Firm Capacity

By: Mike Ye x Ella (AI)

Firm capacity is power expected to remain available through stressed grid conditions: heat waves, supply shortfalls and emergencies. For AI infrastructure, it is the difference between a megawatt on paper and a megawatt that keeps accelerators running when the grid is under strain.

exmxc ranks power in four levels:

Nominal capacity: a planned or contracted megawatt.
Connected capacity: a load physically authorized to draw from the grid.
Firm capacity: power expected to remain available through stressed conditions.
Sovereign capacity: firm capacity reinforced by self-generation, storage, fuel security, contractual rights and operational flexibility.

The gap between connected and firm capacity is curtailment exposure: the risk that a load is cut when the system is constrained. Grid operators have begun to claim authority to curtail data centers and other large loads with backup generation before broader load shedding. The strategic asset is therefore no longer the right to connect. It is the right and the physical ability to stay energized when the system is tight.

The distinction carries into valuation. Accelerators without reliable power are stranded or impaired capital, so compute capacity should be counted by the megawatts that are firm and usable through stressed periods, not by those connected on paper. Contracted demand flexibility, such as virtual power plants and demand response, can add system capacity, but it should not be counted as delivered firm power. Uprating existing nuclear plants is one of the faster ways to add firm megawatts.