After the 3 November elections, part of the price of building an AI data center in the US is set by five levers that states and their localities control: permits and zoning, grid and power costs, tax incentives, water, and state land. A summary of a Goldman Sachs note, posted on X by the Walter Bloomberg account and reported by us, said Texas, Georgia and Michigan “could introduce restrictions on tax breaks, electricity, water and environmental permits”. That covers four of the five; the fifth, state land, surfaced in Texas this week.
This explainer maps each lever to who decides it and to the responses builders have made so far, using only our reporting from 3 to 6 November. By our count, seven of the nine named AI builds in that coverage, 78%, sit in Texas, Georgia and Ohio, from OpenAI’s Stargate to Meta and the Tesla-SpaceX Terafab.
Business Pill · Who Pays for the Grid
A short explainer on who pays for the grid upgrades that data centres need. It teaches the general idea only and does not discuss this story.
The key insight: As we read it, the cost of an AI data center in the US is now partly set by five state levers: permits and zoning, grid and power costs, tax incentives, water, and state land. A Goldman summary posted on X named four of them for Texas, Georgia and Michigan, and Texas raised the fifth this week near SpaceX’s Starbase; seven of the nine named builds in our coverage, 78%, sit in Texas, Georgia and Ohio, by our count.
Data center politics in the US runs on three levels: local ballots, statewide races and federal pledges, as we explained on 4 November. At the local level, data center limits led 40 of 41 measures on Ballotpedia’s 3 November list. At the state level, the Wall Street Journal said candidates promising to freeze data-center development won statehouses in Texas, Georgia, Ohio and Michigan, as we reported.
Analysts size the exposure in two ways. Texas, Georgia and Ohio account for more than 40% of all planned US data-center capacity, according to Evercore ISI, the Journal reported. Goldman Sachs puts about 45% of planned US capacity in states likely to tighten data-center rules, with Texas alone about a quarter, investingLive reported; that makes Texas about 56% of the capacity Goldman sees at risk, by our arithmetic. The two are different measures from different firms.
Permits are decided by county boards, local ballots and state agencies. Of the 41 local measures, 21 sought a ban or pause, by our reading of Ballotpedia’s descriptions, and ten of those would bar only data centers above a size or power line, as we reported.
Texas already has state-level pauses. Starting in August, Abbott began issuing pauses on data center approvals at the state grid and then at the environmental permitting agency, so information about the facilities could be audited, the Texas Tribune reported; the audit was expected to last until December. Governor-elect Gina Hinojosa told CNN the state should go further: “We will have laws to protect Texans, and we need to stop what we are doing with data centers until we have those protections”, Reuters reported, as we covered.
Lever 2: Grid and Power Costs
In Texas the grid operator, ERCOT, and the Public Utility Commission decide who connects and who pays. ERCOT’s September board materials show about 194 GW of large loads conditionally kept in its Batch Zero study, 2.1 times the grid’s record peak demand of 91,134 MW, by our arithmetic, and its energization of new data centers is already paused pending reports due 10 December, as we reported.
Hinojosa’s affordability page says she “will direct the Public Utility Commission to freeze the rates residential customers pay until we have laws to ensure that data centers and other high-load users pay for themselves”. OpenAI’s blueprint makes a similar ask of regulators: “State utility commissions should require large electricity customers to cover the costs they add to the grid”.
Lever 3: Tax Incentives
Tax breaks are set by legislatures. Hinojosa’s platform says House Bill 2712, signed in 2015, gave hyperscale data centers “20-year tax breaks”, and ending tax subsidies is among the priorities she set out for lawmakers, the Tribune reported. Locally, only one of the 41 ballot measures, in Perry, Ohio, dealt with tax incentives.
Lever 4: Water
Water rules sit with localities and states. One of the 41 local measures, in Grove City, Ohio, named a water threshold, and protecting communities’ water supplies is on Hinojosa’s list of priorities for lawmakers, as the Tribune reported it.
Lever 5: State Land
States also own land. The Texas Tribune had previously reported that a draft agenda for the Texas Parks and Wildlife Commission included a discussion about the possible disposition of approximately 1,300 acres of public land near SpaceX’s Starbase, and the item was later removed, as we reported. Hinojosa said she had contacted the department’s executive director and requested that no action be taken during the transition, Click2Houston reported.
In a CNN clip, by our own machine transcript, she said: “We can have a strong working relationship with Elon Musk and his companies, but it has to be on our terms. It cannot be just running over the people of Texas.” The clip does not mention data centers.
