Tech giants spent $380B on AI in 2025, poaching researchers with $250M offers

Tech giants are spending billions to poach top AI talent

Google, Amazon, Microsoft, and Meta collectively spent $380 billion on AI in 2025. That figure is projected to surge to $650 billion this year, funding both physical infrastructure like data centers and a fierce competition for elite researchers.

The battle for talent has reached staggering levels. Meta reportedly offered a single AI researcher a compensation package worth $250 million over four years. Companies are also spending billions on "reverse-acquihires," poaching star staff from startups without buying the companies.

Academia is facing a massive brain drain

Since ChatGPT's launch in 2022, a significant "AI brain drain" from universities to industry has accelerated. A 2025 study found that young, highly cited scholars are especially likely to leave.

Researchers about five years into their careers, whose work ranks among the most cited, were 100 times more likely to move to industry the following year than ten-year veterans with average citation counts. This model was based on data from nearly seven million papers.

This exodus threatens academia's distinct roles: curiosity-driven innovation, independent critique, and ethical scrutiny. The tech industry's fixation on top individual talent also risks undermining science as a collaborative, team-based endeavor.

The myth of the lone genius engineer

The astronomical salaries for AI talent buy into the old software industry legend of the "10x engineer"—a single person supposedly capable of ten times the impact of their peers. Tech firms find this narrative increasingly attractive as they bet AI will replace many entry and mid-level jobs.

Google's marketing for its Gemini 3 Pro model, boasting "PhD-level reasoning," directly appeals to executives seeking to replace people with AI. However, this lone-genius narrative is fundamentally at odds with how modern science works.

Research consistently shows science is a team sport. A large-scale study of publishing from 1900 to 2011 found papers from larger collaborations have greater impact than those from smaller teams. Analyses of highly cited scientists and Nobel laureates show the same pattern: their highest-impact work tends to involve many authors.

Why collaboration matters more than ever

As scientific problems grow in scope and complexity, teamwork has become essential. The trend is clear in the data:

• Papers from larger teams consistently have greater impact.
• The most highly cited scientists produce their best work with many co-authors.
• The average team size for Nobel laureates' publications has steadily increased over time.

Big Tech's strategy of skimming the very top individual talent risks eroding this collaborative foundation. It prioritizes a narrow, profit-driven vision of innovation over the broader, curiosity-driven exploration that academia uniquely provides.

The risks of an AI-driven research model

An overreliance on AI-driven modelling presents distinct dangers for science. When profit motives dominate, research can become myopic, focusing on short-term commercial applications rather than foundational questions.

This shift threatens the independent ethical scrutiny crucial for a technology as powerful as AI. Without strong academic counterweights, the development and deployment of AI systems may lack necessary oversight and consideration of societal impacts.

The current trajectory, where industry vacuums up top talent and frames progress around individual genius or AI replacement, challenges the very idea of science as a public, collaborative good. The future of innovation depends on preserving a vibrant, independent academic ecosystem.