The anticipated $10 trillion investment surge in artificial intelligence (AI) infrastructure in the United States is encountering significant challenges, including project delays, public opposition, and investor skepticism. These issues threaten to disrupt the extensive data centre expansion that underpins the country’s AI ambitions.

Over the past several years, a rush to secure computing capacity has fostered what industry insiders describe as a “Wild West” environment, with rapid funding flowing into emerging companies and due diligence often abbreviated. Leading AI research organizations such as Anthropic and OpenAI have reportedly prioritized speed over short-term cost efficiency to capture competitive advantage.

However, after more than three years of rapid growth, reality is setting in. At least 47 major AI data centre projects have experienced delays, cancellations, or have been blocked in 2024, according to regulatory intelligence data that is considered a conservative estimate. Additionally, more than 1,000 communities across the United States have sought to impose moratoria or restrictions on AI infrastructure construction this year, reflecting growing local opposition to the scale and impact of these developments.

The planned expansion relies heavily on a new class of “neo-cloud” data centre providers tasked with financing and delivering unprecedented construction efforts. Yet, rising interest rates and increased borrowing costs, combined with a volatile bond market, have heightened financial risks. Executives in the sector report that securing financing for data centres has become substantially more difficult. Investor confidence has been shaken, with shares of companies such as CoreWeave—a cloud services start-up—declining nearly 25% over a recent week. Debt instruments tied to AI infrastructure providers have also suffered sharp sell-offs.

Obtaining necessary permits and utility approvals has become increasingly problematic. Regulators are reportedly denying grid connections for new projects unless developers commit to supplying their own power, adding to construction complexities and costs. These hurdles contribute to delays that put contractual obligations at risk, as tenants—including Big Tech firms and AI labs—could potentially cancel agreements if projects are not completed on schedule.

Despite these challenges, AI labs and major technology companies remain committed to massive expenditure on computational capacity. Estimates from economist Stijn van Nieuwerburgh of Columbia Business School project U.S. spending on AI infrastructure to exceed $10 trillion between 2025 and 2032—an amount roughly equivalent to 3.6% of annual U.S. GDP during that period.

Market sentiment has been further dampened by postponements or cancellations of planned initial public offerings by several AI infrastructure developers and cloud providers. Notably, OpenAI’s decision to delay its IPO has contributed to uncertainty regarding the sector’s near-term prospects.

In sum, the U.S. AI infrastructure boom faces a critical juncture as financing challenges, regulatory obstacles, and community pushback converge, potentially impeding the delivery of the large-scale computing capacity that emerging AI technologies depend upon.