A widely cited thesis has been circulating in energy and data center circles: that one-third of U.S. grid capacity sits unused, and that virtual power plants, batteries, and demand flexibility could unlock 100–300 GW of new data center load without building a single new power plant.
The Duke University study underlying this claim is more careful than the popular framing suggests. Its actual finding: if data centers agreed to curtail demand for approximately 40 hours per year, roughly 100 GW of new industrial load could interconnect without new generation or transmission. That’s a meaningful finding — and a significant assumption.
The assumption is that curtailment is a clean administrative action. At the node, it isn’t.
When a hyperscale data center drops load abruptly, it generates its own power quality disturbance back onto the T&D interface — a sub-cycle event, not a clean administrative curtailment. NERC’s May 2026 Level 3 Essential Actions Alert — its highest urgency tier — was triggered by documented instances of data centers dropping more than 1,000 MW of load in seconds in the Eastern Interconnection and Texas Interconnection. NERC explicitly stated these events were not predicted by current planning studies.
This is the gap between the administrative map and the physical terrain.
FERC’s June 18, 2026 show cause orders to all six RTOs/ISOs tell the same story. If 300 GW of clean capacity were simply waiting in available windows, FERC would not need emergency Section 206 orders requiring tariff reform by August 17, 2026.
Administrative capacity accounting tells you what the spreadsheet says. Sub-cycle observability tells you what the physics says. The full analysis examines three specific technical failures in the administrative capacity thesis — and what the regulatory record actually shows.
Leave a Reply