Bank of America has cautioned that the current wave of artificial intelligence investment could mirror the boom-and-bust patterns of previous capital cycles, as intensifying competition, falling prices, and excess capacity threaten to undermine returns despite the technology's long-term potential.
The warning arrives as billions of dollars continue to pour into AI infrastructure, including data centers, advanced semiconductors, and computing capacity. This spending spree has positioned AI as one of the largest industrial bets in U.S. history, but doubts are growing over whether companies can generate sufficient returns to justify the outlays.
Historical precedents such as railways, telecommunications, and the internet created immense economic value, yet investors often suffered losses when capital flooded into projects whose returns failed to cover costs. Bank of America analyst Sebastian Raedler noted that the AI industry may be repeating this pattern, where strong initial returns encourage aggressive investment, eventually leading to overcapacity and pricing pressure.
“Most major capital cycles of the past have ended in disappointment, even though the underlying technology often lived up to expectations,” Raedler said. The concern is that a technology can transform industries and deliver broad economic benefits without generating attractive returns for every company investing in it.
High profit margins in the early stages of an investment boom typically attract more entrants and additional capital. As capacity expands and competition intensifies, businesses can lose pricing power, eroding the returns that made the investment attractive in the first place. AI is already witnessing stronger competition at the model level, particularly from Chinese open-weight models. Average token prices have fallen roughly 50% since the summer, adding to worries about whether AI developers can sustain the profitability needed to support their infrastructure spending.
Investor unease is reflected in the market. The Philadelphia Semiconductor Index has declined for seven consecutive sessions, leaving it about 13% below its recent high. Interactive Brokers chief market analyst Steve Sosnick told Reuters that investors are increasingly questioning “the speed, the efficacy and the profitability of the rapid build-out in AI.” Sosnick added that a slowdown in data-center construction would eventually reduce demand for hardware, as customers would no longer be racing to buy equipment at almost any price.
Such a shift could affect chipmakers and other suppliers that have benefited from the AI infrastructure surge. If companies scale back construction plans, equipment purchases could weaken, particularly if falling AI prices make it harder to justify additional computing capacity. As infrastructure spending continues, companies are under pressure to demonstrate that their AI products can generate enough revenue and cash flow to support the investments behind them.
Despite the concerns, Federal Reserve Chair Kevin Warsh offered an optimistic view on Friday regarding AI's potential economic benefits. “It's obviously hitting the demand side of the economy before it's going to hit the supply side of the economy,” Warsh said. “We don't know in the next 12 or 18 months its (AI) effect on employment, but if you believe like I believe, which is it will make economies and countries that put it to good use more productive, that it's going to make it so that we will have more wealth, then that has to be a very good thing.”
Warsh's comments highlight the possibility that AI could improve productivity and generate broader economic gains as businesses adopt the technology. However, the potential long-term benefits do not eliminate the financial risks associated with the current investment cycle.
This article is for informational purposes only and does not constitute financial advice.
