X frames AI data centers as national supremacy or grid parasites; the reliability half of the fight is now a filed FERC and NERC proceeding on emerging large loads, not a thread.
MSM such as Bloomberg covers demand growth and blackout risk, less the reliability standards process where the large-load rules get written.
X frames co-located AI data centers as either the engine of national AI supremacy or a private drain on a public grid.
The fight over powering artificial intelligence has a second front, and it is being contested over reliability in a Federal Register notice rather than a thread.
The paper argued on June 28 that FERC weighs who pays when AI plants bypass the shared grid, a cost question convened at a 2024 commissioner-led technical conference on loads co-located at generating facilities. [2] The question has since grown a reliability half. In February 2026, the Federal Energy Regulatory Commission published notice that its staff would attend a North American Electric Reliability Corporation session on emerging large loads — the industry's term for the gigawatt-scale data centers now bending planning assumptions built for a slower-growing grid. [1]
A staff-attendance notice is dull by design, and that is the point. It places the AI-load debate inside the reliability standards process, where NERC develops the rules that keep the bulk power system stable and FERC oversees them. [1] The venue is not a stage for a verdict; it is where the operating limits of a new class of customer get studied before they become mandatory.
The precedent, as before, is built one grid operator down. PJM, the largest U.S. system operator, runs the interconnection-process reform where a co-located or large-load proposal is studied and where "using the grid" is defined for a customer that claims not to. [3] That is allocation and reliability work at once — assigning both the cost and the risk to someone named in a filing.
On X, a co-located data center is the cornerstone of national AI supremacy or a parasite draining public power for private compute. Mainstream coverage — Bloomberg, the trade press — reports the demand curve and the blackout warnings, both real. The record asks the prior questions both skip: when a gigawatt attaches to the system, can the grid stay reliable, who studies it, and where is that study filed. [1][2]
This is the divergence the paper keeps. The public argument is whether AI deserves the power. The institutional question is narrower and longer-lived: whether the standards process can absorb loads this large without shifting cost or risk onto households, and it is being answered in proceedings that outlast any thread. [3]
The feed will crown a winner tonight. FERC and NERC will decide, in filings that are studied and appealable, whether the grid bends or breaks under the new load — and that answer will still be on the record a decade from now. [1]
-- THEO KAPLAN, San Francisco