Large flexible load is an operating category
ERCOT created a process for large flexible load customers as data centers, cryptocurrency mining operations, and industrial facilities increased their demand for power. The category matters because very large customers can affect grid reliability differently from ordinary commercial growth.
The EIA analysis separates approved demand capacity from the electricity that facilities are expected to consume. That distinction keeps the forecast tied to operating behavior instead of assuming every approved megawatt runs continuously from the first day.
The queue is not the load
Projects can appear in several stages before they become live demand: early plans, submitted studies, plans under review, approved capacity, and operating facilities. A large application total therefore describes potential pressure on the grid, not the amount of electricity already being served.
ERCOT estimates large flexible load from approved demand capacity and the electricity consumed by operating facilities. Forecasts should assign a probability and a schedule to each stage, then separate the capacity approved by ERCOT from the share that facilities actually use through the day.
Curtailment changes the value of capacity
Some large-load facilities can reduce electricity consumption during periods of high system demand or low generation availability. Participation in energy and ancillary-service markets can give ERCOT another operating tool when wholesale power prices signal a tight system.
Flexibility is not identical across data centers. Workloads, service commitments, cooling systems, and backup arrangements affect whether a facility can curtail, how long it can remain reduced, and how quickly it can return without creating another demand spike.
Who benefits and who is exposed
ERCOT benefits when large customers provide accurate schedules and credible demand response, because generation and reliability planning can reflect actual operating choices. Flexible facilities may also benefit when market participation compensates them for reducing load during the most expensive hours.
Developers are exposed when a site depends on capacity that remains under review, while generators are exposed when they build around an application queue that does not become sustained demand. Other customers can face higher wholesale prices if large loads arrive faster or operate less flexibly than the forecast assumes.
ERCOT LFL demand
The forecast uses approved capacity.
STEO uses approved LFL demand capacity from ERCOT and historical electricity consumption to forecast regional electric load. Large-load forecasts should distinguish proposed customers, ERCOT approvals, operating facilities, and actual demand.
A capacity team can apply that structure by updating each project's status as evidence changes. An approved connection should move into the operating forecast only with a credible energization date and utilization profile. A facility that can curtail should also show how much load is flexible and under which grid conditions that flexibility is available.
Generators and grid planners need the same conversion view. Building around the full application queue assumes that every project becomes sustained demand, while ignoring approved projects assumes the opposite. Tracking approvals, operations, actual consumption, and market participation separately gives a more useful range for generation and reliability decisions. Each STEO revision can then follow changes in approved capacity, utilization, and expected electricity demand.
For operators, the practical test is whether the forecast records both consumption and flexibility. A large facility that is online but can reliably reduce demand during a stressed interval creates a different system obligation from a facility with the same normal load and no curtailment capability.
Forecast uncertainty
A key limitation is that future LFL demand depends on ERCOT approvals, facility operations, utilization, and voluntary curtailment. Approval timing, facility completion, utilization, and demand response can all move the result, so one forecast cannot prove how much data-center or mining load will operate in a specific hour.
A useful planning model therefore keeps at least three ledgers: projects seeking approval, capacity approved for operation, and measured electricity consumption. Mixing those ledgers inflates certainty and makes it harder to see whether the next constraint is grid approval, construction, or available generation.
Bottom Line
Approved capacity is not live demand.
Texas has a large pipeline of LFL demand, but generation planning depends on approved capacity, utilization, and timing. The strongest data-center power cases will show where a project sits in the ERCOT process, how its utilization develops, and what curtailment it can provide when the grid is tight.
