AI Data Center Veto Points and the Lowest-Exposure Approval Route

Where AI Data Centers Get Stopped, How Authorization Risk Moves and How Developers, Hyperscalers and Investors Should Sequence Approvals and Capital

Oracle · Blue Owl · Stack Infrastructure · Blue Fern Development · Federal Energy Regulatory Commission · Electric Reliability Council of Texas · Dominion Energy · AEP Ohio · Georgia Power · Loudoun County · Cle Elum · Washington State · Texas · Pennsylvania · New York · Virginia · Ohio · Georgia · Arizona · New Mexico · Wisconsin

Two installments of the AI Infrastructure Authorization Series. The first maps where a data center can be stopped, and the second turns that map into a route.

Full publications: https://www.mindcast-ai.com/p/aidc-veto-points · https://www.mindcast-ai.com/p/aidc-clearing-veto-points

Oracle's Project Jupiter in New Mexico holds its county approvals and could cost up to $165 billion. Reuters reported on September 24, 2026 in “Oracle Cites ‘Force Majeure’ to Shield Itself on Controversial Data Center, Bloomberg Reports” that Oracle had sent a force majeure notice to the campus developer. The approved campus cannot run until a fuel pipeline crosses state land, and the State Land Office rejected the route in March and again in July.

Central finding: A data center passes a chain of veto points rather than one approval, and clearing one point moves the remaining risk to whoever holds the next. Projects fail when capital becomes irreversible before the approval that decides the project becomes durable, so the lowest-exposure route resolves that approval first and keeps every other commitment reversible.

The two papers work as a pair. AI Data Center Veto Points Show Where Projects Get Stopped and How Authorization Risk Moves maps the risk. The Lowest-Exposure AI Data Center Approval Route for Developers, Hyperscalers and Investors sets out what to do about it.

How Authorization Risk Moves and How the Route Answers It

Six mechanisms carry the argument across both papers. Each changes when a developer, tenant or lender should commit the next dollar.

Ten veto nodes. The first paper identifies ten points where an institution can stop a project, from use eligibility and water to interconnection, fiscal terms and expansion. Five settings make the same nodes behave differently by state. Presence and holder decide who can act, sequencing and vesting decide when, and override decides whether a developer can force a node open.

Cleared is not vested. A cleared approval exists today, while a vested or durable approval survives the next rule change. Cle Elum, Washington passed an emergency moratorium three days after a developer unveiled its project, yet the site's rules may have vested under a 2002 development agreement. The gap between clearance and vesting is where capital sits most exposed.

The Critical Veto Path. The second paper names the set of open approvals that can stop a specific project and that other commitments depend on. A grid-connected campus puts power on the path, while onsite generation swaps power for air and fuel approvals, as Jupiter shows.

Two route regimes. Where a physical resource such as power or water decides viability, the route is a problem of structural bottlenecks. Where a discretionary local vote decides viability, the route is a problem of strategic timing, because an announcement can provoke a moratorium before the project's rights become durable.

Necessary and avoidable exposure. Some capital must go at risk early to hold a grid position or a site, and the rest buys no right at all. Virginia regulators now require large loads to sign 14-year contracts with minimum charges of 85% of transmission and distribution demand, and six regional grid operators are due to file new large-load terms by November 16, 2026.

Levers and archetypes. Current rules and observed projects produce at least nine recurring state archetypes, from Virginia's return to local review to New Mexico's fuel-route gate. Four parties hold the levers that optimize a route, and the anchor tenant is best placed among private parties because it alone can hold sites across archetypes.

What the Full Publications Add

The two papers run about 13,800 words across thirty sections and carry twelve MindCast Simulation Predictions with bands, falsifiers and named verification sources. The first paper adds the ten-node table and four state arrangements side by side, plus a node-by-node first-move map and five tests of veto migration. The second adds nine state archetypes, a 600 megawatt worked route through a mid-project redesign and a capital release schedule for seven classes of capital. Both papers carry stakeholder risk mitigation with owners, deadlines and residual exposure for every prediction.

