Data center construction limits could prevent an AI investment bubble

A natural brake on the boom

Recent US midterm elections have introduced a new variable into the AI investment cycle: regulatory and local government pushback. As the demand for data centers—the physical facilities that house the servers powering AI—skyrockets, many states are beginning to push back against the rapid construction.

Brett Ryan, an economist at Deutsche Bank Securities, told Bloomberg that these moratoriums or regulatory pressures might actually be a positive sign for market stability. He suggested that if these hurdles slow down the buildout, it could prevent the AI investment from blowing into a massive bubble. This perspective views government intervention as a necessary cooling mechanism for an overheated industry.

Ryan noted that these new limits provide a convenient reason to cancel projects that were already looking financially unstable. By creating a 'natural brake,' the current political climate may help ensure that the growth of AI infrastructure remains sustainable rather than speculative. This shift suggests that the era of unchecked expansion may be meeting a significant wall of local opposition.

The geography of the limits

The impact of these new limits is concentrated in specific high-growth regions. According to estimates from Goldman Sachs cited by Bloomberg, approximately 45% of planned data-center capacity is located in states where newly elected Democrats campaigned on imposing limits on new projects.

Texas stands out as a primary area of concern for the industry, accounting for 25% of the total planned capacity. The state's governor-elect, Gina Hinojosa, has already vowed to halt new approvals for these facilities. Other states, including Georgia and Ohio, are also seeing proposals for stricter conditions or statewide pauses on construction.

While these moves are significant, some analysts believe the restrictions may be temporary. Sarah Bianchi, a strategist at Evercore ISI, told Bloomberg that she expects these state-level moratoriums to last for no more than a year. However, even a short-term pause can have a lasting effect on the long-term investment forecasts for the entire sector.

Winners and losers of the slowdown

The effects of slower construction are expected to be split between different types of companies. Anna Wong of Bloomberg Economics noted that a slowdown could deepen financing stress for leveraged developers and 'neocloud' companies, which are smaller cloud service providers that rely on heavy debt to build infrastructure.

Conversely, the situation might favor 'hyperscalers'—the massive tech giants like Amazon, Google, and Microsoft. Because these companies already own significant infrastructure, tighter computing capacity could actually preserve their pricing power. They can afford to wait out the construction delays while smaller competitors struggle to find space to build.

Amazon Web Services (AWS) chief executive Matt Garman recently commented on the situation, stating that the company is in a 'pretty good spot' regarding its expansion plans. He suggested that while some others may have more work to do to navigate the new restrictions, the largest players are currently well-positioned to weather the regulatory storm.

The long-term investment outlook

Despite the new hurdles, economists do not anticipate an immediate collapse in the AI sector. Much of the capacity scheduled to come online in the next year is already under construction, meaning the current momentum is difficult to stop overnight. The real impact of the election results will likely be felt in the 2028 outlook.

Deutsche Bank had already been projecting a slowdown in investment growth by 2028. Following the recent elections, Brett Ryan stated that his 'knee-jerk reaction' was that he would not need to raise his 2028 forecast anytime soon. This indicates that the market was already pricing in a cooling period, and the new regulations may just be confirming that trend.

The scale of the current spending is still massive, with monthly construction spending nearly doubling to an annualized $100 billion. While the pace of growth may be moderating, the sheer amount of capital currently flowing into information processing equipment and software continues to provide a significant boost to the US economy, even as it eclipses the peak of the 1990s dot-com boom.

From chips to permissions

This story highlights a fundamental shift in what limits the AI industry. For a long time, the primary constraints on AI development were the availability of advanced chips and the amount of electricity required to power them. Now, the permission to build the physical structures is becoming a primary binding constraint.

As the industry matures, it moves from a phase of pure technological possibility to one of physical and political reality. Operators and investors must now navigate a complex landscape of local zoning laws, environmental concerns, and state-level politics as much as they navigate hardware specifications. This transition marks a new chapter in the AI supercycle.

For those following the industry, the next few months will be critical to see which states actually enact these moratoriums and how they are worded. Whether these limits result in a permanent slowdown or a temporary bottleneck will determine the trajectory of AI infrastructure spending over the next three to five years.

The data at a glance

To understand the scale of this shift, it is helpful to look at the specific numbers involved in the current construction boom. The following figures illustrate the magnitude of the infrastructure being built and the concentration of that growth in the United States.

• $100 billion: The annualized amount of monthly construction spending on data centers, which has nearly doubled over the last 12 months.
• 45%: The share of planned data-center capacity located in states where newly elected Democrats campaigned on construction limits.
• 25%: The portion of total planned data-center capacity located in Texas alone.
• $50 billion: The estimated annualized spending on data-center construction from one year ago, based on the current doubling of costs.

Key takeaways

The AI boom is facing a new type of friction. While the demand for AI remains high, the physical ability to build the necessary infrastructure is being challenged by local governments and new regulations.

This creates a bifurcated market: large tech giants with deep pockets may maintain their dominance, while smaller developers face higher costs and more difficulty securing permits. For the broader economy, these limits may serve as a necessary 'brake' to prevent a speculative bubble from bursting too violently.

Investors and observers should watch for the specific language of new state moratoriums in Texas, Georgia, and Ohio. These policies will determine whether the 2028 forecast for AI investment remains steady or faces a significant downward revision as construction projects are canceled or delayed.