- Monday Momentum
- Posts
- AI Just Lost Its First Election
AI Just Lost Its First Election
How data centers became a ballot issue that voters rejected almost everywhere it appeared, and why the binding constraint on AI may no longer be chips or capital, but consent
Justin Wright
November 09, 2026 • Est. Reading Time: 6 minutes
Happy Monday!
Voters rejected data center development in all but one of the 41 ballot measures on the issue, according to Redbud Consulting's tally. (Source: Al Jazeera)
For most of this year, the debate about AI's limits was conducted by labs, investors, and regulators. Would the chips arrive, would the capital keep flowing, or would the models become dangerous? On November 3, a different group weighed in.
Across the country, 41 local and state ballot measures asked voters about data center development. According to a tally by Redbud Consulting reported by Reuters and Al Jazeera, voters sided against data centers in all but one. In Texas, Democrat Gina Hinojosa won the governorship, the first Democrat to do so in more than 30 years, after campaigning on a statewide pause on data center approvals. In South Carolina, Senate candidate Annie Andrews, who called for a two-year pause on new data center projects, beat an incumbent she nicknamed "Data Center Darline."
Two days later, the market caught up. On November 5, the Philadelphia Semiconductor Index had its worst day since 2020. By the end of the week, the valuation talk for Anthropic's IPO had fallen to roughly $1.2 to $1.3 trillion, and OpenAI had cut its fundraising target by a third. The AI industry has spent the year arguing about extinction risk and pacing. Voters turned out to care about something more immediate: their electricity bills.
Voters rejected data center development in 40 of 41 ballot measures on November 3, and Texas elected a governor who pledged a data center moratorium in a state holding about 27% of planned US data center capacity. Pro-AI super PACs largely failed to change outcomes. Chip stocks then had their worst day since 2020, Anthropic's IPO valuation talk fell to about $1.2 trillion from $2 trillion in September, and OpenAI cut its raise. The constraint on AI has moved from chips to capital to public consent.
What the Vote Showed
The results were broad, local, and often bipartisan in character, even where the candidates who benefited were Democrats.
Data Centers on the 2026 Ballot
Signal | Result |
|---|---|
Local and state ballot measures | Data centers lost 40 of 41, per Redbud's tally |
Texas governor | Hinojosa (D) wins on a data center moratorium pledge |
Texas's share of planned capacity | About 27% of planned US data center gigawatts |
South Carolina Senate | "Data Center Darline" loses to a pause advocate |
Voter priority | 56% called data centers a crucial or major factor |
Pro-AI super PACs | Little measurable effect on general elections |
The ballot measures were mostly about zoning, review requirements, and local restrictions, and Ohio alone accounted for about half of them. The lone exception, in Harwood, North Dakota, saw voters reject a review requirement for large industrial projects. Everywhere else, the side asking for more control over data centers won.
The polling points the same way. A post-election poll found 56% of voters called data centers a crucial or major factor in their decision, though outlets attribute the figure to different pollsters, so treat the exact number with some caution. In a Quinnipiac poll, 72% of people who opposed a local data center cited electricity costs as a reason.
The money did not save the industry either. Leading the Future, the pro-AI super PAC network that launched with more than $100 million, spent only about $7.4 million on general election races, according to Fortune, and lost several high-profile ones. Time concluded that public skepticism of AI made the industry's money a liability for some of the candidates who received it. Fast Company offered a more measured read, noting AI-backed PACs still helped elect dozens of candidates in both parties.
It Was the Bills
The backlash was not mainly about superintelligence, job loss, or chatbot safety, it was about electricity prices.
Residential power prices have risen more than 10% since 2024, according to reporting by Forbes. A Federal Reserve Bank of Dallas working paper estimated that data centers have pushed wholesale electricity prices up 6.9%. That is a cost voters see every month, and data centers became a visible, local symbol for it.
This should not surprise anyone who has followed the year. In July, New York became the first state to pause new hyperscale data centers, citing residential electricity prices up nearly 68% since 2019. In August, Pew found 52% of Americans were more concerned than excited about AI. Even Texas's outgoing Republican governor, Greg Abbott, had already frozen permitting and grid connections for new data centers during an audit. The election did not create the backlash, it brought into light something that already existed.
There is an important caveat. Electricity prices are also rising because of the Iran war and tariffs, so data centers are not the only cause. But politics does not run on precise attribution. A new building the size of several warehouses that draws as much power as a city is easy to see, and easy to blame.
Then the Market Repriced
The financial reaction came fast. On November 4, AI infrastructure stocks fell on the Texas result. Wells Fargo estimates that roughly 27% of planned US data center gigawatts sit in the state, and analysts now expect its existing freeze to extend well into 2027.