Read the veto map · Read the approval route model

MindCast Simulation Predictions

MindCast carries twelve Simulation Predictions across the two papers: seven in the veto map and five in the approval route. The seven below span both.

• Veto map S1 (60–70%): Cle Elum's moratorium fails to reach the Blue Fern project by December 31, 2026.

• Veto map P2 (70–82%): A Pennsylvania data center project has state permit issuance or a review clock held because a local approval remains unresolved or denied by December 31, 2027.

• Veto map P3 (72–82%): Texas resumes data center grid-connection approvals under standing audit rules by July 31, 2027.

• Veto map S4 (65–80%): At least 60% of eleven tracked local moratoriums convert into written terms at expiration by July 30, 2028.

• Approval route P1 (70–80%): At least two of the six federally regulated grid operators file large-load rules requiring a readiness deposit, financial security or cost-recovery commitment before firm service by December 31, 2026.

• Approval route P2 (65–75%): At least two further large data center projects produce public force majeure, change-in-law or tenant deferral notices tied to an approval delay by December 31, 2027.

• Approval route S2 (75–85%): At least three local moratoriums adopted between October 1, 2026 and June 30, 2027 follow within 90 days of the first public report of a specific proposed data center.

Each prediction carries a deadline, a falsifier and a named verification source.

What the Analysis Means for Each Stakeholder

Each stakeholder faces a different commitment decision. The callouts below pair each decision with the mitigating action the full papers detail.

💼 Developers. The decision is which approval to seek first and when to announce. Map the Critical Veto Path before any land closes, and hold land by option until the decisive approvals vest.

🏢 Hyperscalers and anchor tenants. The decision is how much capital to commit across candidate sites. Hold options in sites under different state rules, and tie delivery dates to the durability of the approval that decides each site.

📊 Lenders and investors. The decision is when to release construction capital. Treat an approval list as a claim to audit, and release construction draws only against durable decisive approvals.

⚖️ Counsel. The decision is whether to call a project approved. Identify the vesting trigger at every node, and name the decisive approvals in every tenant and financing agreement.

⚡ Utilities and grid operators. The decision is how much security to demand before firm service. Refundable security and a non-firm service path lower the capital developers must strand without weakening screens against speculative load.

🏭 Equipment suppliers. The decision is how to structure reservations for long-lead equipment. Transferable production slots and refundable reservations let customers hold scarce equipment without stranding capital.

🏛️ State policymakers. The decision is whether to add holds or sequence approvals. Binding early use determinations and coordinated review calendars let developers learn their decisive approval before committing capital.

🏘️ Local governments. The decision is when to write rules. Ordinances written before a project arrives reach more projects than pauses adopted after an announcement.

Conclusion

Jupiter holds its county approvals and still faces a tenant notice over a fuel route the state has twice denied. The veto map explains why clearing the county did not make the campus safe. The approval route shows how naming the decisive approval and funding only the exposure needed to reach it turns open-ended uncertainty into dated decisions.

About MindCast AI

MindCast AI applies Predictive Behavioral Economics + Dynamic Game Theory through Cognitive Digital Twins of the institutions that authorize AI infrastructure. Developers, hyperscalers, lenders and equipment suppliers use the work before committing capital. Clients commission a Veto Profile for one site or a portfolio with a capital release schedule and Simulation Predictions tied to the dates that matter for the decision. Contact [email protected] or visit www.mindcast-ai-simulation.com.

Related Works

• AI Data Center Veto Points Show Where Projects Get Stopped and How Authorization Risk Moves

• The Lowest-Exposure AI Data Center Approval Route for Developers, Hyperscalers and Investors

• The Data Center Authorization Market: A 50-State Regulatory Atlas

• The Model AI Infrastructure Authorization Code

• The Data Center Authorization Price

• AI Data Center Moratoriums Are Forecastable — and They End in Pricing Rules, Not Bans

• The Grid-Anchored Clean Power Bargain

• The Clean Compute Match

• The Authorization Market — Standardized Bargaining, Rationed Power, and the Competition to Build America’s AI Infrastructure

• AI Data Center Authorization Bargaining Power

• AI Datacenter Edge Computing: Ship the Workload Not the Power