Data centers have emerged as a national issue during this midterm election cycle, and candidates for the House and Senate have seized on it in their campaigns.
But state and local officials actually have more sway over data center construction and energy infrastructure, so the winners of lower-profile races may hold the economy’s future in their hands.
Hyperscalers have been spending hundreds of billions of dollars a year, with forecasts for 2027 topping $1 trillion. JPMorgan has estimated AI-related investment was responsible for about one-fifth of GDP growth over the past year. Analysts at ING believe AI will account for a third of growth in 2026.
With Election Day just over a week away and early voting already in full swing in many states, expectations for big Democratic gains could weigh on the stock market, which has sold off in recent days on fears the AI boom will go bust.
If there is a further stock slump, it will be because a “blue wave” at the state and local levels will usher in lawmakers opposed to data centers, Thierry Wizman, global FX and rates strategist at Macquarie, said in a note on Monday.
“For the direction of the U.S. economy, the composition and direction of governance at the state level may matter more than whether the Democrats ‘win’ the House or the Senate,” he added.
That dynamic was on display on Tuesday, when Amazon canceled a plan to build data centers in Loudoun County, Virginia, after reaching an agreement with local officials amid widespread backlash from residents.
Amazon Web Services bought property for the project in February for $427 million, but the purchase reportedly blindsided county leaders who were already growing uneasy with the explosion of data centers in recent years.
Amazon vowed not to build data centers on the land and will repurpose most of it for mixed-use and residential development, while also donating 25 acres for a community arts center and recreational facility plus $5 million to a housing fund.
And in Texas, another AI industry hot spot, Republican Gov. Greg Abbott has turned from a data center booster to a skeptic as he faces a tough re-election race against Democratic state representative Gina Hinojosa.
Michael Pearce, chief U.S. economist at Oxford Economics, also observed that midterm races for state capitals and town halls represent a bigger risk to the data center buildout than what happens in Washington, D.C.
The Trump administration has already used regulations and executive orders to facilitate construction, and any attempt to pass comprehensive legislation in Congress will be slow, he added.
By contrast, local officials pose more immediate threats via moratoriums, environmental reviews, and bottlenecks in connecting new data centers to the power grid, he said in a note on Thursday.
The key midterm races for data centers are in Texas and the Carolinas as well as Midwest states like Ohio, Illinois, and Wisconsin, according to Pearce. And the outcome of the midterms could have implications for the presidential election cycle.
“If this year’s elections show that an anti-data center stance wins political contests, it could put data center regulation at the heart of the 2028 campaign,” he said.
Perhaps in anticipation of a more hostile political climate, data center construction shifted sharply higher in the middle of the year, Pearce wrote. Meanwhile, developers can soften local opposition if they can show that a data center will lower local tax bills or fund schools and roads.
Still, even if lawmakers manage to pass a moratorium on new data centers, it won’t immediately slow the economy. Most of their economic value comes from the computing equipment inside, and that spending lags construction by six to eight quarters, Pearce estimated.
“This year’s surge will keep supporting business investment over the next couple of years, whatever happens on the ground,” he predicted.
Any slowdown in spending may actually be beneficial for the AI boom, according to Daniel Skelly, a portfolio manager at Morgan Stanley Wealth Management.
Despite the recent stock market volatility, the sector still enjoys solid growth and strong earnings, while the long-term AI investment cycle remains intact, he said in a note Wednesday.
“Also, ongoing power deficits, a shortage of people to build data centers, and political pushback against them may elongate the cycle and lower the odds of a blow-off top,” Skelly added.
This story was originally featured on Fortune.com