Voters favored limits on data-center development in 40 of 41 local ballot contests on Nov. 3, 2026, per Ballotpedia’s certified-style tally published Nov. 6—the sole exception was Harwood, North Dakota, which rejected an added industrial-review step that would have constrained projects including hyperscale campuses. The revised count upgrades Culled’s election-night floor of 33 passing or leading measures and helps explain why AI and semiconductor stocks sold off sharply into the weekend—the binding constraint is LABOR and community politics at the permit layer, not only federal oversight after Democrats seized Congress.
Congress can talk about a national moratorium; counties can cap megawatts. By the time Ballotpedia closed its Nov. 6 rollup, the local map had hardened into near unanimity—enough for analyst capex models to treat referendum risk as a feature of every greenfield site, not an Ohio outlier. The residual test is throughput: whether canceled projects rise faster than ballots multiply.
Columbus, Nov. 8 — Tuesday’s counts are now almost complete, and the pattern Culled flagged on election night hardened into arithmetic: voters favored limiting data-center development in 40 of 41 local contests on Nov. 3, according to Ballotpedia’s Nov. 6 rollup. The lone break was Harwood, North Dakota, where residents defeated a measure that would have added review requirements for certain industrial developments, data centers included—a procedural loss for restriction advocates, not a vote to invite hyperscale load onto a village of fewer than 800 people.
That 40-of-41 score matters because it upgrades an earlier Stateline-era floor of 33 measures passing or leading with a denominator tied to Ballotpedia and Data Centers on the Ballot tracking: 41 contests across 10 states on the general-election ballot, Ohio alone carrying 21. Investors who treated Ohio as the stress test now inherit a national permit map where rejection of pro-development measures counts the same as approval of bans, moratoriums, and acreage triggers.
What voters actually passed
Ballotpedia sorts the wins into buckets that capex models rarely encode. Voters approved 27 measures that limited or opposed development outright, including five full bans—California, Florida, and three in Ohio—plus 12 partial bans, three six-month moratoriums in Maine, zoning rules in Adams County, Ohio, non-binding opposition language in Illinois, and enforceable voter-approval gates such as Grove City, Ohio, Issue 28, which ties large power and water footprints to ballot consent.
They also defeated 13 measures that would have favored sites: Wilmington, Ohio, lost four rezoning attempts; Illinois saw five non-binding “allow development” advisories fail; a Boulder City, Nevada, parcel-use change fell. Adding defeats of pro-buildout items to wins for curbs is how you reach 40 restrictive outcomes from 41 races without ideological symmetry—Republican exurbs and Democratic townships share substation anxiety.
The platform constraint is no longer a zoning hearing; it is copy-paste charter language that travels county to county.
Markets read permits as duration
CNBC’s Nov. 6 wrap tied the tally to a Thursday rout in AI and chip equities—Nvidia down 9.3%, AMD 17.3%, Micron 17.5% in one session—as analysts labeled the buildout a “war of attrition” where scarce power favors the deepest balance sheets. Barclays credit strategist Dominique Toublan argued bipartisan local pushback makes the cycle pricier, not necessarily dead; Goldman Sachs still pencils roughly $7.6 trillion of AI capex from 2026 through 2031. The ballot map pressures duration—the clearing window between grid connection, water rights, and return on megawatt spend.
Congressional Democrats’ blue sweep added a federal narrative CNBC and desk notes linked to national-moratorium chatter, but no statute text followed over the weekend. STATE governors had already moved: Texas and Pennsylvania executives campaigned or ordered against rural siting and tax breaks before voters marked “yes” on local curbs. That layering—federal headline risk plus county enforceability—is what separates this week’s multiple compression from a routine rotation.
Falsifiers for investors
The falsifier is project flow. Reuters-quoted analysts noted restrictive votes may redirect rather than halt campuses; if 2027 interconnection queues in permissive states absorb Ohio’s redirected pipeline, ballots stay politically loud but operationally marginal. If International Data Center Authority-style counts of canceled or stalled projects rise past $170 billion of announced spend blocked since 2024, the 40-of-41 result becomes a recurring tax on greenfield IRR.
Harwood reminds that unanimity is not eternity: one procedural win for industry does not repeal hundreds of local moratoriums already on the books. It does prove that copy-paste opposition can miss a town that fears reviews more than data halls. For the PLATFORM layer, the working assumption after Ballotpedia’s close is simpler: assume referendum risk on the next transformer, not only on the next Fed minutes.
Washington can schedule hearings; Sunbury, Grove City, and Maine moratorium towns already scheduled the buildout’s politics.
Continue reading
Sources
Ballotpedia News Nov. 6 on 40 of 41 contests; Ballotpedia tracker page; CNBC Nov. 4 and Nov. 6 market wraps; Culled Nov. 4 ballot coverage.
