US interest rates pose biggest threat to billion-dollar AI investments

A report by consultancy Gavekal, authored by analyst Will Denyer, warns that rising US interest rates have become the primary threat to the current cycle of investment in artificial intelligence infrastructure. After the Federal Reserve raised its benchmark rate on Wednesday (16), 10-year Treasury yields surpassed 5% for the first time since 2007, increasing the real cost of capital for AI projects.

Despite recent calls from Dario Amodei, CEO of Anthropic, for the AI industry to slow its development—a stance supported by leaders from OpenAI and xAI—Denyer believes this will not halt the flow of resources. He argues that any lab that slows down will be overtaken by competitors, making a coordinated pause unlikely.

According to Denyer, the Fed’s monetary tightening is already driving up financing costs and could push the so-called Wicksellian spread—the difference between the real return on capital and its cost—to a critical level. If this indicator falls below the historical median of the past 20 years, the risk of a widespread investment slowdown increases significantly.

The impact of regulatory restrictions is also limited. Moratoriums in New York and Texas, along with other local barriers, are expected to delay only about 2.3 gigawatts of planned capacity in the United States. Meanwhile, projections indicate an additional 38 gigawatts in 2027, compared to 18 gigawatts in 2026 and 10 gigawatts in 2025. Consultancy SemiAnalysis estimates that the sector will continue to expand, driven by demand for chips and energy.

Competitive pressures sustain progress

Jensen Huang, CEO of Nvidia, says there is no reason to slow down, especially with Chinese competition putting pressure on the market. Denyer notes that, regardless of decisions made by leaders like Amodei, Altman, and Musk, there will always be someone pushing the technological frontier forward as quickly as possible.

Gavekal compares the current situation to the fracking boom in the United States, where fears of political backlash did not prevent oil production from growing. The same could happen with data centers, which remain essential for AI expansion.

Demand for computing capacity remains strong

So far, there are no signs of reduced demand for processing power. Companies across the AI supply chain are reporting strong results, and the return on investment for the US corporate sector reached a record high in the last quarter. This contrasts with the telecommunications bubble of the late 1990s, which collapsed when returns failed to sustain investments.

Although rising interest rates do not yet justify a massive shift from stocks to fixed income, bonds already serve as a hedge against the risk that high capital costs could end the AI investment cycle. Paradoxically, a potential decline in AI investment would have deflationary effects, making bonds an attractive alternative even in an extreme scenario.