Investors are grappling with doubts about whether the substantial capital flowing into artificial intelligence (AI) infrastructure will yield the returns promised by proponents of the technology. Recent market developments underscore these concerns, as CoreWeave, a leading independent AI computing provider in the United States, saw its market value drop by about 25% amid worries over its $50 billion debt exposure. The turmoil has also affected Japan’s SoftBank and raised apprehensions among major technology players including Meta and prominent semiconductor manufacturers.

The AI industry’s expansion has largely been driven by heavy investment in data centers and computing power, based on the assumption that demand for AI computing resources is effectively unlimited. However, some economists argue that this demand may be overstated. Torsten Slok, chief economist at Apollo, has suggested that customer willingness to pay for compute resources might diminish sharply if providers begin charging rates that reflect the true costs of AI infrastructure. He warns that users could reduce their usage or turn to more affordable, open-source alternatives, potentially leaving major hyperscalers to shoulder rising capital expenditures amid declining profit margins.

At the heart of this uncertainty is skepticism over whether AI will generate the anticipated productivity gains. Historically, advances in computing have shown uneven effects on economic productivity. Despite the proliferation of computers and digital technology over recent decades, overall growth in productivity across advanced economies has fallen short of expectations. This phenomenon echoes the observation made by economist Robert Solow decades ago, who noted a disconnect between the widespread adoption of computers and measurable improvements in productivity statistics.

One analysis highlights that information technology more often redistributes economic value rather than expanding it, favoring network-driven social interactions, cost-cutting disintermediation, and shifts toward asset-light business models. While these changes can accelerate adoption and connectivity, they also bring challenges such as increased competition, wealth concentration, and other negative externalities that may offset productivity gains. Complexity scientist Geoffrey West has examined these patterns, noting that as cities grow, economic benefits and associated costs both increase disproportionately—a dynamic that IT technologies exacerbate. This effect can lead to a fragile balance where social and economic acceleration does not necessarily translate into improved collective well-being.

Moreover, the pervasive use of digital connectivity may have unintended societal impacts. For example, while the global ease of communication has broken down geographical and cultural barriers, it has also potentially undermined social cohesion and community integration. Critics argue that digital connectivity can reduce incentives for cultural assimilation and foster social fragmentation, trends that may extend into AI’s influence on trust and security, as improved algorithms also empower bad actors such as scammers and fraudsters.

Businesses that initially gained productivity advantages from innovations like smartphones or AI often see those benefits dissipate as competitors quickly adopt similar technologies. This dynamic creates a constant “arms race,” where continuous innovation is required merely to maintain competitive parity rather than achieve lasting margins. Slok emphasizes that companies are willing to invest heavily in AI if it can sustain superior profitability, but in a market where capabilities become rapidly commoditized, the incentive to pay premium prices diminishes, raising questions about the long-term economic viability of large-scale AI infrastructure projects.

Although proponents contend that AI-driven productivity improvements may yet materialize, industry experts caution that the current evidence calls for measured expectations. The risk remains that the anticipated AI revolution could slow or fall short of its transformative promise, leaving behind significant financial and societal costs.