What are data center ballot measures? Data center ballot measures include binding referendums and nonbinding advisory questions about development, restrictions or public support. Their legal effects differ: some can change an approval or law, while advisory questions express voter preferences for officials to consider. On November 3, 2026, voters decide 41 of them. For anyone underwriting a data center site, data center ballot measures are a specific kind of entitlement risk: one whose effect depends on the measure, local law, approval history and any remaining administrative or judicial proceedings. For broader context on site level risk work, see our complete guide on AI real estate due diligence.
Key Takeaways
- Ballotpedia tracks at least 50 local data center ballot measures in 13 states in 2026, and 41 of those 50 are decided on November 3.
- Ballotpedia classifies all six earlier outcomes as favoring limits rather than development. They include different kinds of measures, including a nonbinding advisory question, and are not a project-approval base rate.
- Ohio alone accounts for 21 of the 50 measures across 18 distinct jurisdictions, making it the single largest concentration of referendum risk.
- Augusta Township, Michigan voters overturned a rezoning their own township board had already approved, which is the result that should change your diligence.
- Review each measure's legal effect, petition and election deadlines, and potential challenges when negotiating entitlement contingencies and deposit terms.
What Is on the Ballot on November 3
According to Ballotpedia's tracking of data center related ballot measures in 2026, voters will decide at least 50 local measures across 13 states this year: Arkansas, California, Florida, Illinois, Kansas, Maine, Michigan, Nevada, New Hampshire, New Jersey, North Dakota, Ohio, and Wisconsin. Of those 50, 41 measures, or 82 percent, are on the November 3 general election ballot, with six decided earlier in the year and three scheduled later. Smart Cities Dive reported on October 29 that the November 3 measures span 10 states.
The geographic concentration matters more than the headline count. Ohio accounts for 21 of the 50 measures, 42 percent of the national total, spread across 18 distinct jurisdictions. Illinois has seven, four of them in Madison County. The remaining 22 sit thinly across 11 states, led by Maine with four and Wisconsin and Kansas with three each.
The measures cut in both directions, which is easy to miss in the coverage. Some would limit or prohibit development or impose new requirements on it; others would allow or facilitate it. Alhambra, California votes on prohibiting data centers inside city limits, Edgewater, Florida on prohibiting new data centers and expansions, and Boulder City, Nevada on whether to permit them in the Eldorado Valley Transfer Area.
The 2026 Scorecard Is 0 for 6, and It Was Not Close
This is the part that should change how you price a site. As of October 30, 2026, voters had decided six of the 50 measures. Ballotpedia's scorecard records zero outcomes favoring development and six limiting it. In every case, a majority of voters favored greater limits, either by approving new restrictions or by rejecting a measure that would have facilitated development.
The margins are the real signal:
- Canterbury, New Hampshire (March 10): a measure to prohibit data centers within the town passed 322 to 63, or 84 percent to 16 percent.
- Port Washington, Wisconsin (April 7): voters approved requiring voter approval before the city creates, adopts or approves a Tax Incremental District with projected base value of at least $10 million or projected project costs exceeding $10 million.
- Monterey Park, California (June 2): Measure NDC, amending the city's land use element to prohibit data centers, passed 10,388 to 1,372, or 88 percent to 12 percent.
- Augusta Township, Michigan (August 4): a referendum to uphold an ordinance rezoning agricultural residential land to general industrial for a proposed data center was defeated 269 to 2,314, or 10 percent to 90 percent.
- Wrightstown, Wisconsin (earlier in 2026): a nonbinding advisory question on committing municipal infrastructure and services to a large scale data center was defeated 151 to 1,033, or 13 percent to 87 percent.
- Nottingham, New Hampshire (September 8): a 12 month moratorium on accepting data center permits passed 1,335 to 143, or 90 percent to 10 percent.
Six local votes are a small, selected and heterogeneous sample, not a forecast of tomorrow's 41 or a probability that a particular project will fail. The margins describe those votes; they do not establish what caused the outcomes or rule out turnout effects. The voting also tracks national survey work: a Reuters/Ipsos poll released in June 2026 found 77 percent of respondents worried data centers would raise their electricity rates, and Brookings noted in August 2026 that these concerns cut across parties. That is national opinion context, not proof of the cause of any local election result.
Why This Is Different From a Moratorium
We have covered legislative restrictions at length, including the New York data center moratorium and the broader wave of state level data center moratorium bills. Those are legislative actions, with procedures that differ from a binding referendum or an advisory vote. The available paths for revision, public participation, negotiated conditions or challenge depend on the jurisdiction and the particular action; none guarantees an approval.
