State and Local Races May Shape Data Center Expansion and AI Growth in Midterm Elections

State and local officials have more influence over data center construction and energy infrastructure than federal lawmakers, so lower-profile midterm races may shape the economy's future. Data centers have become a national campaign issue for House and Senate candidates. Hyperscalers spend hundreds of billions of dollars annually, with forecasts putting spending above $1 trillion in 2027. JPMorgan estimated that AI-related investment generated about one-fifth of GDP growth over the past year. ING analysts believe AI will generate a third of growth in 2026.

Election Day is just over a week away, and early voting is well underway in many states. Expectations of large Democratic gains could pressure stocks, which have fallen in recent days amid concerns that the AI boom will collapse. Thierry Wizman, Macquarie's global FX and rates strategist, said in a Monday note that further stock declines would reflect Democratic gains installing state and local lawmakers opposed to data centers. Wizman argued that state-level governance may matter more for the US economy's direction than Democratic control of either congressional chamber.

Amazon canceled plans for data centers in Loudoun County, Virginia, on Tuesday after an agreement with local officials amid widespread resident opposition. Amazon Web Services paid $427 million for the project property in February. The purchase reportedly surprised county leaders, who were already increasingly concerned about rapid data center expansion in recent years. Amazon pledged not to construct data centers on the property and will convert most of the land to mixed-use and residential development. Amazon will also donate 25 acres for a community arts center and recreational facility, plus $5 million to a housing fund. In Texas, Republican Governor Greg Abbott has shifted from supporting data centers to expressing skepticism while facing a difficult re-election contest against Democratic state representative Gina Hinojosa.

Michael Pearce, chief US economist at Oxford Economics, said state and local midterm races pose a greater risk to data center expansion than developments in Washington. Pearce said the Trump administration has already eased construction through regulations and executive orders. Pearce added that any effort to enact comprehensive congressional legislation will proceed slowly. In a Thursday note, Pearce said local officials present more immediate risks through moratoriums, environmental reviews and delays connecting new data centers to the electricity grid. Pearce identified Texas, the Carolinas, Ohio, Illinois and Wisconsin as locations of key midterm contests for data centers. Pearce said electoral success for candidates opposing data centers could make data center regulation central to the 2028 presidential campaign.

Pearce wrote that data center construction accelerated sharply around the middle of the year, perhaps in anticipation of a less favorable political environment. Developers can reduce local opposition by demonstrating that a data center will cut local tax bills or finance schools and roads. However, even a moratorium on new data centers would not immediately weaken the economy. Pearce estimated that computing equipment accounts for most of the economic value of data centers, with equipment spending following construction by six to eight quarters. Pearce predicted that this year's construction surge will therefore support business investment over the next couple of years, regardless of developments affecting construction.

Daniel Skelly, a portfolio manager at Morgan Stanley Wealth Management, said slower spending could benefit the AI boom. In a Wednesday note, Skelly said the sector retains solid growth and strong earnings despite recent stock market volatility. Skelly said the long-term AI investment cycle remains intact. Skelly argued that electricity shortages, insufficient data center construction workers and political opposition may extend the investment cycle and reduce the likelihood of an unsustainable market peak.