In the days following the midterm election, stock markets have moved sharply in both directions, with technology and AI-related stocks serving as the primary driver of this movement. Among the most closely watched issues for financial markets is the future outlook for data center construction.
The buildout of data centers has played a significant role in driving economic activity and corporate profits, as companies develop the infrastructure needed to support AI applications. The results of the midterm election, along with related ballot measures, indicate that voters broadly favor placing restrictions on this type of construction across various levels of government.
These developments add to the pressures already facing technology stocks. While this sector has contributed meaningfully to portfolio returns in recent years, it also faces ongoing questions about valuations, interest rates, and whether large-scale AI infrastructure investments will ultimately deliver returns. That said, volatility in these parts of the market is not unusual, and returns over longer time horizons have remained strong. What do the latest election results mean for markets, and how can investors keep a broader perspective?
Voters broadly supported restrictions on data center construction
As the accompanying chart shows, opposition to data center construction among voters was widespread. So far this year, there have been 47 local measures directly related to data centers, with 41 voted on during the midterm election across ten states. The results indicate that voters supported restrictions on data center construction by wide margins, with only one exception.
These measures varied in scope, addressing issues such as zoning, electricity and water usage, and regulations specific to data centers. For example, one measure in Kansas banned "high-impact" data centers, defined as facilities using more than 10 megawatts of electricity or 100,000 gallons of water per day, for 10 years. Towns in New Hampshire and Maine approved temporary moratoriums on new construction.
The one exception in the latest election was Harwood, North Dakota, a town of roughly 850 people, where about three-quarters of voters rejected stricter review requirements on large industrial projects. This outcome clears the way for construction to continue on a $3 billion data center campus. The reported reasoning was that a data center would likely have been built nearby regardless of how residents voted. By permitting it within town limits, the community could capture tax revenues and other economic benefits.
The tension between large infrastructure projects and the needs of local communities is not new, but it is accelerating due to the rapidly growing demands of AI applications.
Historical parallels include the construction of oil and gas drilling sites, where communities have long weighed environmental and quality-of-life concerns against the promise of jobs and tax revenues. The same "Not In My Backyard" (NIMBY) dynamic also arises with highways, power plants, cell phone towers, and similar projects. Data centers represent the latest example of a large industrial buildout affecting residential communities, raising concerns about power usage, water consumption, and the scale of these facilities.
Another parallel is the construction of large job-creating factories, which many towns and cities historically competed to attract. An extreme version of this was the industrial towns of the 19th and 20th centuries, where manufacturing plants, mills, and production lines were not just central to local economies, but had entire communities built around them. In many cases, these facilities brought economic benefits including jobs, tax revenues, and local investment. In hindsight, these towns had a mixed track record, and history shows that high dependence on a single economic engine can create challenges, alongside issues such as environmental concerns, noise, and strain on local infrastructure.
Data centers are following a similar pattern, with one key distinction: they are typically capital intensive, requiring substantial investment and equipment, but are not necessarily labor intensive. The fact that data centers do not tend to be a significant long-term source of local employment appears to be an important factor behind the resistance reflected in ballot results.
Data center construction has been a meaningful contributor to economic activity
Why have data centers become such a central focus? It helps to first understand what they are. When users visit a website or app, conduct a search, send an email, make an online purchase, or interact with an AI model, those requests are processed by servers. A data center is essentially a large warehouse packed with servers, ranging anywhere from a few dozen to 1,000 acres in size. These servers operate continuously, requiring electricity, cooling, security, and other supporting systems. Together, they form the physical infrastructure that underpins all digital services, a role increasingly dominated by AI activity.
Spending on these facilities has grown considerably in terms of its contribution to economic activity. As the chart above illustrates, data center construction spending, not including the hardware inside, has risen sharply since the launch of ChatGPT in late 2022. The Census Bureau estimates it has reached $96 billion on an annualized basis, well above the roughly $49 billion spent on all other office construction. This buildout is one reason the gross private domestic investment component of Gross Domestic Product (GDP), which measures business spending on structures, equipment, and more, accelerated to 3.3% in the third quarter, according to the Bureau of Economic Analysis.
It is important to distinguish between the potential long-term benefits of AI that many are anticipating, such as productivity gains, improved profit margins, and new business models, and what is currently driving AI-related trends. At present, it is the physical construction of infrastructure that has been propelling economic activity and financial market performance in the AI space.
This mirrors the railroad boom of the 19th century, where laying rail lines created jobs and demand for steel and other materials, separate from the longer-term economic gains that faster transportation eventually delivered. Restrictions on data centers may affect the pace of construction activity, but the longer-term trajectory of AI adoption will ultimately be determined by whether AI applications genuinely boost productivity and deliver broader economic benefits.
Technology and AI stocks have experienced notable volatility
As the chart illustrates, the Magnificent 7 group of technology stocks, which includes Meta, Amazon, Apple, Alphabet, Nvidia, Microsoft, and Tesla, has generated strong returns over the past several years. These stocks have gained around 140% since 2022 and rose about 340% from the start of 2023 to their recent peak, outpacing both the S&P 500 and the Nasdaq. However, the chart also reveals significant pullbacks along the way, including the current period of volatility.
The market's reaction has not been uniform across all AI-related companies. Semiconductor makers and so-called "neoclouds," which are newer public companies providing cloud computing services, experienced sharper declines in the days following the election. By contrast, large "hyperscalers," the companies funding the AI buildout, did not see their share prices pull back as much, although they have faced pressure due to concerns over cash flows.
Pullbacks of this nature are not unusual, even for companies with strong long-term fundamentals. In 2022, when interest rates rose rapidly, the Magnificent 7 declined by nearly 50% before recovering and delivering healthy returns for long-term investors. More broadly, transformational technologies have often encountered public and regulatory resistance during their buildout phases.
Today's volatility also reflects genuine uncertainty. It remains unclear whether new restrictions will prove widespread and lasting, or whether economic incentives will lead many communities to support data center development over time. For investors, this serves as a reminder that AI does not represent a single investment, but rather a long-term theme with the potential to support portfolios over time.
The bottom line? The midterm election has raised new questions about the pace of data center construction across the country. However, recent volatility is not unusual, and maintaining a long-term perspective with a well-balanced portfolio remains the best way to navigate these periods.