Product1 distinct publisher3 min readUpdated
Washington Post data has AI and data centers outranking crypto, Israel and manufacturing on candidate websites. Siting and power decisions now carry electoral risk.
The Product Desk · Product desk

Compiled by The Product DeskSomething wrong?How this is made
Candidates in 40 percent of current U.S. races now list AI and data centers as issues on their campaign websites, according to Washington Post data reported by Gizmodo [1]. On those sites the topic outranks cryptocurrencies, Israel, manufacturing, and race and racism, and sits just below guns and gun violence [2][3], which means the siting, interconnect, and deployment decisions inside an AI roadmap are now inputs to somebody else's campaign.
The Post also found AI has become an issue in 90 percent of U.S. states [4]. The framing splits hard by party. Among candidates who raise AI, Democrats account for 90 percent of mentions of broad AI harms [5], 84 percent of mentions of data centers [6], 77 percent of mentions of police and military use [7], 76 percent of child safety mentions [8], 72 percent of jobs mentions [9], and 70 percent of mentions of broad AI benefits [10]. Republicans dominate one subtopic: China and national security, at 76 percent versus 24 percent for Democrats [11]. Of the seven subtopics reported, six are majority-Democratic and only the China frame is majority-Republican [12].
That asymmetry is the operational detail. A national-security frame is permissive about building; a harms-and-data-centers frame is about who approves what, where. Both can be true at once, in the same country, which is how a company ends up with federal tailwinds and a hostile township.
Michigan Senate nominee Abdul El-Sayed is the clearest example on the Democratic side, with a policy section titled "First, Do No Harm: AI Under Democracy" that lists specific harm-prevention actions plus proposals for democratic control and private ownership, and a separate policy document on data centers [13]. On August 1, 2026 he posted that he had stood with local and state elected officials to say no more AI data centers should be approved until federal guardrails protect communities like Saline Township "from being run over by the likes of DTE, Oracle, and OpenAI" [14]. Naming an incumbent utility and two named vendors is a different risk class than a zoning hearing.
The downside for incumbents is also visible. J. Stuart Adams, president of the Utah State Senate, moved to revamp his image on data centers in the closing days of his campaign and then unexpectedly lost last month [15]. Late repositioning did not work.
The speed matters more than the level. A Heatmap Pro survey found public opinion on data centers swung 49 points over nine months [16], an average of roughly 5.4 points a month [17]. The source does not state the direction of that swing [16], though it is cited in the context of backlash. Either way, a 49-point move inside three quarters is faster than most site-selection and power-procurement cycles, and faster than the political intelligence most product organisations maintain. Nine months ago the register was Senator Bernie Sanders filming himself interrogating a smartphone running Claude and afterwards saying he had spoken to an "AI agent" [18].
Three things to watch. Whether the partisan split holds after the general, because a durable China frame keeps federal permitting friendly while local approvals get harder. Whether website language converts into actual moratoria, as El-Sayed has demanded [14]. And whether more incumbents follow Adams and lose on this specific issue [15], which is the signal that decides how loudly the next state senate president negotiates over a substation.
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
J. Stuart Adams, president of the Utah State Senate, quickly sought to revamp his image with regard to data centers in the waning days of his campaign before unexpectedly losing last month.
A recent Heatmap Pro survey found that over the course of nine months, public opinion on data centers swung by 49 points; the reported figure does not specify the direction of the swing.
According to new Washington Post data, 40% of current races in the U.S. involve candidates who list AI and data centers as issues on their websites.
In terms of presence on candidate websites, the Post found AI and data centers are more important than cryptocurrencies, Israel, manufacturing, or race and racism.
Guns and gun violence rank just a tiny bit higher than AI and data centers on candidate websites.
The Post found AI has become an issue in 90 percent of U.S. states.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Single publisher relaying two unpublished datasets
Every quantitative claim traces to one article that summarizes a Washington Post analysis and a Heatmap Pro survey without reproducing sample sizes, coding rules, dates, or the direction of the 49-point swing. The only primary artifact in the cluster is El-Sayed's dated campaign post; the Adams example is asserted without electoral data. Internally the figures are specific and mutually consistent, which supports a middling rather than low score.
No adoption evidence in cluster
The cluster reports political salience on campaign websites, not technology deployment, procurement, usage, or market uptake. No release, deployment, benchmark, pricing, licensing or usage disclosure is present, so adoption cannot be measured without inferring facts the sources do not contain.
Framing outruns the demonstrated consequences
The measured salience figures are plausibly reported, but the article's leap to AI as a 'decisive, life-or-death political issue' and the cluster framing that roadmaps have not priced this in rest on one unexplained loss and an undirected 49-point survey figure. Website mentions are evidence of rhetoric, not of enacted policy, blocked permits, or changed capex, so the claims are modestly overstated relative to what the cluster demonstrates.
Campaign self-presentation relayed by ad-supported tech media
The underlying material is largely candidate-controlled: campaign websites and a candidate's own post naming DTE, Oracle and OpenAI, all produced to win votes. Those are relayed by a consumer tech publisher whose framing favors striking comparisons (AI outranking racism or Israel) and a horror-movie lede. No named company or survey author is quoted in reply, so competing incentives are unrepresented.
Directionally credible, thinly sourced
One publisher, no methodology for either dataset, and no corroboration of the causal story cap confidence. Confidence is not lower because the reported statistics are specific and internally consistent, one claim is anchored to a dated primary artifact, and the derived arithmetic is straightforward.
product
A school agenda shipped with "Vitoiis" and a planet named Marc, and no one read it first1 distinct publisher
science
Text watermarks land on 2 December. The detection they imply does not.1 distinct publisher
build
The $559M-versus-$12.3B quarter matters more than the $65B run rate4 distinct publishers
science
Claude's watermark is a compliance artefact, not a cheating detector1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 16, 2026