Inside ERCOT’s Merchant BESS Economics: Where Revenue Stacking Actually Pays
In two and a half years, ERCOT has become the world’s largest proving ground for merchant battery storage. Installed battery capacity crossed 12 GW in Q3 2025 (Modo Energy), with more than 30 GW in the interconnection queue and no signs of the pipeline slowing. Texas now dispatches more grid-connected BESS than any market on earth, and merchant operators in that market have generated the cleanest data set we have on what revenue stacking actually looks like when software meets grid operations at scale.
The headline finding is simple. Operators still optimizing primarily for energy arbitrage are leaving the most profitable product on the table. Since the ERCOT Contingency Reserve Service (ECRS) launched in June 2023, the revenue curve for battery storage in Texas has been reshaped by that single product category. Top-quartile operators have built their revenue stacks around ECRS; bottom-quartile operators have not. Hardware is identical. The difference sits in the dispatch stack and the trading discipline around it.
This piece unpacks what merchant BESS economics in ERCOT actually look like in 2026, why ECRS has rewritten the ranking, and what that means for operators building, buying, or refinancing storage assets in the Texas market. It is a companion to this sprint’s core whitepaper on the software dispatch gap in BESS, which maps the broader control-stack anatomy across all US markets.
ERCOT by the Numbers — 12 GW and Counting
Merchant battery deployment in ERCOT has been close to vertical for three years. Installed capacity crossed 2 GW in mid-2022, 5 GW by mid-2023, 9 GW by mid-2024, and 12 GW in Q3 2025 (Modo Energy). Forward pipeline exceeds 30 GW in the ERCOT interconnection queue. Texas now runs more grid-connected BESS than any other ISO/RTO in the world, by a wide margin.
The buildout has been driven by three reinforcing forces. First, ERCOT’s market design: an energy-only construct with no separate capacity market, which rewards assets that can both arbitrage real-time price volatility and deliver ancillary services on demand. Second, extreme weather exposure, which has produced the price spikes that finance merchant BESS pro formas (Winter Storm Uri in 2021 reset underwriting assumptions permanently). Third, a protocol and interconnection path that, while imperfect, has been materially faster than California’s Rule 21 process or PJM’s queue.
The practical implication for software teams: at 12 GW installed and climbing, Texas is where the operational data lives. Dispatch strategies that work in ERCOT are being exported to CAISO, PJM, and MISO as those markets tighten their storage participation rules. The Texas lessons are becoming the template.
How Revenue Stacking Works in ERCOT
ERCOT clears six products that a BESS can bid into, and the economics of merchant storage in Texas is the economics of sequencing across those products. The day-ahead and real-time energy markets are the obvious layer: charge when nodal LMPs are low, discharge when they are high, earn the spread. Four ancillary service products layer above energy: RegUp and RegDown (frequency regulation, cleared every five minutes, fast ramp), Responsive Reserve Service (RRS) (frequency response within seconds of a disturbance), Non-Spinning Reserve (NonSpin) (10-minute response), and ERCOT Contingency Reserve Service (ECRS) (a 10-minute-response product launched in June 2023 that now dominates the profitable end of the stack).
