Never before has the global economy funneled such colossal capital into a new technology. Investment in Artificial Intelligence (AI) is already eclipsing the amounts once spent on the development of railroads or the Internet. According to PricewaterhouseCoopers (PwC), cumulative global spending on data centers alone could reach $30 trillion by 2050—a figure approaching the value of all outstanding US Treasury bonds.
The Gap Between Investment and Productivity
Despite the fervor, economists are raising serious flags. JPMorgan reported in August that broad productivity gains in the US remain "elusive," questioning the sustainability of current AI company valuations. To justify Nvidia’s valuation, for instance, US productivity would need to grow by 3% to 5% annually for a decade—nearly double the Congressional Budget Office’s projection of 1.75%.
Bain & Company notes that existing markets are insufficient to bridge the funding gap. Tech giants known as "hyperscalers"—including Google, Amazon, and Microsoft—will need to generate more than $4.2 trillion in new revenue over the next five years just to fund their infrastructure expansion.
Corporate Ambition and Labor Market Shifts
Industry leaders remain undeterred. Anthropic plans to spend $518 billion in the coming years, a sum 100 times its projected 2025 revenue. Its CEO, Dario Amodei, has predicted that AI could eliminate half of entry-level office jobs within five years. Early signs are already emerging; Stanford research shows that hiring for 22-25-year-olds in AI-vulnerable sectors, such as accounting and paralegal work, is 19% lower compared to roles AI struggles to replicate.
"Historical precedents suggest that technology-driven booms often end when infrastructure growth ceases to yield sufficient returns." — JPMorgan
While the risks are high, history suggests that the infrastructure left behind—much like the railroads after the Panic of 1873 or the fiber optics after the dot-com bubble—will continue to provide long-term utility, even if the initial investors face ruin. The critical question remains whether AI applications will emerge fast enough to service the staggering debt fueling this revolution.
