Global AI spending hits record highs while economists question future profits

More money is currently flowing into artificial intelligence than past technological revolutions like railways or the internet. Cumulative global spending on data centres alone could reach $30 trillion by 2050, according to PwC projections. Individual firms are also committing massive sums, with Anthropic planning to spend $518 billion in coming years, which is more than 100 times its 2025 revenue. Backers argue that AI will be more transformational than steam engines, while tech leaders like Anthropic's Dario Amodei and OpenAI's Sam Altman speak of immense future wonders. Jasjeet Sekhon of Google DeepMind noted that recursive self-improvement is a key part of the investment thesis to deliver major productivity gains.

Financial Pressures

Economists warn that these massive outlays and high valuations rely on assumptions of vast productivity gains with little historical precedent. JP Morgan noted in August that broad-based productivity gains remain elusive. A Bain and Company study stated that productivity from existing markets will not be enough to cover costs, and companies need to find more than $4.2 trillion of new revenue in the next five years. Columbia Business School economist Stijn Van Nieuwerburgh estimated that the US AI sector would need to generate $3.55 trillion in annual revenue by 2032 to earn a 10% return on investment. Furthermore, heavy debt funding means a modest drop in demand could cause large financial losses, according to Van Nieuwerburgh. Diane Coyle of Cambridge University pointed out that past technological revolutions typically took 10 to 50 years for productivity impacts to show.

Employment Impacts

Anthropic's economics team modelled growth scenarios for 2030 ranging from a modest 2.4% to an extreme 15.4%, warning that higher growth could mean more job losses. Amodei previously forecast that AI could wipe out half of all entry-level white-collar jobs within five years. Stanford University researchers reported in August that employment for workers aged 22 to 25 in AI-exposed industries like accountancy and paralegal work was 19% lower than for jobs AI finds hard to replicate.

Context

Large-scale technology booms throughout history have frequently outpaced short-term financial returns before delivering long-term economic changes. Even when past investment bubbles burst, the underlying infrastructure remained in place to support future productivity growth.