With midterm elections approaching in November, the US is witnessing intensifying public and political opposition to data center projects and the wider AI industry, even as corporate and private capital investment in AI continues to accelerate. With the backlash cutting across the country’s usual partisan lines, Felix Cook examines the phenomenon and how sagacious investors are centering political-risk analysis and investigation in their pre-transactional due diligence.
On September 1, US President Donald Trump took to Truth Social with a characteristically blunt message about the local politics of the US AI boom. “The only reason that communities throughout the USA should not want Data Centers is if they want to end up being backwards and poor,” Trump posted. “If we kill the Golden Goose, you will only have yourselves to blame!” For market observers, the necessity of a presidential intervention on the subject illustrates a new reality: US public opinion has shifted rapidly against the AI industry this summer in an unusually bipartisan backlash, even as AI garners an increasing portion of corporate and private capital investment. At the same time, many of these investments are dependent on complex webs of tax incentives and other fragile political inducements. For investors and deal makers already stretching to assess the economics, return, and long-term viability of emerging technology projects, increasing political risk adds complex but critical considerations into their diligence process.
An AI revolt
Per Gallup, the percentage of Americans who believe AI does more harm than good has increased from 31% in 2025 to 39% in 2026, rising to 47% among adults 18-29. Most Americans express little or no trust in businesses to use AI responsibly, with 73% trusting businesses ‘not much’ or ‘not at all.’ In an August poll by the Annenberg Public Policy Center, 76% of Americans stated that big tech should be more regulated and 60% said the AI rollout should be slowed. Anthropic CEO Dario Amodei has acknowledged shifting public sentiment on social media, writing that the AI industry faces “a crisis of trust” in which the public “always suspect that we are cooking up some new way to screw them over.”
Most relevantly for investors, building the physical infrastructure of AI – the capital-hungry data centers that power AI technology – is also increasingly unpopular, a rapid turnaround from just a few years ago when data centers were widely welcomed as harbingers of economic regeneration. In May, Gallup found 7 in 10 Americans oppose constructing data centers in their local area, including 48% who are strongly opposed. Local and national opposition campaigns have crystallized around environmental and cost-of-living concerns. Approximately half of opponents cite data centers’ use of resources in opposing projects; 20% mention concerns around higher utility costs and public subsidies; and 16% mention noise, air, or water pollution. By a striking, 18-point margin, Americans would now prefer building a nuclear power plant in their neighborhood than an AI data center.
Warning lights
There’s a hoary political aphorism in Washington, DC: A hit dog hollers. For all of President Trump’s public boosterism, some read his statements as evidence that mounting public opposition to AI is becoming a political problem for the President and his Republican Party ahead of pivotal midterm elections. With Trump’s approval ratings reaching record lows, even his political allies are viewing his ‘Let Data Reign’ AI policy with growing nervousness. US Senate Majority Leader John Thune has declined to endorse Trump’s comments about data centers, saying, “These entities need to be self-sufficient…they shouldn’t be going places where there isn’t support for them at the local level.” Thune’s own state, South Dakota, is one of a growing number of reliably Republican jurisdictions where lawmakers are under pressure to suspend or retract data center incentives in the face of opposition from rural communities. Another is Texas, where Governor Greg Abbott this month paused approvals for new data centers while state regulators audit tax breaks and utility usage. The backlash at the local level has been less visible but even more striking: last week in tiny Independence, Missouri, best known as the hometown of President Harry Truman, 68% of voters voted to immediately recall a city councilman over his support for a multibillion-dollar data center. Similar stories are playing out across the country, as usually sleepy local elections give voters a direct way to express their disquiet with the AI revolution.
With Republican incumbents on the back foot, Democrats have taken up the AI backlash as an issue that energizes their supporters while dividing the opposition. The localism of data center protests makes them a potent ‘wedge’ issue in an era in which all politics has become increasingly nationalized. In Michigan, where an open Senate seat is in contest, Republican candidate and Trump ally Rep. Mike Rogers came out last week in support of a one-year pause on building new data centers, after his Democratic opponent, Abdul El-Sayed, called for stricter regulation and a construction moratorium until Congress can enact national standards. Democrats who once positioned their states as hubs for data center investment have rolled back their stances, with Pennsylvania governor Josh Shapiro issuing an executive order requiring local community approval for projects despite previously wooing billions of dollars in tech investment, and Virginia governor Abigail Spanberger signing a first-in-the-nation electricity consumption tax on data center operators in July. Those Democrats who have resisted this turn have paid a political cost: in April, Maine Governor Janet Mills vetoed a proposed statewide moratorium on data center development, and went on to suffer a high-profile primary defeat.
Structural vulnerabilities
These political headwinds are important for deal teams assessing AI-focused transactions as their economics can be unusually dependent on public mood. Polling suggests that Americans still see few permanent jobs or broader economic benefits from data centers, while their footprint in terms of land use, noise, and utility draw is highly visible. These projects therefore lack the usual constituencies of local support that surround a conventional industrial asset: employees, labor unions, and upstream and downstream businesses to advocate on their behalf in the political sphere. At the same time, with the long-term economics of AI still forming, many projects are currently underwritten by long-term property tax abatements, sales tax exemptions, and negotiated utility rates. These political inducements must be honored and renewed across multiple election cycles and administrations at the local, state, and federal levels, and a change in any of these commitments could abruptly convert an attractive asset into a stranded one. For investors, this means a deal’s value and projected returns may rest on arrangements that are politically contingent rather than contractually secure.
Political risk in the sector will worsen before it gets better. With some prominent tech figures having embraced Trump at the start of his second term, industry anxiety has been rising regarding the possibility of hostile oversight and damaging investigations should Democrats retake the House of Representatives, as currently widely projected. The end of the midterms will mark the beginning of the 2028 presidential cycle, and with skepticism of AI and big tech an increasingly powerful current in the Democratic primary electorate, candidates will stake out competitively tough stances. If a Democrat does retake the White House, they are likely to have a far more combative posture towards the industry than the previous Biden administration, wherever they sit on the party’s ideological spectrum. Nor can the industry reliably depend on Republican patronage: Trump’s Vice President, JD Vance, who rose to prominence on his Appalachian conservative populism, has been notably more muted on the issue than his boss. Other would-be inheritors of Trump’s MAGA movement have taken openly hostile positions. Steve Bannon, a former Trump White House adviser and prominent thinker on the American right, has called for greater AI regulation and criticized Trump’s embrace of big tech, saying, “They have to be confronted…They’re totally out of control.”
Conclusion
Understanding the shifting landscape of political risks surrounding AI projects requires expert intelligence gathering and on-the-ground insights beyond standard legal and financial due diligence. This kind of investigation can confirm not only the present political conditions and value-adds to a transaction – the existence of a tax abatement, for example, or preferable utility contract – but also the composition of the body which granted it, the key constituencies supporting it, and the likelihood of positive or detrimental change within short, medium, and long-term time frames. While the fundamentals of AI-focused transactions may remain attractive, deal teams, portfolio managers, and their counsel are increasingly tracking key political and regulatory figures, election forecasts and polling, ballot initiatives, local hearings, and social media and local news reporting, which often flags changes in public sentiment well ahead of elections and legislation. Where open-source information is insufficient, human intelligence and other beyond-public-record inquiries are proving critical in converting intelligence and rumor into actionable insights and forward planning. Taken together, these efforts ensure political risks are properly priced at entry, managed during investment, and framed effectively on returning to market.