| ERCOT product | Response / duration requirement | Typical 2023 contribution to merchant BESS revenue | Typical 2025 contribution to merchant BESS revenue | Software-layer requirement |
|---|---|---|---|---|
| Day-ahead & real-time energy | Requirement: Nodal LMP clearing; 5-minute real-time intervals. | 2023: ~50–60% of merchant BESS revenue. | 2025: ~25–35%. | Software: Price-forecasting pipeline; day-ahead and real-time bid engines; nodal LMP routing. |
| RegUp / RegDown | Requirement: 5-minute dispatch; fast ramp. | 2023: ~15–20%. | 2025: ~10–15%. | Software: Sub-5-minute dispatch telemetry; continuous regulation signal processing. |
| Responsive Reserve Service (RRS) | Requirement: Sub-second frequency response. | 2023: ~15%. | 2025: ~10%. | Software: Frequency-triggered autonomous dispatch; fast under-frequency relay logic. |
| ERCOT Contingency Reserve Service (ECRS) | Requirement: 10-minute response, sustained output; launched June 2023. | 2023: ~5–10% (ramp from June launch). | 2025: ~30–50% for top-quartile operators. | Software: Guaranteed sub-10-minute response from any allowed SOC window; demonstrated-reliability tracking against ERCOT award records. |
| Non-Spinning Reserve (NonSpin) | Requirement: 10-minute response. | 2023: ~5–10%. | 2025: ~5–10%. | Software: Reserve-commitment scheduler; state-of-charge reservation logic. |
| Revenue delta: energy-only vs. co-optimized (all six products) | — | 2023: +20–30% multi-market uplift. | 2025: +35–50% multi-market uplift on identical hardware. | Software: Mixed-integer co-optimization; joint SOC path solver across product markets. |
What makes ERCOT distinct is the interaction between products. Unlike markets with deep capacity payments, merchant storage in Texas earns most of its revenue from ancillary service awards and short windows of energy scarcity. An operator that runs a battery purely for day-ahead energy arbitrage captures the baseline spread and nothing else. An operator that co-optimizes across energy + RegUp/RegDown + RRS + ECRS captures revenue across every operating hour that the battery has capacity and state of charge to spare. That is the delta. It is structural, and it is software.
The trading discipline required to run this co-optimization is non-trivial. Every bid into an ancillary product reserves state of charge that cannot be deployed for energy. Every real-time energy dispatch consumes state of charge that cannot back an ancillary commitment. A naïve stack that bids the same battery into multiple products in parallel, without the joint optimization that reconciles them, gets caught short and pays performance penalties that erase the revenue gain. The ERCOT market is unforgiving to operators whose software cannot solve that joint problem at dispatch-clearing speeds.
ECRS Is Now the Profitable Product — Here’s Why
When ERCOT launched ECRS in June 2023, it introduced a 10-minute-response product with a procurement pattern that short-duration batteries are uniquely positioned to serve. Two-hour and four-hour BESS assets that met the response requirement were, effectively, the lowest-cost qualifying resource in the market. Clearing prices rose steeply through summer 2023 and 2024 as ERCOT procured larger ECRS volumes under tightening reliability conditions. By 2025, ECRS alone was routinely contributing 30–50% of total merchant BESS revenue for top-quartile operators in Texas.
The structural reasons are worth understanding, because they determine which operators can actually capture the product. ECRS rewards fast, sustained response (the battery must deliver within 10 minutes and hold output) and forecastable availability (bids are evaluated against the operator’s demonstrated ability to deliver, not against cell specs alone). An operator whose dispatch stack cannot guarantee sub-10-minute response from any allowed state-of-charge window cannot bid competitively into ECRS. An operator whose settlement layer cannot clean reconcile ECRS deliveries against awarded MW is visible to ERCOT as an unreliable resource and gets deprioritized in subsequent procurement rounds.
The result has been a pronounced sorting of the ERCOT BESS operator landscape. Operators who invested early in dispatch latency, state-of-charge forecasting, and ancillary-award processing are clearing ECRS at volumes that have visibly outperformed energy-only peers on identical hardware. Operators who treated ECRS as a marginal add-on to an energy-first dispatch strategy have systematically underperformed the product’s economic opportunity.
Why the Software Stack Determines Which Projects Win
A merchant BESS in ERCOT is a software business. The cells are procured from the same tier-one suppliers every operator uses; the inverters are commodity-grade; the interconnection path is the same. What varies, project to project, is the control stack that turns physical capacity into bid-ready revenue across the six-product market. Four capabilities separate top-quartile operators from the rest in Texas.
Dispatch latency. ERCOT real-time markets clear every five minutes; ancillary products require sub-10-minute or sub-seconds response depending on the product. A BESS whose software stack takes 30–60 seconds to translate an ECRS award into an inverter command is structurally excluded from the fastest-paying product.
Price and reserve forecasting. ERCOT nodal prices and ancillary clearing prices are volatile, and bid strategies depend entirely on forward price views. Operators running multi-source, short-horizon price forecasts bid more sharply; operators running stale or single-source forecasts systematically lose to better-informed counterparties.
Co-optimization across the six-product stack. Bidding the same battery into energy and three ancillary products without joint optimization is how operators win both and then fail to deliver one of them. The optimization problem is a mixed-integer program under uncertainty; the software that solves it at dispatch-clearing speeds is the moat.
