BofA warns of global economic risks from AI data center boom

The risk of an AI correction

Bank of America Global Research economists recently warned that a sustained correction in the AI trade represents a major risk to the global economy. They noted that as investors become increasingly concerned about the sustainability of the AI trade, the risks associated with rising capital expenditure grow larger. A significant slowdown in this sector could directly impact economic growth and the wealth effect that has supported consumer spending.

The report suggests that the current boom is tied to the massive amount of borrowing required to fund the buildout of infrastructure. If the stock gains that have propped up spending begin to reverse, the broader economy could feel a sharp contraction. This warning comes as the market begins to show signs of divergence between AI-heavy stocks and the rest of the market.

Analysts have noted that while AI stocks have driven much of the recent market activity, the rest of the market has started to pull its own weight. For example, during a recent earnings season, the S&P 500 excluding AI stocks rose 5% while the broader index fell 3%. This suggests that the economy is increasingly reliant on the success of a few specific technologies.

The physical cost of data centers

The physical infrastructure required to power AI is expanding at a record pace. According to the Census Bureau, spending on data-center construction hit a preliminary annual rate of $84.95 billion in August, which is a 73% increase from the previous year. This massive investment is necessary to house the hardware that processes the vast amounts of data required for generative AI models.

The scale of this investment is significant, with AI-related hardware spending estimated to be about 1.9% of U.S. GDP. Major tech companies are on track to spend roughly $800 billion on capital expenditures this year, a 96% increase. Goldman Sachs analysts expect this spending to climb even further to approximately $1.2 trillion next year.

However, this rapid expansion is creating a physical squeeze on resources. Experts expect the pressure on construction workers and materials to intensify by 2027. While the buildout creates jobs, it also creates friction in the supply chain and puts a strain on the availability of materials needed for traditional construction projects.

Local backlash and political friction

As the physical footprint of data centers grows, local communities are beginning to push back. On a single Tuesday, voters in 10 states will decide on 41 different local ballot measures regarding data centers. These measures range from new requirements for developers to outright bans on construction in specific areas.

The political backlash has already reached high levels of government. In Texas, Governor Greg Abbott halted new data-center approvals pending an audit following complaints from rural communities. Meanwhile, in Ohio, a senator is facing scrutiny over his past support for these facilities as the state prepares for 21 different data-center measures on the ballot.

Despite this local friction, the construction is moving forward. While some states have already passed measures against development, many analysts believe the focus may shift from local bans to federal regulation. Goldman Sachs suggests that while a national moratorium is unlikely, there may be a move toward basic federal oversight and self-testing requirements for the industry.

Impact on the labor market

The integration of AI into the workforce is becoming increasingly visible in recent months. Data shows that 24.6% of U.S. businesses now use AI in at least one business function, with a significant portion of firms expecting to adopt it within the next six months. Adoption is particularly high in the finance, professional services, and information sectors, where it is approaching half of all businesses.

This adoption has a measurable impact on employment trends. Industries with higher AI adoption are currently showing weaker job growth and higher unemployment rates. Specifically, jobs in marketing, graphic design, customer service, and some tech roles are seeing the most significant impact from these technological shifts.

The numbers reflect a growing trend of displacement. Layoffs attributed to AI reached 4,000 in September and have totaled 207,000 since May 2023. Conversely, the construction of the data centers themselves is creating a different kind of labor demand, with construction jobs tied to these facilities up by about 315,000 since 2022.

The road ahead for AI spending

While the AI boom continues to dominate headlines, other parts of the economy are showing signs of cooling. The Goldman Sachs Analyst Index, which tracks the sentiment of stock analysts, fell below 50 for the first time since April 2025. New orders for many products have fallen significantly, reaching their lowest levels since May 2020.

Looking toward 2027, analysts identify AI spending and adoption as a primary risk factor, second only to potential conflicts in Iran. The focus of the debate is expected to shift from the physical construction of data centers to the regulation of the technology itself. There appears to be little appetite for highly restrictive rules, but some form of federal oversight is likely.

The ultimate success of the AI trade depends on whether the massive capital expenditures can translate into sustainable economic growth. If the investment fails to produce significant returns, the broader economy could face a correction as the wealth effect from stock gains reverses.