Never has so much money flowed into a new technology as is pouring into artificial intelligence, Reuters reported on Saturday — eclipsing even the sums that railways and the internet absorbed when those revolutions were drawing in capital.
The numbers are dizzying. PwC projects that cumulative global spending on data centres alone could top $30 trillion by 2050, almost matching the value of outstanding US Treasuries, and dwarfing what was spent on the railroad or dotcom booms even after adjusting for inflation. Anthropic alone, just one of the major players in the race, plans to spend $518 billion in the coming years, according to an IPO prospectus seen by Reuters — more than 100 times its 2025 revenue.
**Where is the return?**
Lurking behind the projections, economists warn, are assumptions about vast, broad-based productivity gains and future profits for which there is little evidence so far. JP Morgan wrote in August that broad-based productivity gains in the US, which leads the AI race, "remain elusive", raising questions about the sustainability of AI valuations. It estimated that US productivity gains would need to run at 3% to 5% a year for a decade to justify Nvidia's valuation — far above the Congressional Budget Office's baseline expectation of 1.75% a year.
A Bain study found that productivity gains from existing markets would not be enough to justify current outlays, and that the hyperscalers rolling out infrastructure worldwide — Google, Amazon and Microsoft — plus others in the race needed to find more than $4.2 trillion of new revenue in the next five years. "The question is whether the applications arrive in time to pay for it," the study said.
Columbia Business School economist Stijn Van Nieuwerburgh estimates US AI investment could run as high as about $9 trillion from 2025 to 2032 — roughly 3.2% of US GDP each year — and that the US AI sector would need about $3.55 trillion in annual revenue by 2032 to earn a 10% return. Because much of the AI buildout is debt-funded, he warned, even a modest deterioration in demand or asset values could produce far larger losses.
History offers both caution and comfort. Cambridge economist Diane Coyle noted that past revolutionary technologies usually took 10 to 50 years to feed through into productivity. But she added that the infrastructure endures: trains still ran after the Panic of 1873 bankrupted railroad barons, and the internet did not shut down after the dotcom bust.
Sources:
- Reuters: https://www.reuters.com/business/re...sform-world-before-money-runs-out-2026-10-03/
The numbers are dizzying. PwC projects that cumulative global spending on data centres alone could top $30 trillion by 2050, almost matching the value of outstanding US Treasuries, and dwarfing what was spent on the railroad or dotcom booms even after adjusting for inflation. Anthropic alone, just one of the major players in the race, plans to spend $518 billion in the coming years, according to an IPO prospectus seen by Reuters — more than 100 times its 2025 revenue.
**Where is the return?**
Lurking behind the projections, economists warn, are assumptions about vast, broad-based productivity gains and future profits for which there is little evidence so far. JP Morgan wrote in August that broad-based productivity gains in the US, which leads the AI race, "remain elusive", raising questions about the sustainability of AI valuations. It estimated that US productivity gains would need to run at 3% to 5% a year for a decade to justify Nvidia's valuation — far above the Congressional Budget Office's baseline expectation of 1.75% a year.
A Bain study found that productivity gains from existing markets would not be enough to justify current outlays, and that the hyperscalers rolling out infrastructure worldwide — Google, Amazon and Microsoft — plus others in the race needed to find more than $4.2 trillion of new revenue in the next five years. "The question is whether the applications arrive in time to pay for it," the study said.
Columbia Business School economist Stijn Van Nieuwerburgh estimates US AI investment could run as high as about $9 trillion from 2025 to 2032 — roughly 3.2% of US GDP each year — and that the US AI sector would need about $3.55 trillion in annual revenue by 2032 to earn a 10% return. Because much of the AI buildout is debt-funded, he warned, even a modest deterioration in demand or asset values could produce far larger losses.
History offers both caution and comfort. Cambridge economist Diane Coyle noted that past revolutionary technologies usually took 10 to 50 years to feed through into productivity. But she added that the infrastructure endures: trains still ran after the Panic of 1873 bankrupted railroad barons, and the internet did not shut down after the dotcom bust.
Sources:
- Reuters: https://www.reuters.com/business/re...sform-world-before-money-runs-out-2026-10-03/
