Five major technology companies are spending an estimated $371 billion this year on data centres for artificial intelligence, while critics warn that the investment may never generate sufficient returns to justify the costs.
The spending has raised concerns about overbuilding, reports Bloomberg Businessweek. In 2025 AI technology is expected to generate $60 billion in revenue, according to Azeez Azhar and Nathan Warren, who write the AI-focused newsletter Exponential View. McKinsey projects data centres will require $5.2 trillion in spending by 2030 to meet AI demand.
Bain calculated that Big Tech would need $2 trillion in additional annual revenue to pay for data centre expenditures by 2030 and projected a shortfall of $800 billion annually even under ideal circumstances.
Harris Kupperman, founder of Praetorian Capital Management LLC, a self-described contrarian hedge fund with about $300 million in assets under management, said the economics appear unsustainable.
“I believe it’s a bubble,” Kupperman said. “Will there ever be a payback on this stuff? I think the answer is ‘highly unlikely.'”
The MIT Media Lab found that 95 per cent of AI investments have produced no measurable returns, whilst McKinsey reported that almost 8 in 10 companies adopting generative AI see no significant bottom-line impact.
Graphics processing units, essential computer chips that account for significant data centre costs, depreciate quickly, with a shelf life of just a few years before requiring reassignment to basic AI tasks. Data centres typically take 2 to 3 years to build, but connecting them to energy sources can take up to 8 years, according to Boston Consulting Group.
Erik Gordon, a clinical assistant professor of entrepreneurial studies at the University of Michigan, said the first signs of fading optimism would likely appear in venture capital funding for AI startups.
“You might see the size of funding rounds go down,” Gordon said.
George Gilder, the writer and tech guru, said current data centre development represents overbuilding.
