How many hours of demand response can a Texas data center actually offer tenants?

AI tenants doing diligence on a Texas data center site increasingly ask one specific question: how many hours of demand response (getting paid by the grid to turn off when it’s strained) can you actually offer us? The answer is not one number. It is three separate constraints stacked on top of each other — one mandatory grid program, one voluntary grid program, and one air-pollution rule — each with a different operational meaning and a different ceiling. Most data center operators do not separate them in the tenant pitch, which makes the “flexible” claim either vague or unprovable when a tenant procurement team runs the math.

The combined number is what underwrites the LOI. Here are the three.

Constraint 1: mandatory curtailment via PCLR

ERCOT (Electric Reliability Council of Texas, the grid operator) offers Provisional Controllable Load Resource (PCLR) status: a binding agreement to actually shut load off when ERCOT says so, in exchange for getting connected to the grid faster through a fast-track process called Batch Zero. When ERCOT calls a PCLR event, the load really curtails — no running backup generators instead, no opt out. ERCOT dispatches these through SCED (Security-Constrained Economic Dispatch, the every-five-minute market that decides which generators run) based on local grid congestion at the data center’s electrical node. How often it actually fires depends on the site’s “shift factor” (how much that specific node contributes to local congestion) measured against historical congestion-price patterns.

Expected curtailment varies wildly by location. A site in the Panhandle near the wind-export bottleneck plausibly sees hundreds of hours/year of forced shut-off. A site in the Coastal Bend or East Texas may see fewer than 50. The geographic dispersion is large enough that “PCLR enrollment” means very different things at different parcels. The technical breakdown lives in the companion post on ERCOT PCLR and Batch Zero.

Constraint 2: voluntary grid programs (ERS and CLR)

Emergency Response Service (ERS) is ERCOT’s voluntary demand-response program: loads commit to be available to shed power in four-month windows and get paid for being on call. ERS deployment is rare in practice: a 15 to 30-minute test event each year is guaranteed, and real grid emergencies that trigger ERS historically average closer to zero than to ten per year. Total annual hours an enrolled site is actually asked to shed are typically well under twenty.

Voluntary (non-provisional) Controllable Load Resources, the regular opt-in version of CLR, also participate in ERCOT’s ancillary services markets — the “reserves” products the grid buys in case generation falls short (Responsive Reserve Service, ECRS, Non-Spin). Calls are more frequent than ERS but still infrequent operationally; deployment events for well-bid CLRs typically cluster in the tens, not hundreds, per year. For a 60 MW Texas data center, voluntary CLR plus ERS revenue can run well into seven figures annually with operational disruption measured in tens of hours, not hundreds.

ERS and voluntary CLR can be stacked on top of PCLR in principle, though the rules about how to count overlapping events matter. They are not a substitute for PCLR — the voluntary programs trade revenue for the right to say no, where PCLR trades a binding shut-off obligation for getting energized faster.

Constraint 3: how many hours your backup generators can legally run

This is the constraint most often forgotten. Diesel backup generators are subject to Title V of the Clean Air Act — the federal program that regulates “major sources” of air pollution — when their potential to emit (the maximum they could legally pollute if run at full allowed hours) exceeds 100 tons/year of any “criteria pollutant” (the six pollutants EPA tracks: NOx, particulate matter, sulfur dioxide, carbon monoxide, ozone precursors, lead) in an “attainment area” (a region currently meeting EPA air quality standards). The threshold drops in non-attainment areas. Williamson County is in the Austin-Round Rock MSA and currently attainment/unclassifiable for the 2015 ozone NAAQS (National Ambient Air Quality Standards), so the federal 100 ton/year threshold applies to NOx (nitrogen oxides, the main pollutant from diesel combustion).

For a typical AI-HPC data center with 100 MW of Tier 4 final diesel backup generators — the current strictest EPA non-road emission standard at 0.40 g/bhp-hr NOx (grams of NOx per brake-horsepower-hour, the engine work unit EPA regulates) — the runtime budget at full load before NOx alone triggers Title V applicability is:

100 MW × 1.34 hp/kW × 1000 = 134,000 hp
134,000 hp × 0.40 g/bhp-hr = 53,600 g/hr NOx
53,600 g/hr ÷ 907,185 g/ton = 0.059 tons/hr
100 tons/year ÷ 0.059 tons/hr ≈ 1,690 hours/year

Roughly 1,500 to 1,700 hours/year of full-load runtime headroom on NOx alone, before Title V applicability is triggered. Other regulated pollutants (fine particulate PM2.5, carbon monoxide, hazardous air pollutants or HAPs) bind at different runtime numbers; in practice NOx is usually the first constraint for diesel-backed data centers, and dispersion modeling (the EPA-required simulation of how pollutants travel from the smokestack to the surrounding air) can fail before annual emission totals do for sites near sensitive receptors like schools or hospitals. The number is also a function of which EPA tier the gensets meet — Tier 2 gensets emit roughly 16x more NOx per unit of work than Tier 4, and the runtime budget shrinks proportionally.

What the tenant pitch can credibly say

Stack the three constraints — mandatory grid curtailment, voluntary grid programs, and air-permit runtime cap on backup generators — and the operator-side numbers fall out in plain ranges:

The credible offer to an AI tenant in Texas is roughly 100 to 300 hours/year of curtailable load via PCLR plus voluntary CLR, with several hundred hours of remaining backup generator runtime headroomunder the Title V cap for non-curtailment use cases (pausing AI training during peaks and running on backup generation, planned maintenance windows, scheduled inference ramps). That is a real flexibility budget. It is not “unlimited shut-off for unlimited duration,” which is what the marketing word “flexible” sometimes implies.

For a tenant procurement team, the two numbers that matter are: (a) committed curtailable hours/year through ERCOT programs (mandatory plus voluntary), and (b) remaining backup-generator runtime headroom under the air permit. Both are knowable. Most operators do not compute or expose them, which is why “flexible AI/HPC” in a marketing one-liner often does not survive a tenant’s procurement review.

What Cliff is building

Cliff computes these numbers per site. Given a parcel, a generation stack (turbines, diesel backups, BESS), and a target enrollment posture, we return expected PCLR curtailment frequency (modeled from the site’s shift factor against historical congestion prices), ERS and voluntary CLR revenue and event modeling, and Title V runtime headroom by pollutant under the local attainment-status profile. Every input is cited to a primary source (ERCOT filing, TCEQ permit, EPA rule). If you have a Texas project and tenants asking how many hours of flex you can actually offer, send us the location.

Primary sources

• ERCOT PGRR145 (Batch Zero / PCLR mechanism)
• PUC Project 58481 (16 TAC §25.194 large-load interconnection standards)
• ERCOT Emergency Response Service program
• TCEQ Austin-Round Rock current attainment status
• EPA Title V operating permits program
• EPA Tier 4 nonroad diesel engine emission standards