The fight over America’s AI buildout is no longer just about zoning, water rights, or power rates; it has become a contest over narrative terrain, where influence operations seek to magnify local anxieties into national headwinds that can slow a strategic technology race.
At a Glance
- Senior national-security officials warn that foreign adversaries are amplifying anti–AI and anti–data center sentiment to impede U.S. AI infrastructure growth.
- The strongest public evidence points to amplification and covert content generation—not to foreign creation of the broader grassroots backlash.
- Genuine local concerns—land use, water, grid strain, and cost—predate the foreign-influence frame and animate much of the opposition.
- The strategic stakes are real: data centers are the industrial base for AI capability; delays can translate into lost competitive advantage and resilience.
What’s being claimed—and what the evidence actually shows
U.S. national-security voices have increasingly described a pattern: foreign adversaries exploit open online ecosystems to push or reinforce narratives that make it harder to site and build data centers and power infrastructure. Emil Michael, the Defense Department’s Under Secretary for Research and Engineering, has said adversaries are running propaganda and “psyop” campaigns to turn the U.S. public against data centers and AI, explicitly tying information tactics to the AI readiness race. Other officials and political leaders have echoed that framing, pointing to foreign-directed propaganda, dark-money vectors, and malign influence that converge where local permitting fights are tightest.
Where the public record is strongest is on amplification, not origination. Open-source researchers and reporters who have surveyed the online landscape describe state-aligned outlets and covert social accounts boosting anti–data center talking points and, in some instances, fabricating content—tactics consistent with known playbooks in energy and technology debates. Analysts, however, have not produced definitive, at-scale evidence that such efforts created the opposition movement or materially redirected broad U.S. public opinion; the weight of expert commentary remains that the core backlash is organic, with foreign actors riding existing currents.
How the information tactic works: amplify the friction that already exists
Psychological operations in the modern sense rarely invent a grievance from whole cloth; they identify fissures already under stress and apply repetition, velocity, and targeted reach. Data centers offer many such seams. They are land- and capital-intensive, require large and often time-shifted electricity loads, and in many regions draw on strained water systems for cooling. Communities weigh property-tax revenue and jobs against truck traffic, substation expansions, line upgrades, and noise. Those costs and tradeoffs are real, and they provide credible scaffolding for any message that seeks to slow, stall, or splinter coalitions. When foreign amplifiers enter that information space—by recirculating imagery of large facilities near residential areas, overclaiming grid impacts, or seeding doubt about national benefits—they do not need to persuade a majority nationwide; they need only tilt a handful of tightly contested local processes across dozens of jurisdictions to inject drag into the national buildout.
That is precisely why officials connect the dots to national strategy. Compute capacity is the industrial base of modern AI. The marginal data hall that does not get built by 2027 is the missing training run in 2028, the delayed deployment in 2029, the brittle logistics node in 2030. The competition is cumulative; delays compound, especially where supply chains for transformers, high-voltage equipment, and skilled labor are already constrained.
Where the skepticism is well-aimed
Healthy skepticism focuses on two claims that often get blurred in public rhetoric: first, whether foreign actors are present in the information ecosystem at all; second, whether they are determinative. On the first, multiple journalistic and analytical accounts acknowledge foreign-linked content and amplification in English-language channels around AI and data centers. On the second, experts caution against conflating “detectable activity” with “decisive effect.” Graphika’s analyst Dina Sadek, for example, told WIRED the firm had not seen organized, scaled influence operations driving the debate, with limited exceptions; the broader opposition appears to draw chiefly from local, material concerns over land, water, and rates. PolitiFact’s shorthand is similar: claims about sweeping causality are exaggerated relative to the evidence of impact on public opinion.
That distinction matters. Overstating causation can delegitimize legitimate community objections, alienate stakeholders needed for siting compromises, and hand adversaries a propaganda gift—portraying the U.S. as dismissive of citizen voice. Understating the amplification risk, however, ignores a clear lesson of the last decade: targeted narratives can shape the tempo and texture of regulatory and permitting fights even when they do not convert broad national majorities.
The policy problem hiding in plain sight: infrastructure governance by a thousand cuts
America’s compute expansion now depends on a governance system designed for a different era—fragmented authorities, sequential permits, contested environmental reviews, and infrastructure upgrades planned on decadal cycles. That makes the system exquisitely sensitive to narrative shocks. A viral claim about spiking electricity bills in one county becomes a reference point for hearings across a region. A doctored image of cooling towers circulates faster than a utility’s 10-year integrated resource plan can be explained, let alone updated. Even when planners do the work—on heat-reuse, non-potable water, substation siting, and time-of-use load shaping—the capacity to communicate those tradeoffs at scale lags. Foreign amplifiers exploit that asymmetry. Skeptics are right that they did not invent the asymmetry; officials are right that adversaries can turn it to strategic effect.
This is also why the “data center” label itself is doing too much work. Some facilities are training clusters with peaky, high-density loads; others are inference farms optimized for steady-state services. Curtailment flex, on-site generation, thermal storage, direct-to-transmission interconnects, and waste-heat districting are not abstractions—they are engineering levers that change the local equation. Communities respond differently when they see specific mitigations tied to enforceable conditions.
Implications: defend the discourse, earn the permit
If the goal is to build fast and smart without sacrificing legitimacy, two lines of effort matter. First, harden the information environment. That does not mean policing dissent; it means investing in credible, preemptive, and locally specific communication—on water sourcing, plume abatement, traffic, and rate design—and equipping trusted intermediaries (co-ops, municipal utilities, county staff) to answer with facts at the speed of narrative. Federal and state partners can support with standardized disclosures and model agreements that surface mitigations up front. Doing so reduces the surface area for exploitation without collapsing debate into “for or against AI.”
Second, upgrade the siting and interconnection machinery. Co-locate with brownfields and existing substations where public acceptance is higher; use enforceable community-benefit agreements that tie tax receipts to visible amenities; preference designs with measurable flexibility so operators can ramp down during peak residential demand without service penalties. The more tangible the local upside and the tighter the operating envelope, the less oxygen there is for outside amplification to take root.
Bottom line: a real influence risk layered onto a real local debate
The strongest claims—that foreign actors are the prime movers of U.S. anti–data center sentiment—are not substantiated by the open record. The most defensible claims—that adversaries are present, probing, and amplifying friction to slow U.S. AI infrastructure—are supported both by officials’ warnings and by what we know of modern information operations. Treat the movement as fundamentally domestic in origin, but not hermetically sealed. Then design policy and project practice for that mixed reality: neutralize manipulation by earning trust on the substance.
Sources:
foxnews.com, thehill.com, cnbc.com, politifact.com, npr.org, finance-commerce.com
3/ For months, Republicans have pushed this stupidity.
Trump: "a lot of people say it's China."
Burgum blames "foreign psyops campaigns."
Scalise says they're investigating.
Sen. Cotton asked DOJ to investigate foreign influence targeting AI infrastructure.— TJHSST ⚔️ Defeating #DarkMoney (@TJHSSTParents) October 6, 2026
This is a header
This is a paragraph.