Curtailment Clauses: The Impossible Choice
What you are actually signing when you accept a curtailment clause for faster grid access, and how to design operations to survive forced load reduction without killing critical workloads.
Who this is for
- ■Hyperscale and colocation data center operators evaluating curtailable interconnection service.
- ■Crypto mining and EV charging depot decision makers weighing speed to power against firm service.
- ■Greenfield industrial manufacturers and smelters needing more than 20 MW of new grid capacity.
- ■Anyone about to sign a large-load interconnection agreement in PJM, ERCOT or SPP.
Should we accept mandatory curtailment provisions to accelerate grid access, and how do we design operations to survive forced load reduction?
Curtailment clauses did not appear because grid operators got greedy. They appeared because the queue broke. In Q1 of 2026, 198 GW of large load applied for interconnection in ERCOT alone. ERCOT's entire installed capacity is about 103 GW. Nearly twice the grid, asking to plug in, in a single quarter.
PJM is running the same math from the other direction: about 40 GW of generation retiring between 2022 and 2030 while load climbs, with a projected capacity shortfall of 15 GW by 2030. That is the setup. When the grid cannot serve everyone, someone gets curtailed. Curtailment clauses decide who.
31%of this guide, read. The rest of it is below.
- 02 The mechanism The curtailment hierarchy and where you sit
In PJM there are two categories that matter for large loads taking faster access. Non-Capacity-Backed Load, or NCBL, accepts curtailment before traditional load when the market cannot serve everyone. In exchange, you get on the grid without waiting for new generation. Interim Network Integration Transmission Service, or Interim NITS, goes further: you interconnect before new transmission is even built to serve you.
PJM curtailment orderInterim NITS curtails first, then NCBL, then traditional load. Your tier decides when your power gets cut. Texas SB 6 mandates that any new large load at or above 75 MW interconnected after December 31, 2025 must comply with mandatory ERCOT-directed curtailment during emergency conditions. You also pay for transmission studies, minimum $100,000, and infrastructure upgrades on top.
In SPP the structure is three-tiered. CHILLS provides 7 years of non-firm transmission service as a bridge for loads ready to energize before transmission upgrades finish. PALS, Price Adaptive Load Service, is market-based and non-firm, and it requires full curtailment within 5 minutes. That is a hard engineering requirement. Crypto mining can technically meet it. A hyperscale compute facility serving financial or government workloads generally cannot.
ISO or Utility Curtailment mechanism Speed or exposure PJM Interim NITS, then NCBL, then traditional Event-driven, tier-based ERCOT (SB 6) Mandatory during emergency conditions 75 MW SPP PALS Dispatches to zero on reliability events 5 minutes Idaho Power Remote disconnection, summer peaks 225 hours Entergy Arkansas Interruptible service, per agreement 20 events/year A recognizable large-load tariff archetype is emerging across the DELTa database. Three objectives: assign incremental costs to the large load to prevent cost shifting, implement long contract terms and minimum bills to reduce stranded-asset exposure, and define exit rules that are actually enforceable.
The archetype, in ranges1.5$M/MWCollateral, high end per MW80%Typical minimum bill vs contracted capacity20yearsLongest tariff terms observed20%Typical change band before exit feesThese are the numbers showing up in filings. Your utility may land differently, which is exactly why you read the schedule before you negotiate.For a 100 MW facility at the high collateral end, you are looking at 150 $M in posted security. Dominion specifically requires 1.5 $M/MW for a 14 years contract. Of the 77 filings reviewed in DELTa, 44 filings include additional collateral requirements beyond base rates, and 33 filings include numeric minimum-bill provisions.
Dominion's GS-5 rate class for loads at or above 25 MW is the most aggressive upstream cost assignment yet: 85% take-or-pay on contracted transmission and distribution capacity, plus 60% on generation demand, regardless of actual usage. Under GS-5 you cannot curtail your way out of the bill. You owe the infrastructure charge either way.
- 03 What it costs you The tariff archetype, in dollars
A recognizable large-load tariff archetype is emerging across the DELTa database. Three objectives: assign incremental costs to the large load to prevent cost shifting, implement long contract terms and minimum bills to reduce stranded-asset exposure, and define exit rules that are actually enforceable.
The archetype, in ranges1.5$M/MWCollateral, high end per MW80%Typical minimum bill vs contracted capacity20yearsLongest tariff terms observed20%Typical change band before exit feesThese are the numbers showing up in filings. Your utility may land differently, which is exactly why you read the schedule before you negotiate.For a 100 MW facility at the high collateral end, you are looking at 150 $M in posted security. Dominion specifically requires 1.5 $M/MW for a 14 years contract. Of the 77 filings reviewed in DELTa, 44 filings include additional collateral requirements beyond base rates, and 33 filings include numeric minimum-bill provisions.