A binding referendum can add a separate voter decision even after a board has acted; an advisory vote does not itself reverse an approval. The Augusta Township result is the cleanest illustration in the dataset. The township board had already passed the rezoning. The developer had its approval. Voters then reversed it by a 90 to 10 margin. If your diligence memo treats a favorable board vote as the end of entitlement risk, Augusta Township shows why the memo needs a jurisdiction-specific check. It does not predict the outcome or legal effect of the Ohio measures.
The Port Washington measure points at a second exposure. It did not itself ban a data center; it required voter approval for qualifying Tax Incremental District actions under its base-value or project-cost thresholds. Where incentives are load bearing in your returns, a procedural measure can impair a deal without ever mentioning data centers.
How to Screen a Site for Referendum Risk
Referendum risk is unusually tractable because it is public, dated, and jurisdiction specific. Practical steps:
- Check the referendum rules first. Determine whether the jurisdiction permits citizen initiative, zoning referendum, or advisory questions, and what signature count and filing window apply. A jurisdiction without the mechanism carries materially different entitlement risk.
- Calendar the petition window against your approval date. Augusta Township's exposure existed because a rezoning could be referred to voters after passage. Have local counsel identify the applicable petition window, the effect of a timely filing, and any other approvals or challenges. Expiry of one window does not clear every entitlement risk.
- Read the neighboring jurisdictions. Ohio's 21 measures sit in 18 jurisdictions and four of Illinois' seven are in Madison County. Use that concentration to investigate the local measures and campaigns. Counts alone do not establish common organizers or predict that opposition will spread.
- Price the contingency instead of describing it. Discuss refundable deposits, extension rights and election contingencies with counsel where referendum exposure could affect closing. The appropriate terms depend on the deal and seller agreement.
- Underwrite the ratepayer politics alongside the land. Electricity costs are a documented public concern, which connects to the large load tariff and ratepayer debate shaping utility proceedings.
If you are building a repeatable screen for this across a pipeline rather than one deal, The AI Consulting Network does exactly this kind of jurisdiction level diligence design.
A Hypothetical Illustration of the Cost
A hypothetical example with stated assumptions, not a market observation. A developer contracts a 150 acre parcel at $60,000 per acre, a $9.0 million price, with a 5 percent deposit of $450,000 that goes nonrefundable the day the township board approves the rezoning, and a 45 day post approval referendum window nobody calendared. If a valid petition suspends or subjects the approval to a vote and voters reject the rezoning under the assumed local procedure, the deposit may be exposed, depending on the contract and applicable law. A negotiated extension or refund contingency may reduce that particular risk but can have a price or other tradeoff. Closing a petition window does not eliminate litigation, later approvals or other closing risks. The 45 days are an example assumption, not a legal deadline for any named jurisdiction.
What to Watch After the Votes Are Counted
Three things matter by week's end. First, whether the 0 for 6 pattern holds across 41 measures, or whether the earlier results reflected small, highly motivated electorates. Second, which measures supporting development pass and what legal or advisory effect each result has. Third, whether Ohio officials propose any state-level response. The number of local measures alone does not establish that preemption will follow.
CRE investors who want help translating these results into a site screening standard can reach out to Avi Hacker, J.D. at The AI Consulting Network.
Frequently Asked Questions
Q: How many data center ballot measures are there in 2026?
A: Ballotpedia tracks at least 50 local data center related ballot measures across 13 states in 2026. Of those, 41 are on the November 3 general election ballot, six were decided earlier in the year, and three are scheduled later.
Q: Have voters approved or rejected data center development so far?
A: As of October 30, 2026, voters had decided six measures, and Ballotpedia classified all six as outcomes favoring limits rather than development. The measures differ in legal effect and include a nonbinding advisory question; the tally is not a project-approval probability. Five had vote shares of roughly 84 to 90 percent on the side opposing development or supporting restrictions.
Q: Can a ballot measure reverse a zoning approval a board already granted?
A: In jurisdictions that permit zoning referendums, yes. Augusta Township, Michigan voters defeated a referendum to uphold a rezoning the township board had already approved, by 10 percent to 90 percent in August 2026.
Q: How should referendum risk change a purchase agreement?
A: Ask local counsel to identify petition and challenge windows, whether a vote is binding, and what approvals remain. Negotiate deposit, extension and termination terms for the specific exposure. Neither board approval nor expiry of a petition window alone clears every entitlement risk.