Anti-AI Politics Is About to Get Big

NextFin News - Artificial intelligence spent the past three years winning the culture war and losing the neighborhood. A Gallup poll in March found 71% of Americans oppose a data center being built in their local area — including 48% who are strongly opposed — a higher level of local resistance than even nuclear power plants, where Gallup's historical peak opposition was 63%. Heatmap Pro's polling is more severe still: three-quarters of Americans would oppose a data center near where they live. The backlash is no longer a mood. It is becoming an organized, bipartisan political force with a playbook borrowed from the climate movement, a roster of state-level victories, and a growing list of candidates and elected officials who have made opposition to AI infrastructure a central plank.

The question this piece pursues is not whether Americans are anxious about AI — the polling answers that — but whether that anxiety is about to acquire the machinery of a durable political movement, and what that means for the companies racing to build the physical backbone of the AI boom. The answer points to a structural shift in the industry's social license, one that will raise the cost of capital and the timeline for AI infrastructure even as the underlying demand for computing power keeps climbing.

The Situation: A Movement, Not Just a Mood

In August 2026, a group called Irreplaceable — as in, people are irreplaceable by machines — formally launched with an audacious premise: that the diffuse public unease about artificial intelligence can be organized into a national movement the way climate activism was. Its leadership reads like a roll call of the climate left. Phil Aroneanu, the executive director, is a co-founder of 350.org. Jeremy Ornstein, a campaign strategist, is a former organizer with the Sunrise Movement. Their stated view is that the climate movement initially floundered by acting as policy wonks, and that modern movements are built on what people feel rather than what they think.

"People get it," Ornstein said. "People have the intuition that if we create machines more powerful than people, that could go dangerously and terribly wrong."

The tactical inheritance is visible. In July 2026, hundreds gathered in what was likely the largest anti-AI demonstration in San Francisco's history, marching from OpenAI's headquarters to Anthropic's and finally to Google's — with chant sheets, a marching band, and a slushy truck deployed to "freeze" AI. The group's priorities span the economy (mitigating automation's effects), civil liberties (opposing government and corporate surveillance), and catastrophic-risk guardrails (preventing blackouts from AI-powered hacking or bot-engineered pandemics). Climate-oriented organizations — including the environmental-justice arm of the NAACP, the Sierra Club, and Earthjustice — are supporting local data-center battles through lawsuits, model legislation, and protests.

The raw material for organizing is abundant and getting more so. Pew Research found in June 2026 that 63% of U.S. adults believe AI is advancing too quickly, while just 2% say it is moving too slowly. A separate Pew survey in late June found 52% of Americans are more concerned than excited about the increased use of AI in daily life, up from 37% in 2021. Job anxiety is the sharpest edge: 73% of adults under 30 expect AI to take away jobs over the next two decades, up from 61% in 2024, and 71% of all adults now hold that view, up from 64% two years earlier. A tracking project run by Athena Insights found that as of late August, 74% of Americans say AI is making job security worse, compared with 15% who say it is making it better.

Yet adoption continues apace. About half of U.S. adults now use AI chatbots, up from a third in 2024. This is the central tension of anti-AI politics: the technology is being embraced in daily life even as its physical and economic footprint is being rejected in communities.

Layer 2: Why This Time the Backlash Has Teeth

The Mechanism: AI Has a Physical Address

For most of the AI boom, the industry's externalities were abstract — job displacement in some distant future, bias in unseen algorithms, misinformation in feeds. Those are hard things to organize against. Data centers are different. They have a site, a fence line, a water connection, a substation, and an electricity bill that shows up on neighbors' utility statements. The backlash has found its target because AI, for the first time, has a physical address.

That physicality converts a collective-action problem into a local one. A data center can draw as much power as a small city and, by industry estimates, consume up to 3 million gallons of water a day for cooling. Residents do not need to understand transformer architectures to object to a six-building, 5-million-square-foot-plus complex going up across from a high school, as is proposed in South Whitehall Township, Pennsylvania. They need only read their electric bill. This is why the opposition is so broad: it recruits people who have no ideological position on AI at all.

The result is that "permission to operate" has become a scarce fourth input for AI infrastructure, sitting alongside chips, power, and capital. In the first three months of 2026 alone, local opposition blocked or delayed at least 75 data center projects worth roughly $130 billion, according to Data Center Watch — the most in any three-month period since the group began tracking in 2023, and roughly equal to all of the pushbacks recorded in 2025 combined. Over the two years through March 2025, the group counted $18 billion in projects blocked and another $46 billion delayed.

