Classical forum hall with golden accents

AI data centers

2 essays in the Forum tagged "AI data centers".

← All essays
Watching the Cracks

Both Sides of the Cushion: AI Debt, Captive Insurers, and the Four Percent

Daniel Oliver of Myrmikan Capital published a letter on August 14, 2026 tracing the specific institutional path by which an ordinary American's paycheck finances the artificial intelligence build-out through three channels the saver never selects: a 401(k) into index funds holding hyperscaler equity, a pension into investment-grade bonds now 14 percent tied to AI, and a life insurance premium into a private-equity-owned insurer buying private credit secured against graphics processors. The letter is the most rigorously documented account of that chain in print, and this essay draws on it heavily and with attribution throughout. Oliver's central warning concerns a number: United States life insurers report roughly 11.0 trillion dollars in assets against 10.6 trillion in liabilities, an equity cushion of approximately four percent, held by an industry that now owns 849 billion dollars of the two-trillion-dollar private credit market and is increasingly controlled by the same private equity sponsors originating the AI debt it buys. His argument is that credit losses on that debt consume the cushion. This essay accepts that argument and adds the half of it his analysis does not reach. The cushion is measured against liabilities carried at prescribed statutory valuation rates rather than at what it would cost to extinguish them in the market. Article 41 of this catalog established, following Antal Fekete, that a falling interest-rate structure raises the liquidation value of fixed long-duration obligations and that conventional accounting does not record the resulting loss. Life insurance reserves are precisely such obligations. Which means that in the specific scenario Oliver forecasts — the Federal Reserve printing to arrest an AI debt collapse, driving rates down — the four percent cushion is attacked simultaneously from the asset side by credit losses and from the liability side by a mechanism that appears in no statutory filing. The two failures are not sequential. They share a trigger.

Daniel OliverMyrmikanlife insuranceprivate equityprivate creditAI data centersAthenereinsuranceLaw of LiabilitiesWatching the Cracks
Watching the Cracks

Eminent Domain, AI Data Centers, and the Erosion of Property Rights

In May 2026, Georgia Power began invoking eminent domain to acquire 330 properties along a 35-mile transmission corridor — Project Wansley — that reporting has tied to at least four AI data centers. Between 20 and 30 homes are slated for demolition. Hundreds of homeowners face permanent easements with 500-kilovolt towers feet from their bedroom windows. The Georgia case is the most visible current example of a structural pattern that has been accumulating since the Supreme Court's 2005 Kelo decision and that the framework's housing analysis has been identifying as a specific saleability risk. This essay traces eminent domain from Magna Carta through Kelo through the current AI-infrastructure expansion, examines the New Albany Company's privatized-governance model in central Ohio as the most developed example of the pattern, and applies the framework's saleability analysis to what the cases reveal about property rights in 2026.

eminent domainKeloproperty rightsGeorgia PowerProject WansleyNew AlbanyWexnerMengerFeketeAI data centerssaleabilityweaponization risk