Key Takeaways

  • Voters passed data center moratoria in 43 out of 44 local ballot measures during the 2026 midterms, locking up permitting nationwide.
  • Debt markets tightened abruptly, ending cheap borrowing and thrusting AI infrastructure into an equity funding era.
  • Specialized neoclouds face an existential squeeze because they lack the balance sheets to fund buildouts without debt.
  • Hyperscalers like Amazon are pulling away, capturing compute market share while operations migrate to locations like Harwood, North Dakota and Canada.

Local Ballots Block Regional Grid Expansion

Local politics just collided directly with AI compute demand. In the 2026 midterm elections, local communities voted down data center construction almost everywhere they had the chance. As John Coogan reported, “The results are in data center moratoria went 43 to 1. That was the final score. 43 to 1.” (play from 22:25)

Voters across the country signaled that they will fight server farm construction over power grid and water strain. California saw its own political threat, where Jordi Hays observed that “Prop 40 as it was written was going to destroy the model for building technology companies in California.” (play from 9:50) While founders spent years obsessing over chip allocations and software architecture, municipal permits emerged as the real bottleneck.

This sudden wall of local resistance forces immediate geographic migration. When 43 out of 44 local measures reject new facilities, compute builders cannot remain in legacy regions. Coogan noted the immediate operational consequence: “All hyperscalers must now converge on Harwood North Dakota.” (play from 22:31) Builders are looking beyond typical tech corridors toward Canada and rural North Dakota, where power and space remain open to development.

The Credit Crunch Traps Neoclouds

Permitting blocks arrived at the worst possible moment for specialized cloud providers. For two years, neoclouds used cheap debt to buy clusters of high-end GPUs and lease them to startups. That model required constant access to low-interest loans. Now, credit markets have shut down.

Coogan highlighted this shift plainly: “The most obvious reason why the stock market has been rocky over the last few weeks is the tightening of credit markets. We are needless to say full-on into the equity funding era.” (play from 9:01)

Without debt, neoclouds cannot buy real estate, build power substations, or purchase new hardware racks. Their rapid growth depended entirely on cheap borrowed money. Coogan pointed out the grim irony facing these challenger providers: “Neoclouds are on the edge. They have been at the forefront of the credit concerns. From a certain perspective being unable to build data centers has a silver lining in that they couldn't get the money to pay for them anyways.” (play from 11:13)

When borrowing dries up, only companies with massive cash reserves survive. Neoclouds that promised lower hourly GPU rates now hold expensive commitments on hardware they cannot scale or house.

Balance Sheets Crown Hyperscalers

Ben Thompson's analysis of the midterm results points to a clear structural winner: the established hyperscalers. Amazon, Microsoft, and Google do not need debt markets to approve a new facility. They generate billions of dollars in operating free cash flow every quarter. When credit contracts, balance sheet equity becomes an unassailable moat.

Coogan summarized the shift: “The downside for the Neoclouds, upside for hyperscalers. Yeah, Jeff Bezos is doing well. Amazon's way up today. They are more likely to grow their share of AI compute than they were before.” (play from 14:26)

When local permitting requires years of litigation, political lobbying, and grid upgrades, small cloud firms fail. A hyperscaler can absorb a two-year permitting delay, fund municipal infrastructure improvements, or run high-voltage transmission lines into Harwood, North Dakota. AI compute demand continued growing after the midterms, but the votes concentrated compute supply inside large corporate balance sheets.

What to Do With This

Audit your compute infrastructure dependencies by Friday afternoon. If your training runs or inference pipelines rely on long-term reserve contracts with debt-financed neoclouds, demand written proof of their data center delivery timelines and power allocations. Begin testing secondary deployments on Amazon Web Services or another hyperscaler so that a sudden facility pause or credit freeze does not take your product offline.