The Political Vehicle: Bipartisan, and Winning

The second reason this backlash has teeth is that it does not respect the usual ideological fault lines. A review of public statements by elected officials in districts with large data center projects found that 55% of the politicians who took public positions against the projects were Republicans and 45% were Democrats. On the left, the arguments run through environmental impact and resource consumption; on the right, through tax abatements, grid strain, and distrust of big-tech elites. Both wings converge on power consumption.

That convergence is now showing up in election results and candidate platforms. In Ohio, the Republican nominee for governor, Vivek Ramaswamy, released an "Ohioans-first data center pledge" calling for a moratorium on data center construction until legislation passes, and even proposed that data center companies cover the power bills of nearby residents. "A top concern I hear from Ohioans across our state — second only to property taxes — is the accelerating pace of data center expansion," he said. "The data center industry has badly failed to earn the trust of millions of everyday Ohioans." His Democratic opponent, Amy Acton, released her own "conditional moratorium" plan and attacked Ramaswamy for being too close to big-tech billionaires.

In Wisconsin, Democratic gubernatorial candidate Francesca Hong built a broad coalition around a data center moratorium. "It's time to press control+alt+delete to put a stop to this AI data center crisis," she said in a campaign video. In Michigan, Democrat Abdul El-Sayed won his Senate primary after rallying in front of the OpenAI-Oracle data center site in Saline Township. "Unless and until we protect local communities from the power of corrupting money and politics," El-Sayed said, "we're going to continue to see projects like this get pushed on to local communities despite themselves." In Michigan's 7th Congressional District, William Lawrence won a Democratic primary against two more moderate candidates after data centers became a central issue; four data centers had been proposed in the district since he declared his candidacy. "It wasn't something that I expected to be part of the campaign last August," Lawrence said, "because data centers weren't on our radar."

The victories are not confined to campaigns. In Arizona, Governor Katie Hobbs signed a state budget in June 2026 that includes a three-year moratorium on data centers receiving tax breaks, following protests in Chandler and Ahwatukee. In Oregon, House Bill 4084 took effect on June 5, 2026, barring new data centers from the state's largest property tax incentive program until 90 days after the 2027 legislative session adjourns. Oklahoma has moved to impose a moratorium on data centers with an electrical load greater than 100 megawatts until November 1, 2029, and South Dakota has proposed a one-year pause on hyperscale construction and expansion. North Carolina repealed the sales-and-use exemption for electricity used at certified data centers, effective for billing periods starting on or after August 6, 2026, while Virginia imposed a $0.011-per-kilowatt-hour tax on data center electricity through June 2028, capped at $600 million a year.

At the local level, the consequences for officials who back projects have been severe and memorable. In Cascade Locks, Oregon, voters recalled two Port Authority officials in June 2023 for supporting a $100 million data center project from Roadhouse Digital; the new board canceled the project the following month. In Warrenton, Virginia, residents voted out every town council member who had supported Amazon's proposed data center in the November 2024 election. In Independence, Missouri, voters recalled First District Councilman John Perkins on September 1, 2026, with nearly 70% of his district voting to remove him, after he backed tax incentives for a Nebius AI data center. Through early September, Ballotpedia had identified 42 recall efforts targeting 79 officials related to data center projects in 2026. For a local politician anywhere in America, the lesson is legible: approving a data center is now a career-risk vote.

Cyclical or Structural? This Is a Regime Shift

The critical question for investors and executives is whether this is a cyclical wave of sentiment that will recede as the technology proves its benefits, or a structural change in the political environment for AI. The evidence points to structural.

A cyclical backlash requires a short-term trigger and a mean-reverting mechanism. The triggers here — job anxiety, utility bills, water use, noise — are not short-term. They are the direct, durable consequences of the technology's physical footprint, and they intensify as the buildout accelerates. Goldman Sachs Research forecasts that global power demand from data centers will rise 50% by 2027 and as much as 165% by 2030 compared with 2023 levels, meaning the very thing that provokes opposition is on a steep growth path. There is no mean reversion baked into that trajectory.

Nor does this fit the pattern of past tech backlashes that faded as adoption grew. Social media's "techlash" was about content and privacy — issues that users could individually ignore while continuing to scroll. Data centers cannot be ignored by the communities that host them. The opposition is organized, too: at least 142 activist groups across 24 states are now coordinating against data center construction, and more than 300 state-level data center bills were filed across 30-plus states in a six-week span in 2026. That is the infrastructure of a movement, not a mood.

The strongest evidence for the structural read is the asymmetry between the speed of deployment and the speed of consent. AI companies can break ground on a gigawatt-scale campus in months; earning local permission takes years, if it happens at all. That mismatch does not self-correct. It compounds, because every blocked project makes the next proposal more politically costly, and every recalled official teaches the next one to hesitate.

