Pulse24 Original
Local Voters Rejected 46 of 47 Data Center Ballot Measures This Year. Texas Already Froze New Permits.
November 13, 2026
Nearly every local ballot measure on AI data centers failed this year, and Texas quietly paused new permits back in September. Wall Street's reaction over the past two weeks shows that fight is now splitting AI's cash-rich winners from its leveraged builders.
Forty-six of the 47 local ballot measures on AI data center construction that went to a vote this year failed, according to Ballotpedia's tally through November 4. That number alone could be read as local noise. The state carrying the single largest concentration of planned AI infrastructure already acted on the same instinct without needing a ballot at all.
[[IMG1]]
What Changed
Texas Governor Greg Abbott ordered state agencies in August to audit pending data center requests. By September, he had directed the Texas Commission on Environmental Quality to stop issuing final permits for new projects, a pause that has now run more than two months with no confirmed end date before December. Gina Hinojosa, the state's governor-elect, campaigned on a statewide moratorium on new data centers that would leave existing and under-construction sites untouched. Texas isn't acting alone. Ohio's incoming governor has floated a similar moratorium, and Morgan Stanley analysts have flagged her as a material risk to that state's buildout. In Congress, Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez have proposed a nationwide construction freeze until lawmakers add guardrails on electricity costs and community approval, while New Mexico Senator Martin Heinrich's GRID Savings Act would stop utilities from passing grid-upgrade costs for massive AI campuses onto ordinary ratepayers.
Markets reacted fast, then started to stabilize. Two trading days after the election, on Thursday, November 5, the Nasdaq composite shed roughly $2.5 trillion in value in a single session, and the semiconductor ETFs SOXX and SMH each fell 12%. A CNBC poll found 56% of voters called data centers a crucial or major factor in how they voted. Chip stocks have since clawed back part of that drop, which is its own signal: investors appear to be pricing this as a cost and timeline problem rather than an existential one.
Why It Matters
The through-line across Wedbush, Barclays, and Goldman Sachs research notes is the same: this backlash is local and bipartisan, and that combination makes the AI buildout pricier and slower rather than smaller. Barclays' credit strategists, including Dominique Toublan, have been making the case for months that scarce power access favors the biggest balance sheets over smaller, more leveraged builders. Wedbush's Dan Ives now expects hyperscaler capex growth to peak in 2027, a timeline he has pulled forward from his earlier view of 2028 or 2029, largely because permitting friction is compressing the runway for new projects.
That reshuffling is already visible in which companies are being named as exposed and which as insulated. Oracle carries the most direct Texas risk among the major cloud players: its fiscal first-quarter revenue rose 30% to $19 billion and net income climbed 63% to nearly $4.8 billion, but that result included $1.4 billion in interest expense, and total debt of more than $125 billion now sits well above its $67 billion book value. CoreWeave, whose own earnings call this month was spent largely defending its balance sheet to bondholders rather than fielding questions from stock analysts, has no legacy revenue base to fall back on if financing gets more expensive. Alphabet, by contrast, keeps coming up in these notes as the name best positioned to absorb permitting delays without touching its balance sheet. The financing side of this story is already showing up in credit markets, where a single Texas data center bond helped push high-yield spreads to their widest since May 2025.
What To Watch Next
The Texas Commission on Environmental Quality's permit freeze is due for some kind of resolution before December, and whatever Hinojosa does with her moratorium proposal once she takes office will set a template other states may copy. Ohio's governor-elect will face a similar early test. Congress returning under full Democratic control raises the odds that Heinrich's grid-cost bill, or something like it, gets real floor time, which would matter more for utilities and ratepayer-exposed data center operators than for the hyperscalers themselves. Credit spreads on AI-linked debt are probably the cleanest real-time gauge of how investors are actually pricing this, more useful than any single stock's daily move.
The Pulse24 Take
None of this looks like the moment the AI buildout stalls. What it resembles is a sorting mechanism, one that raises costs fastest for whoever is least equipped to absorb them. Permitting friction and local opposition function economically like a tax: diffuse, uneven across states, and easy to miss until it surfaces in a capex guidance cut or a bond spread. Markets fell sharply right after the midterms, then partly recovered, and that pattern reads less like panic than like repricing. Cash-rich hyperscalers can wait out a permit freeze. Highly leveraged neoclouds can't, and that gap is likely to widen before it narrows.
How we read the data
Curious how we get from raw data to a take like this? Our Trader's Toolkit walks through the tools we lean on.
Explore the Toolkit