Dominion's GS-5 rate class for loads at or above 25 MW is the most aggressive upstream cost assignment yet: 85% take-or-pay on contracted transmission and distribution capacity, plus 60% on generation demand, regardless of actual usage. Under GS-5 you cannot curtail your way out of the bill. You owe the infrastructure charge either way.
Gas is not automatically the answer, and storage is not automatically enough
The myth is that behind-the-meter gas solves curtailment. It gets you online, but the LCOE is heavy. Ascend Analytics data puts an industrial combustion turbine running at high capacity factor above 120 $/MWh. Even a natural gas combined cycle with a 35 years asset life prices above ERCOT's grid cost forecast in their analysis. Gas as a bridge, yes. Gas as your long-term hedge against curtailment, the math gets ugly.
The other myth runs the other way: that a battery covers everything. Ascend modeling shows a 4 hours BTM BESS paired with a non-capacity-backed load connection covers the majority of expected curtailment events in normal weather years, and storage can effectively receive 100% accreditation by shaving critical peak demand. But multi-day extreme events, Uri in ERCOT, the 2019 polar vortex in PJM, require longer duration or gas backup. Duration sizing is market- and weather-specific.
GasFast to deploy, expensive to run
Gets you online. LCOE for an industrial CT lands above the ERCOT forward curve. Stress test against your market's actual forward gas and power curves before you assume it pencils.BatteryCovers the common case, not the tail
A four-hour BESS handles nearly all curtailment scenarios in most markets. Multi-day extreme events break that assumption. Run tail-risk for your ISO.CombinedBattery for frequency, gas for duration
Storage for the routine curtailment events, gas for the multi-day tail. Sized correctly, you keep critical workloads live in both scenarios.NeitherWorkload migration
For crypto, ramping to zero on price is native. For hyperscale serving regulated workloads, workload migration architecture has to be designed in from day one, not bolted on. - 04 The myth The BTM stack that mitigates curtailment
Gas is not automatically the answer, and storage is not automatically enough
The myth is that behind-the-meter gas solves curtailment. It gets you online, but the LCOE is heavy. Ascend Analytics data puts an industrial combustion turbine running at high capacity factor above 120 $/MWh. Even a natural gas combined cycle with a 35 years asset life prices above ERCOT's grid cost forecast in their analysis. Gas as a bridge, yes. Gas as your long-term hedge against curtailment, the math gets ugly.
The other myth runs the other way: that a battery covers everything. Ascend modeling shows a 4 hours BTM BESS paired with a non-capacity-backed load connection covers the majority of expected curtailment events in normal weather years, and storage can effectively receive 100% accreditation by shaving critical peak demand. But multi-day extreme events, Uri in ERCOT, the 2019 polar vortex in PJM, require longer duration or gas backup. Duration sizing is market- and weather-specific.
GasFast to deploy, expensive to run
Gets you online. LCOE for an industrial CT lands above the ERCOT forward curve. Stress test against your market's actual forward gas and power curves before you assume it pencils.BatteryCovers the common case, not the tail
A four-hour BESS handles nearly all curtailment scenarios in most markets. Multi-day extreme events break that assumption. Run tail-risk for your ISO.CombinedBattery for frequency, gas for duration
Storage for the routine curtailment events, gas for the multi-day tail. Sized correctly, you keep critical workloads live in both scenarios.NeitherWorkload migration
For crypto, ramping to zero on price is native. For hyperscale serving regulated workloads, workload migration architecture has to be designed in from day one, not bolted on.You have more leverage than the tariff sheet suggests, and less time than you think. Bringing your own generation changes the negotiation. In ERCOT, large loads with BTM generation are treated more favorably in the batch study process. Maryland's Utility RELIEF Act lets data centers jump ahead in the queue if they provide their own power. Georgia Power's customer-identified resource provision lets you commit to your own supply.
The window is narrowing. FERC is expected to act on its ANOPR by June 2026. Tariff filings jumped from 14 filings across all of 2018 through 2024 to 53 filings in 2025 alone. Bilateral negotiation gives way to commission-designed defaults. The strategic window to negotiate favorable, project-specific terms is now.
Tariff filings, then and nowThe pace of standardization is accelerating. Bilateral negotiation windows close as filings become defaults. - 1 Confirm your exact tier in the curtailment stack, in writing, before you sign.