The Second-Order Effect: Consent Becomes a Cost of Capital

The first-order effect of anti-AI politics is obvious: some projects get blocked. The second-order effect is more consequential and less discussed. Political consent is becoming a priced input in AI infrastructure finance. A project with contested local approval carries longer timelines, more litigation risk, and higher uncertainty around power and water access. Lenders and equity investors price that. The cost of capital for AI infrastructure rises even for projects that ultimately proceed.

That dynamic favors incumbents. Hyperscalers with billion-dollar balance sheets, established utility relationships, and federal lobbying operations can absorb delays and legal fights that would kill smaller developers. The likely result is a more concentrated industry — the opposite of what a populist movement intends. It also creates a fault line inside the AI coalition itself. Chipmakers such as Nvidia and Broadcom want data centers built as fast as possible; utilities and data center REITs want predictable, permitted sites; hyperscalers want to deploy capital without political friction. Anti-AI politics does not threaten all of them equally, and it will not take long for the industry to start sorting itself into winners and losers of the new political environment.

There is also a geographic arbitrage emerging. States that lean into the backlash — Oregon, Arizona, Oklahoma — are signaling that they will not subsidize the AI buildout on the old terms. States that offer streamlined permitting and stable power may capture investment, but they will face their own local fights. The map of AI infrastructure in 2030 will be drawn as much by local politics as by power prices.

The Counter-Argument, Taken Seriously

The strongest case against the "anti-AI politics is getting big" thesis is that the backlash is NIMBY, not anti-AI — and that Americans are still voting for the technology with their behavior. Chatbot usage has climbed from 33% of U.S. adults in 2024 to 49% in 2026. People object to the data center down the road while continuing to use the services it powers. On this reading, the movement has a coherent enemy only at the local level; it lacks a unified federal agenda, and its bipartisan coalition could fracture the moment the issue reaches national policy.

That fracture is already being tested in Washington. The Great American AI Act, a discussion draft released in June 2026 by Representatives Jay Obernolte (R-California) and Lori Trahan (D-Massachusetts), would preempt state laws "specifically regulating the development" of AI models for three years, with a sunset in December 2029. A related bill, H.R. 5388, would impose a temporary moratorium preempting certain state laws that restrict AI models and systems in interstate commerce. More than 200 state lawmakers have called on Congress to oppose the preemption proposal, arguing it would override state protections without a federal framework to replace them.

If federal preemption passes, it could neuter much of the state-level momentum and reassert the industry's preferred regulatory terrain. The counter-thesis holds that the movement's victories are local, reversible, and ultimately subordinate to federal law and to the economic gravity of the AI race with China.

But even this counter-argument concedes the central point: the politics of AI have moved from the periphery to the center of legislative combat. Preemption is not a dismissal of the backlash; it is a response to it. And preemption fights are slow, litigious, and uncertain — precisely the conditions that raise the cost of capital in the meantime.

What Comes Next: The Signals to Watch

The thesis that anti-AI politics is about to get big will be tested in the 2026 midterm elections and in the federal preemption fight. The falsifying signal is specific: if the Great American AI Act or H.R. 5388 clears Congress before the November 2026 elections, and if anti-data-center candidates fail to flip more than a handful of governorships and swing House seats, then the movement's reach has been overstated, and the backlash remains a local nuisance rather than a national force.

By time horizon, the outlook splits. In the short term — through the midterms — expect more candidates to adopt anti-data-center positions, more ballot measures, and more recalls. The Oregon Data Center Advisory Committee's final report, due in October 2026, will set the template for other states. In the medium term, the industry will adapt: more off-grid power, more behind-the-meter generation, more lobbying, and more site selection in politically favorable jurisdictions. In the long term, the structural question is whether AI can rebuild a social license — whether the industry can demonstrate local benefits (jobs, tax revenue, grid investment) fast enough to offset the concentrated costs it imposes.

The base case is a slower, more expensive, more politically mediated AI buildout, with capital concentrating in the hands of the players best equipped to navigate it. The upside case for the industry is that preemption passes, local opposition is contained, and the buildout accelerates on the strength of demand. The downside case is that the movement wins the midterms in key states, preemption fails, and the political risk premium on AI infrastructure becomes a permanent feature of the sector's valuation.

For all the billions being spent on chips and power, the scarcest resource in the AI race may now be consent. The industry that taught the world to move fast and break things is learning that some of the things it breaks are political — and those do not get fixed by the next software update.

Explore more exclusive insights at nextfin.ai.