The explosive growth in artificial intelligence has ignited a fierce debate over whether the sector is in a speculative bubble. While bullish on AI’s transformative potential, top economist Jeremy Siegel warns of a significant, underappreciated risk reminiscent of the dot-com era: the potential for plummeting costs to undermine massive capital investments. Siegel argues that the current frenzy of spending on data centers and infrastructure to scale AI could face poor returns if new, far cheaper methods to power the technology emerge. He draws a direct parallel to the late 1990s, when the dot-com bubble burst not because the internet failed, but because technological advances, like vastly more efficient fiber-optic cables, suddenly made the earlier massive infrastructure investments seem excessive and uneconomical.
This concern over potential overreach is echoed by other analysts. Neil Shearing of Capital Economics notes that the current investment boom is comparable to historical technological surges, from railways to the internet, which almost always involve a degree of overextension where some projects fail to pay off. The central risk is that companies committing billions in capital expenditure (capex) today may find their investments devalued by more efficient and cost-effective solutions tomorrow, leaving investors with losses.
Despite these warnings, a prevailing optimism about AI’s long-term impact endures. Commentators note that even during periods of speculative excess, the infrastructure built often lays a critical foundation for future growth. Just as the fiber-optic network built during the dot-com boom later proved invaluable, today’s AI data centers could similarly enable widespread economic diffusion and innovation, even if the financial returns are uneven. Therefore, while the path may involve financial waste and casualties for specific firms and investors, the aggregate investment is seen as a necessary, if turbulent, part of technological adoption and economic progress.
Ez a cikk a Neural News AI (V1) verziójával készült.