State-of-charge management across the operating day. The highest-paying hours in ERCOT are concentrated in afternoon peaks and event windows. An operator whose software discharges the battery into the morning ramp because the energy-arbitrage model said so no longer has capacity to bid into the afternoon ancillary market when prices spike. The operators who win the afternoon hours optimize across the market day, not the market hour.
The protocol and integration layer that connects all of this to ERCOT’s market systems, to the EMS, and to the BMS is the same layer that is converging around IEEE 2030.5 for utility BESS programs elsewhere in the US, a topic the companion Wattsmart piece in this sprint examines in detail.
For operators weighing build-versus-buy on the software layer, the lesson from Texas is direct. The aggregator business model and the energy markets and trading platform engineering behind it, explored in the February sprint’s demand response aggregator economics piece, generalize cleanly from DR into merchant storage. The core decision (how deliberately to own the dispatch stack) is the same one that determines capacity market participation outcomes in other ISOs. Texas is just where the feedback loop between software quality and quarterly revenue is tightest.
12 GW Is a Software Challenge Now, Not a Hardware One
ERCOT has done the work of proving that merchant BESS is a viable asset class. Twelve gigawatts of installed capacity, 30+ gigawatts of queue, and three summers of live ECRS data have answered the question the industry was asking in 2022. What that dataset has also made clear is that the spread between top-quartile and bottom-quartile merchant BESS revenue in Texas is consistent, large, and driven by the control stack.
For operators underwriting new projects or refinancing existing ones in ERCOT, the diligence question has already shifted. Cell-level warranties and EPC quality still matter, but they no longer explain the variance in P&L. What explains the variance is dispatch latency on ECRS awards, forecast accuracy on ancillary clearing prices, state-of-charge discipline across the operating day, and the settlement layer that reconciles all of it against ERCOT market awards. Lenders are beginning to ask those questions at financial close; competitive operators are answering them in engineering decisions made 12–18 months earlier.
Texas is running ahead of the rest of the US market in making this visible. CAISO and PJM are following, with their own ancillary products and their own capacity mechanics. The operators whose software stack is built to clear ERCOT’s six-product market at dispatch-clearing speeds are the ones whose next market expansion will be the shortest and cheapest to execute. Hardware, at 12 GW installed in a single ISO, has become the common denominator. The moat is somewhere else.
Frequently Asked Questions
ERCOT installed battery storage capacity crossed 12 GW in Q3 2025, according to Modo Energy’s quarterly tracking, with more than 30 GW of additional capacity in the ERCOT interconnection queue as of early 2026. Texas now operates more grid-connected BESS than any other ISO/RTO globally, and the pipeline is still growing. Most of that installed capacity is two-hour or four-hour duration, positioned to serve both short-duration ancillary products and afternoon peak energy arbitrage.
ERCOT Contingency Reserve Service (ECRS) is a 10-minute-response ancillary product that launched in June 2023. It procures capacity that can respond to contingency events within 10 minutes and sustain output for a required duration. Two-hour and four-hour BESS assets are well-positioned to qualify, provided their dispatch stack can guarantee sub-10-minute response from the state-of-charge window they bid into. Qualification depends as much on demonstrated software-layer reliability as on cell specifications, because ERCOT evaluates bids against the operator’s demonstrated ability to deliver awarded MW, not against nameplate ratings alone. Since launch, ECRS clearing prices have risen materially and the product now contributes a large share of merchant BESS revenue for top-quartile operators.
A BESS in ERCOT earns revenue across six cleared products: day-ahead and real-time energy (nodal LMP arbitrage), Regulation Up and Regulation Down (frequency regulation), Responsive Reserve Service (fast frequency response), Non-Spinning Reserve (10-minute response), and ERCOT Contingency Reserve Service (10-minute response, launched June 2023). Operators running co-optimized dispatch across all six products capture materially more revenue per installed megawatt-hour than operators running energy arbitrage alone. The revenue uplift from multi-product dispatch has routinely been 30-50% on identical hardware since ECRS launched, based on Modo Energy benchmarks, and is driven by the ancillary-product layer rather than by energy spreads alone.