On November 5, the selloff spread. The Philadelphia Semiconductor Index fell about 12%, its worst day since 2020. The Nasdaq 100 lost roughly $2.3 trillion in value, and AMD and Micron each fell around 17%. Weak results from Qualcomm and existing worries about AI spending contributed, so the election was one cause among several. Chips partially rebounded the next day.
Private valuations followed. Bloomberg reported that potential buyers of Anthropic's IPO were discussing a range of $1.2 to $1.3 trillion, down from talk of $2 trillion in September and $1.5 trillion in October. Prediction markets put the odds of a listing before December near 50%, down from about 80%. OpenAI cut its fundraising target from $30 billion to $20 billion and its proposed valuation from $1.4 trillion to $1.25 trillion, closing an initial $11 billion.
Chips, Then Capital, Then Consent
This newsletter has tracked a moving constraint all year. Early on, the question was whether there would be enough chips. By August, it was capital, as Nvidia and Wall Street built a $500 billion financing machine and lenders began charging AI companies more for risk. Last week's vote suggests a third constraint that may matter most: whether communities will agree to host the buildout and pay for the power it consumes.
Chips can be manufactured and capital can be raised. Consent has to be earned, one county, utility commission, and statehouse at a time. That changes the economics of every AI company with large compute commitments, including the $518 billion in largely non-cancelable infrastructure contracts in Anthropic's prospectus. A data center that cannot get a permit, a grid connection, or political support does not get built on schedule, no matter how much money stands behind it.
What This Means for Practitioners
For enterprise buyers, expect regional variation in AI capacity and pricing. If new data centers face delays in Texas, Ohio, and elsewhere, available compute could tighten in some regions even as overall demand grows. Ask your vendors where their capacity is coming from.
For the AI industry, the lesson is that public relations will not fix a power bill. Companies that pay their full share of grid upgrades, bring their own generation, or lower local rates will have a much easier path than those relying on tax breaks and super PAC spending.
For investors, political risk is now a line item. The Texas result alone moved stocks. Siting, permitting, and local politics deserve the same scrutiny as chip supply and revenue growth.
The Bottom Line
The AI industry spent 2026 debating existential risk, pacing proposals, and voluntary safety accords. Voters, given their first real chance to weigh in, voted on something more concrete: whether they want the buildout next door and on their electricity bill. Almost everywhere the question was asked, the answer was no.
That does not end the buildout. Demand for AI is real, and data centers will keep getting built where communities welcome them. But the era when the main obstacles to AI were technical and financial is ending. The next obstacle is political, and it will be resolved in county meetings and statehouses, not in data centers or boardrooms.
In motion,
Justin Wright
If the AI buildout depends on communities agreeing to host data centers and pay higher electricity prices, and voters just rejected that bargain almost everywhere it was on the ballot, who should pay for the power AI needs, and what would a deal that voters would actually accept look like?
Builder's Note: I have spent most of this year writing about AI as a technology and a business. This election is a reminder that it is also infrastructure, and infrastructure needs public permission. As someone who builds on these models every day, I want more compute, not less. But I would rather the industry earn that permission by paying its own way on power than keep losing votes it did not think it needed to win.
How the US midterm elections reflected resistance to data centres - Al Jazeera
Data Center Ballot Measures: What Voters Decided Nov. 3 - Expert Beacon
Under Hinojosa, Texas data center approvals could stay paused - The Texas Tribune
AI data center stocks and Musk's SpaceX decline as Texas governor result raises risks of a slowdown - CNBC
Here's how the candidates backed by AI super PACs fared - Fortune
The midterms weren't the rebuke of AI money many expected - Fast Company
Midterm Voters Just Pulled The Plug On AI Data Centers - Forbes
Global stock rout deepens as AI data centre fears fuel worst US chip sell-off since 2020 - Euronews
Anthropic IPO Valuation Could Fall to $1 Trillion, May Delay - Benzinga
OpenAI cuts fundraising valuation target to $1.25 trillion after investor pushback - Yahoo Finance
Quick Hits
Google is reportedly testing an "AI Contribution" pilot that pays selected publishers for content used in Gemini and AI Overviews, a week after AI Mode's lower monetization sank Alphabet's stock. (Hello Growth)
Iren announced delays to AI capacity deliveries for Microsoft alongside a $15 billion capital spending plan, another sign the buildout is running into practical limits. (AI Tool Daily)
Nvidia reports earnings on November 17, the first major test of AI demand since the election selloff. (top1markets)
If you haven’t listened to my podcast Mostly Humans: An AI and business podcast for everyone yet, new episodes drop every week!
Episodes can be found below - please like, subscribe, and comment!