AI is fueling economic growth, but its expansion is also triggering political battles over energy, water, jobs, and local communities.

It is difficult to escape artificial intelligence (AI). It has seeped into daily life, crept into the office space, and become a key driver in the US economy. Indeed, AI and AI-related companies have and most likely will continue to power US markets higher into record territory.  Despite the AI penetration of every aspect of life, it is also feared as the grim reaper of jobs for humans, an invader of privacy, and, potentially, a political tool to surveil what people read, who they communicate with, and what their views are.

Pros and cons. AI is not going away, and it is having a tremendous impact on national, state, and local economies. It touches upon everything from water rights and energy use to the environment. AI data centers use substantial volumes of water; they are very hungry for energy; and questions have been raised about everything from pollution to the impact of the constant hum emanating from AI data centers. The AI industry also offers jobs, investment in local economies, and greater tax revenues. And it is essential to US national security in its rivalry with China.

AI plays an important role in driving growth in the US economy. According to Goldman Sachs, AI infrastructure spending will be close to $765 billion in 2026, equal to around 2 percent of the US economy. It is estimated that total AI capital expenditures could rise above $1 trillion in 2027 and around $3 trillion by 2035, possibly even higher, according to some sources. As one report noted, the “AI buildout is on track to surpass the Gilded Age’s so-called ‘Railway Mania’ to become the biggest infrastructure project in US history.” 

Where AI Data Centers Are Expanding 

Where are the AI data centers and other projects? According to Cleanview, a market intelligence company tracking US power infrastructure and data center development, there are 954 operating data centers and 1,254 planned data centers. Current operating capacity stands at 42,853 megawatts (MW), with a planned capacity of 342,456 MW. These numbers indicate that there is going to be a large step-up in power needs; the process is just starting.

The states that are leading the way in AI data center development are Texas, California, Illinois, Ohio, Utah, and Virginia. The largest concentration is in Virginia, mainly in Loudoun, Fairfax, and Prince William counties. Loudoun County has even gained the moniker “Data center capital of the world.” 

Construction of AI data centers is unpopular with many people, not to mention unease about what AI portends for the job market. In one Gallup poll, 48 percent of Americans strongly opposed local construction of AI data centers, with another 23 percent somewhat opposed.Only 7 percent strongly favor AI data local construction, with 23 percent somewhat favoring. This means that seven out of every 10 Americans do not want to see large, massive humming structures nearby, impacting water usage, energy consumption, and quality of life. 

While some of these concerns are brushed off as the plaintive voices of neo-Luddites, there is a growing political backlash in local communities in Texas, Utah, and Virginia. Indeed, the perception of massive AI companies and hyper-scalers (Amazon, Microsoft and Google) is that they have little regard for the local communities and are going to force through “progress” over local opposition. As Financial Times’ Camilla Cavendish catches the mood: While some tech bosses talk inspirationally about scientific breakthroughs, other plutocrats breach copyright, stand by while their social media platforms harm mental health, demand more energy to power their ambitions with scant concern about climate change—and then boast with apparent glee about the eradication of white-collar jobs.

AI Data Centers Become a Political Issue

Politics are increasingly front and center with AI data center construction. Congress has been slow to act due to the complicated nature of the issues surrounding AI construction, including permitting reforms, power transmission, and the interrelationship between federal, state, and local regulations. Lobbyists also play a major role in pushing their agendas in Washington as well as putting money behind candidates in elections.

Congress is finally considering a bill to force tech companies to pay for AI’s strains on the power grid, as voters across the country express dissatisfaction with data centers driving up utility costs. Known as the Ratepayer Protection Act, it would require state utilities to consider creating a “large load standard” that would force data center builders to pay for upgrades to the grid needed to power them. While this bill applies federal pressure on large tech companies and other heavy energy users, the enforceable protections for the public depend on state legislatures, utility commissions, and governors. 

Local Communities Push Back Against AI Projects 

Not all the political battles are fought in Washington; many of them are fought in packed county meetings and town halls, as in Texas and Utah. 

In the June 2026 Republican primary in Utah, the president of the Utah State Senate, J. Stuart Adams, lost to Stephanie Hollist, a former university professor. What helped her knock out one of the most powerful persons in Utah politics was the sentiment that Adams and Utah’s political establishment were lacking transparency and ignoring their own voters by approving a data center project pushed by the celebrity investor and “Shark Tank” personality Kevin O’Leary.

One outcome of the backlash against the AI data center juggernaut is the search for new locations in more rural parts of the United States. Virginia and Texas have the most planned data centers (287 and 170, respectively). Regionally, the South has 754 planned data centers and 1,209 existing ones; the Midwest has 419 planned data centers and 655 existing ones; the West has 277 planned data centers and 807 existing ones; and the Northeast has 106 planned data centers and 397 existing ones, according to Pew Research.

Could AI Data Centers Become Stranded Assets? 

There is one other factor to consider: Do AI data centers become stranded assets at some point in the future, when new technology reduces their current role as core to the AI revolution, or will there be a need to repurpose such facilities? A 2026 report from the American Council for an Energy-Efficient Economy (ACEEE) suggests that AI data centers risk becoming stranded assets because of their hyper-specialized infrastructure, rapid hardware turnover, and power limitations.

No one remembers the New Haven and Northampton Canal (also called the Farmington Canal), but in the 1830s, it was seen as a cutting-edge innovation, a catalyst for linking the interior of Connecticut and western Massachusetts to the port of New Haven. Financed by private capital, the canal was eventually eclipsed by new technologies, steam-powered ships on the larger Connecticut River, and by the still more deadly railroad. By the 1850s, the age of canals was over. Considering how history works, is there another technology lurking that will bypass the large humming structures and lower demand on power grids? 

Balancing AI Growth With Long-Term Risks 

The AI revolution and rapid expansion of AI data centers raise tough questions for state and local governments on how to responsibly balance the costs of large-scale construction in their communities and the long-term benefits. While the construction of hyperscale AI data centers requires extensive labor (possibly generating over 4.7 million temporary construction jobs), once built, these centers do not require large workforces. What happens if the AI bubblebursts? This throws out the question for many state and local governments, as well as public utility companies and their regulators—what are the trade-offs between short- and long-term gains, as well as the substantial demands on power and water resources?  

It was the Austrian economist Joseph Schumpeter who stated: “Economic progress, in capitalist society, means turmoil.” What is happening in the AI-driven world strongly reflects this.

About the Author: Scott B. MacDonald

Dr. Scott B. MacDonald is the chief economist for Smith’s Research & Gradings and a fellow with the Caribbean Policy Consortium. Prior to those positions, he worked for the Office of the Comptroller of the Currency, Credit Suisse, Donaldson, Lufkin and Jenrette, KWR International, and Mitsubishi Corporation. His most recent book is The New Cold War, China and the Caribbean (Palgrave Macmillan 2022).

The post How AI Data Centers Are Reshaping America’s Economy, Politics, and Power Grid appeared first on The National Interest.