitment effort:

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.

This scramble is driven by the belief that a single, brilliant individual contributor can make or break a company’s next-generation model, giving them an edge over competitors.

Is it just about salaries?

No. When direct hiring fails, tech giants are resorting to a new type of deal that many in the industry see as an acquisition in all but name. Known as “acqui-hires,” these deals often involve a large company investing in or licensing technology from a smaller AI startup while simultaneously hiring its founders and key employees. This strategy allows the larger firm to absorb the startup’s most valuable asset — its human capital — often while sidestepping the lengthy regulatory scrutiny that a formal merger would attract.

This trend has accelerated in 2025. Besides Meta’s deal with Scale AI, Microsoft effectively absorbed most of the team from Inflection AI for $650 million, and Google struck a deal with Character.AI to license its technology and hire its co-founders in a transaction valued at over $2 billion. These moves show that the cost of acquiring top-tier talent now runs into the billions, reshaping the M&A landscape.

How does this fit into the overall cost of AI?

The eye-watering salaries and acqui-hire price tags are just one component of the colossal spending required to compete in AI. The human capital costs are dwarfed by the capital expenditure on physical infrastructure. According to research from McKinsey, the demand for AI-ready data centers is projected to require $5.2 trillion in investment by 2030.

Individual companies are making staggering commitments. Meta plans to increase its capex to between $64 billion and $72 billion in 2025 alone, largely to support its AI ambitions. Amazon has announced plans to spend over $100 billion on capex in 2025, with the vast majority going to AI capabilities. Much of this spending flows to Nvidia, whose high-margin GPUs are essential for training and running AI models, a dynamic that has been called the “Nvidia tax.” This reliance on a single hardware provider is what makes projects like OpenAI’s $500 billion “Stargate” data center initiative with Oracle and SoftBank—which will also use Nvidia chips—so critical for securing future compute capacity.

What is the end game for this spending spree?

The ultimate prize these companies are chasing is the development of artificial general intelligence (AGI), or “superintelligence”—AI systems that can outperform humans on a wide range of tasks. The potential economic payoff is enormous. A recent legal filing revealed that Meta forecasts its generative AI products could generate between $460 billion and $1.4 trillion in revenue by 2035.

This projected return is what justifies the massive upfront investment in both people and infrastructure. The pressure to deliver is immense, particularly for companies like Meta, whose Llama 4 model launch in April 2025 was considered underwhelming compared to rivals. In a world where competitors like OpenAI and even Chinese startups like DeepSeek are demonstrating startling advances, falling behind is not an option. The astronomical cost of talent is simply the price of admission to the highest-stakes race in modern technology.

Reference Shelf:

• Meta Offered One AI Researcher at Least $10,000,000 to Join Up (The Register)
• OpenAI, Google and xAI battle for superstar AI talent, shelling out millions (Reuters)
• AI Companies Almost Get Bought (Bloomberg)
• Meta raises AI data center capex forecast to up to $72bn, blames Trump tariffs for increased cost (Data Centre Dynamics)