Analysis:AI's race to transform the world before the money runs out

In recent years, there has been an unprecedented influx of capital into artificial intelligence (AI), surpassing previous technological revolutions such as railways and the internet. According to PwC's projections, global spending on data centers could reach $30 trillion by 2050, equal to the value of outstanding US Treasuries. This investment dwarfs previous booms, even after accounting for inflation.

Companies like Anthropic, a major player in the AI race, plan to invest $518 billion in the coming years, which is over 100 times its 2025 revenue. Backers of AI technology argue that it will be more transformative than the advent of steam engines and industrialization. However, there are concerns about the assumptions surrounding vast productivity gains and future profits, with economists questioning the ability of AI to deliver on these projections.

JP Morgan has noted that broad-based productivity gains in the US, a leader in AI, remain elusive, raising questions about the sustainability of AI valuations. Bain & Company suggests that new markets, such as AI-guided robots and new materials for batteries and semiconductors, must emerge to justify current outlays. For the AI race to be profitable, companies need to find more than $4.2 trillion in new revenue over the next five years.

While the potential of AI to transform various aspects of life is not in doubt, the key question is whether the required applications will arrive in time to pay for the investment. Past technology-driven booms have often ended when infrastructure buildouts ceased to deliver sufficient returns. Nvidia, a leading company in the AI revolution, would need US productivity gains of 3% to 5% annually over the next decade to justify its valuation, which is significantly higher than the US Congressional Budget Office's baseline expectation of 1.75% annual productivity growth.