How Builders Have Responded So Far
Paying for power. Google confirmed a $17 billion campus in Wyoming and said: “We will pay for 100% of the power we use, and when we build data centers, we’ll cover the infrastructure needs directly driven by our growth”, as we reported. OpenAI pledged to cover full power costs at data centers it will own and said it “will not use non-disclosure agreements (NDAs) with public officials”, as we reported.
Using buildings already wired. The Information reported that Nvidia submitted a proposal to convert some of Brightspeed’s 130-plus telecom buildings and that “Nvidia estimates the sites have roughly 1 gigawatt of available power”, as we reported. Verizon’s Dan Schulman said of its central offices: “they’re power-ready, already permitted, and we have thousands of them”, RCR Tech reported, as we covered.
Paying locally and disclosing. OpenAI said it will “publish site-level water and electricity use”, and that residents in Effingham County, Georgia, “helped shape a $4.8 million grant to provide free school meals for all 14,000 students in the county”. Google said its Wyoming project would generate “hundreds of millions of dollars in local property taxes for Laramie County and School District 1”, and that the data center will be air-cooled “to minimize strain on local water resources”.
Moving the timing or the place. A Goldman Sachs trading-desk note said corporate decision-makers “may instead pull capital deployment forward into 2027”, as we reported. The Wall Street Journal’s summary said would-be builders are weighing their options, “from going overseas to building in the desert”.
What to Watch
In Texas, ERCOT’s reports are due on 10 December. Hinojosa takes office on 19 January, and the regular legislative session is scheduled to end on 31 May; from her inauguration to that last day is 132 days, by our arithmetic. The Goldman summary posted by the Walter Bloomberg account said Goldman “expects any moratoria to be short-lived”; that is Goldman’s view, not ours.
As we read the week, the levers sit at different levels: localities vote on zoning and water, grid operators and utility commissions price power, legislatures set tax breaks, and state agencies hold land.
As we read the responses reported so far, most concern power and local payments; for permits, the reported answer is to use buildings that are already permitted, and none concern state land.
As we read the Texas calendar, the next fixed dates are ERCOT’s reports on 10 December, the inauguration on 19 January and the scheduled end of the session on 31 May.
Gina Hinojosa, Texas Governor-elect, to CNN, as reported by Reuters, 5 November 2026
“We will have laws to protect Texans, and we need to stop what we are doing with data centers until we have those protections”
Three Implications
FOR DATA CENTER DEVELOPERS Five levers now bear on cost, and a Goldman summary named four of them for Texas, Georgia and Michigan.
FOR HYPERSCALERSGoogle and OpenAI have pledged to pay for the power they use, and OpenAI says it will publish site-level water and electricity use.
FOR READING THE NEXT MOVES Texas has fixed dates: ERCOT’s reports on 10 December, the inauguration on 19 January and the session’s scheduled end on 31 May.
The framework says: “Companies that solve the power equation will dominate the AI era; those that don’t will watch their chips sit idle.”
As we read it, state politics is now part of that power equation, through five levers rather than one. The framework does not discuss the 2026 elections or these states.
Update, 6 Nov, 09:38 UTC: a market-wide count. Cleanview, a market-intelligence firm, counts nine states whose governor or governor-elect has promised or already imposed a pause on new data centers; they hold 158.6 GW of planned capacity, 41% of the US pipeline, with Texas at 88.4 GW. Cleanview also wrote that OpenAI’s entire planned pipeline sits in such states. More in Pause Pledges Hit 41% of Data Center Plans, All of OpenAI’s.
Three firms, three measures, not added or averaged: Cleanview’s 41% of the pipeline in states whose governor backs a pause; Goldman Sachs’s about 45% of planned US capacity in states likely to tighten rules, as reported by investingLive; and Evercore ISI’s more than 40% of planned capacity in Texas, Georgia and Ohio, as reported by the Wall Street Journal.
What Is Not Established
The lever-to-response mapping in this piece is our reading: the sources do not say that each response was made because of a particular lever. No governor-elect has issued an order, and the reports do not say which approvals a Texas pause would cover. The Goldman figures come from summaries of its notes, and the Journal’s list of options is from its summary; we did not read either in full. Meta’s Texas project is described only as 1 GW or more.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Five state levers now bear on the cost of an AI data center: permits and zoning, grid and power costs, tax incentives, water, and state land. A Goldman summary named four of them for Texas, Georgia and Michigan, and Texas added the fifth this week near Starbase. Builders have answered by paying for power, using buildings already permitted, paying and disclosing locally, and, in Goldman’s and the Journal’s telling, weighing timing and location.
A note on sourcing. This explainer is compiled from our own reporting of 3 to 6 November 2026, linked throughout; each fact keeps the attribution of the original report. The grouping into five levers and four responses is our framing. Percentages and day counts marked as ours are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