- 2 Match the curtailment notice period, whether instantaneous, five minutes or hours, to your actual workload shedding architecture.
- 3 Model total financial exposure across the full contract term: collateral, minimum bill, exit fees and take-or-pay together, not separately.
- 4 Run BTM stack LCOE against your market's forward curves and stress test battery duration against tail-risk events specific to your ISO.
- 5 Build regulatory review triggers into the agreement so terms can be revisited when FERC issues standardized rules.
- 05 Your leverage What to ask and how to negotiate
You have more leverage than the tariff sheet suggests, and less time than you think. Bringing your own generation changes the negotiation. In ERCOT, large loads with BTM generation are treated more favorably in the batch study process. Maryland's Utility RELIEF Act lets data centers jump ahead in the queue if they provide their own power. Georgia Power's customer-identified resource provision lets you commit to your own supply.
The window is narrowing. FERC is expected to act on its ANOPR by June 2026. Tariff filings jumped from 14 filings across all of 2018 through 2024 to 53 filings in 2025 alone. Bilateral negotiation gives way to commission-designed defaults. The strategic window to negotiate favorable, project-specific terms is now.
Tariff filings, then and nowThe pace of standardization is accelerating. Bilateral negotiation windows close as filings become defaults. - 1 Confirm your exact tier in the curtailment stack, in writing, before you sign.
- 2 Match the curtailment notice period, whether instantaneous, five minutes or hours, to your actual workload shedding architecture.
- 3 Model total financial exposure across the full contract term: collateral, minimum bill, exit fees and take-or-pay together, not separately.
- 4 Run BTM stack LCOE against your market's forward curves and stress test battery duration against tail-risk events specific to your ISO.
- 5 Build regulatory review triggers into the agreement so terms can be revisited when FERC issues standardized rules.
When curtailable service is worth signing
✓ Sign it- Your workloads can tolerate scheduled or short-notice curtailment without breaching customer SLAs.
- You have, or can finance, a BTM stack sized to your ISO's tail-risk curtailment profile.
- Speed to power is worth more to your business than firm service premium pricing.
- You have modeled the full contract exposure and the numbers survive a worst-case load year.
✗ Walk or renegotiate- Your workloads are regulated, financial or safety-critical and cannot migrate in five minutes.
- The tariff take-or-pay percentages leave you owing most of the bill even at zero consumption.
- You have no BTM resource plan and no capital to build one.
- The exit fees and collateral requirements exceed your acceptable balance sheet exposure over the contract term.
- Decision matrix
When curtailable service is worth signing
✓ Sign it- Your workloads can tolerate scheduled or short-notice curtailment without breaching customer SLAs.
- You have, or can finance, a BTM stack sized to your ISO's tail-risk curtailment profile.
- Speed to power is worth more to your business than firm service premium pricing.
- You have modeled the full contract exposure and the numbers survive a worst-case load year.
✗ Walk or renegotiate- Your workloads are regulated, financial or safety-critical and cannot migrate in five minutes.
- The tariff take-or-pay percentages leave you owing most of the bill even at zero consumption.
- You have no BTM resource plan and no capital to build one.
- The exit fees and collateral requirements exceed your acceptable balance sheet exposure over the contract term.
Questions for your morning huddle- What tier does our interconnection agreement place us in, and have we modeled the curtailment frequency and duration that tier implies for our specific market?
- Does the curtailment notice period in our agreement actually match our workload shedding architecture, or is there a gap between the legal term and the engineering reality?
- Have we run LCOE on our BTM options with our market's forward curves, and stress tested battery duration against tail-risk scenarios for our ISO?
- What are the collateral, minimum bill and exit fee provisions in the tariff we are being offered, and what is our total financial exposure across the full contract term?
The one thing to rememberThe decision on curtailment clauses is not whether to accept curtailment. It is which curtailment structure, at what MW threshold, with what notice requirements, backed by what BTM resource stack, and protected by what exit rights.
Before your next interconnection negotiation, put those five dimensions on one page with your actual numbers. If any dimension is blank, do not sign until it is filled in.
Know if the numbers actually pencil out before you sign anything.
A written second opinion on the project in front of you, whether that is a rate change, new equipment, or a renewable installation.
- 01A short call, to figure out quickly whether we can actually be helpful. If we can't, we'll say so on the spot.
- 02We pull the data, your bills, your rate structure, vendor proposals, project specs.
- 03You get the verdict in writing: whether the payback will materialize, and the opportunities or risks nobody has raised.
Get a Blueprint at blueprint.tac-nrg.com Free for Indiana-based operations spending five figures or more a month on electricity. No obligation. You keep the write-up either way.