Voters Rejected AI Data Centers in 40 of 41 Local Elections
On November 3, voters in 40 of 41 local elections backed limits on data centers, tightening the siting options behind a $720 billion hyperscaler spending surge.
Why it matters
- On November 3, voters in 40 of 41 local elections favored data center restrictions, per Ballotpedia.
- Amazon, Alphabet, Microsoft and Meta plan a combined $720–740 billion in 2026 capital spending.
- AWS reported a $578 billion backlog; Microsoft's commercial RPO rose 73% to $677 billion.
- The Energy Department projects data centers will use up to 12% of U.S. electricity by 2028.
- Amazon announced a $1 billion Built Together investment in U.S. data center communities.
Voters in 40 of 41 local elections on November 3 chose to restrict data center construction near their communities. Ballotpedia counted 27 measures approving restrictions or expressing opposition and 13 pro-development measures defeated. Only Harwood, North Dakota broke the pattern. The tally included five outright bans, 12 partial bans, and three temporary moratoriums.
The stakes are enormous. Amazon, Alphabet, Microsoft and Meta together plan $720 billion to $740 billion in capital spending this year alone — a run rate approaching the trillion-dollar mark Wall Street uses to describe the AI buildout. For months, the market debate focused on whether big tech can afford these plans. November 3 reframed the question: will towns allow the buildings at all? The complaints voters cited were electricity bills, water use, land and strained roads, and the opposition crossed party lines.
The mechanism here is simple. AI demand is real and contracted, but it lands somewhere specific — on a local grid, next to a local road. Siting now depends on permits, and permits favor companies that already have them.
How much are the hyperscalers actually spending?
The four biggest spenders have laid out their 2026 plans in hard numbers:
- Amazon (NASDAQ:AMZN) expects approximately $220 billion in cash capital spending for 2026.
- Alphabet (NASDAQ:GOOGL) guided to $195 billion to $205 billion.
- Microsoft (NASDAQ:MSFT) expects approximately $170 billion for the calendar year.
- Meta Platforms (NASDAQ:META) plans $135 billion to $145 billion.
For scale: back in 2023, Alphabet spent $32.3 billion on capital expenditures. The jump in three years is roughly sixfold for one company.
That spending is straining the balance sheets. Alphabet's free cash flow came in at negative $9.07 billion last quarter, and the company issued $28.7 billion in notes. Meta's free cash flow was negative $2.12 billion. The market has noticed: over the past month, Alphabet shares are down 17.77% and Meta's are down 10.62%. The takeaway: even the richest companies in tech are now borrowing to fund AI infrastructure.
Is the demand actually there?
Yes, and it's signed. AWS reported a $578 billion backlog. Google Cloud reported $561 billion. Microsoft's commercial remaining performance obligations rose 73% to $677 billion. Amazon CEO Andy Jassy said big customers reserve "hundreds of megawatts or sometimes up to a gigawatt of capacity" in advance.
But those megawatts have to land somewhere, and the local cost is becoming visible. The Energy Department projects data centers will use up to 12% of U.S. electricity by 2028. One Denver campus may draw up to 805,000 gallons of water a day.
Amazon's response is telling. Its earnings call had no specific management commentary on local opposition or permitting — yet the company announced a $1 billion Built Together investment in U.S. data center communities. A $1 billion line item tells you how much local goodwill now costs.
Who actually gets hurt when towns say no?
Not the hyperscalers, mostly. All four can move or postpone projects. Smaller developers often can't. For a project funded with debt, a 12-month delay can significantly change the economics. The result is a widening gap: developers that already hold power, permits and financing gain ground on those still waiting to break ground.
The big players are already building workarounds. Microsoft cut dock-to-live times for new GB300s by about 20%, which squeezes more capacity out of sites it already operates. Alphabet is placing its AI chips in partner and customer facilities — a move Google Cloud chief Thomas Kurian said "offsets the need for us to have to prebuild facilities."
The pattern matters beyond these four companies. When the biggest spenders can route around local vetoes and the smallest can't, voter resistance doesn't stop the boom — it concentrates it.
What signals decide 2027?
The companies themselves are signaling continued acceleration. Amazon plans to spend "meaningfully more in 2027," and Alphabet expects capex to "increase significantly in 2027."
When the four report fourth-quarter results, the number to watch is their 2027 capex guidance. Two other indicators matter just as much:
- Whether local permitting shows up in their risk disclosures.
- Microsoft's third-party capacity, already delayed to the second half of the year — another slip would show the vote box reaching the income statement.
For the four hyperscalers, voter opposition means slower and costlier siting. The real trouble falls on developers that have neither power nor permits.
Source: 247wallst.com
More from Sophie Lindqvist
Show full bio
Staff writer covering regulation, policy and their impact on artificial intelligence and machine learning at AI In Context.
611 articles
Related articles
- Data Centers Lost the Midterms: Voters Rejected AI's Building Boom
- Data Center Backlash Becomes a Political Force After Midterms
- Senate Votes 73-25 to Open Debate on Bill Making AI Data Centers Pay Their Own Power Bills
- OpenAI will keep funding AI data centers despite voter rebuke, CFO says
- Senate takes up permitting bill that would make AI data centers pay for their power