How U.S. wind and solar hedge power prices in 2026

In 2019, energy researcher Jay Bartlett pointed to Phoebe Solar in West Texas as an example of a project that could contract most of its power and still sell the remainder on the wholesale market. Two winters later, the project showed why the percentage contracted tells you only so much. Its hedge called for predetermined deliveries, and Winter Storm Uri brought supply interruptions and extreme prices at the same time.

Phoebe sold its actual power at the plant’s node and financially bought the contracted hourly quantity at a hub before receiving its fixed price. Innergex reported a US$24 million cash disbursement in July 2021, covering disputed amounts and a contract account payment, net of unpaid power sales. It separately wrote down the project by C$24.7 million, citing higher expected congestion costs and a higher discount rate. In 2024, Innergex reported paying US$54.5 million to terminate the hedge as part of a Texas portfolio transaction that moved Phoebe to merchant power sales. Settlement mechanics · 2021 cash settlement · 2021 impairment · 2024 annual report.

By 2025, the United States generated nearly five times as much utility-scale solar electricity as it did in 2018. Batteries can shift some of that energy out of the midday hours. Today a buyer can contract for the energy, the right to dispatch the battery, or a share of its market revenue. The question Bartlett posed still works: which risk moves to the other side of the deal, and which stays at the project?

From 2018 to 2025, U.S. wind generation grew 69%, and utility-scale solar generation rose to 4.7 times its 2018 level. Through August 2026, wind was up 7.8% and utility-scale solar 20.0% against the same months in 2025. Those are metered and estimated generation totals from EIA’s Electric Power Monthly, across projects with all kinds of contracts and ownership.

A plant earns the price in the hours when it generates. To calculate its energy capture price, multiply the plant’s output in each interval by the local wholesale price, add the dollars, and divide by total MWh. A simple market average gives a quiet midday hour the same weight as an evening peak. A solar plant cannot make that substitution. When many plants produce at once, they can lower the price in the very hours they depend on. Congestion can push one plant’s price further below the regional hub.

Through September 2026, the solar-weighted price at California’s SP15 hub was 52% of the flat price. At SPP North it was 127%. At ERCOT North, the flat hub price rose from $33.08/MWh to $35.06/MWh against the same months of 2025, while the solar-weighted price fell from $27.11/MWh to $24.40/MWh. A view on where average prices are heading can miss the price at which a generator actually sells. These are hub-price comparisons using system-wide wind and solar output. A plant’s own location and generation can produce a different result.

Separately, Berkeley Lab estimates that in 2025 U.S. utility-scale solar delivered $41/MWh in wholesale energy and capacity value, or 82% of the value of a flat block across its sample. It found $23/MWh in ERCOT and $62/MWh in SPP. That calculation includes capacity where applicable and is a separate measure from the hourly energy comparison above. Berkeley Lab, 2026 solar update.

SP15 was below zero in 15.4% of day-ahead hours, but those hours coincided with 33.5% of CAISO’s reported solar output in the same 2026 window. An exclusion triggered by a negative reference price can touch much more generation than the hour count suggests. Curtailment is a different risk: ERCOT’s monitor estimated 2025 output reductions of about 6% for solar and 4.7% for wind. A contract can treat power produced at negative prices differently from power that never reached the grid. ERCOT market monitor.

A merchant generator sells into the wholesale market. It may leave that price exposure open or offset it with a financial contract. A physical power purchase agreement, or PPA, pays for delivered power. With a virtual PPA, the project still sells the electricity into the grid at its node and settles a separate price difference with the buyer. If the contract references a hub, the project keeps the difference between its own price and the hub price. That difference is called basis.

A fixed block gives the buyer a predictable contract quantity. The project’s actual output can fall short in an expensive hour, even if it meets its annual production forecast. An as-generated swap follows output and avoids that fixed-quantity mismatch, but low output still cuts revenue. A proxy swap uses modeled output from observed weather, so weak wind can be covered while a broken turbine remains the project’s responsibility. Curtailment, negative-price thresholds, timing of settlements and exceptions are all matters of contract language.

A congestion right can help with a specified path. In ERCOT it settles against day-ahead congestion for its source, sink and quantity. It does not cover the project’s real-time price versus day-ahead settlement, or the cost of a volume shortfall. The independent market monitor also notes that the West load zone combines an export-constrained Panhandle and an import-constrained Permian Basin. A zone average can hide both exposures. ERCOT market monitor, 2025.

In September 2026, RWE announced a 433 MW virtual PPA with Oracle from three operating Panther Creek wind farms in West Texas. The announcement names the capacity and projects but gives no strike, settlement location or contracted hourly volume. Compare that with Capital Power’s public filing: at the 178 MW Bloom Wind project in Kansas, its counterparty pays a fixed amount in exchange for modeled weather-linked revenue through June 2027. In October, CME listed futures on ERCOT regional wind generation indexes. Those contracts can address regional output risk, with differences between the index and the plant left open. RWE announcement · Capital Power filing · CME listing circular.

Capital Power describes two contracts on its 101 MW Buckthorn wind project in Texas. One is an as-generated agreement for 55% of output, including energy and renewable energy certificates. The other is a fixed-notional swap sized to cover roughly 85% of the remainder. Both use ERCOT North Hub. The two contracts address different amounts and create different volume exposures, while both leave the project’s price difference to North Hub in play. Capital Power filing.

U.S. utility-scale batteries reached 55.6 GW of power and 152.9 GWh of energy capacity in September 2026, up from 0.9 GW and 1.2 GWh at the end of 2018. Operating capacity, including stand-alone batteries, is preliminary for September. ERCOT changed how batteries compete for revenue in December 2025 when it began clearing energy and ancillary services together in real time. A contract priced on earlier results has to account for that change. EIA generator inventory · EIA 2018 battery survey · ERCOT market change.

In July 2026, ContourGlobal and Tesla announced a long-term PPA for about 1 TWh per year from Arizona’s planned Sterling project: 450 MW of solar and a 360 MW / 1.4 GWh battery, with firm transmission rights into the California ISO market. The parties describe the contracted volume as roughly 90% of as-generated output and do not publish an hourly schedule. In Texas, GridStor and Axpo announced a separate 100 MW battery revenue swap backed by the operating 220 MW / 440 MWh Hidden Lakes battery. At the smaller Cross Trails battery, Energy Vault describes a ten-year offtake with a revenue floor and a share of upside. A buyer can contract for renewable production, battery market revenue, or downside protection. The agreements describe different things. ContourGlobal · GridStor · Energy Vault.

The battery changes the hours a project can sell. Its power rating limits how fast it can discharge; its energy rating limits how long. Dispatch also depends on how full it is, the rules for charging and selling, and how much cycling the contract allows. A Berkeley Lab study of 51 hybrids found that their observed 2024 dispatch captured 62% of the study’s modeled optimized wholesale market value. Contract payments and operating profits are separate from that estimate. Berkeley Lab hybrid dispatch study.

Bartlett classified 3,381 MW, or 48% of U.S. wind additions in 2017, as merchant or the merchant portion of mixed projects. His definition included financially hedged projects. The preliminary 2018 figure was 49%. Counting the entire capacity of any project with a merchant portion would instead give 54% and 55%. Bartlett, Resources for the Future, 2019, pp. 13–15.

Berkeley Lab’s 2026 land-based wind update asks who initially bought the output. It classifies none of the 2025 additions as merchant or quasi-merchant, down from about 20% of 2017 additions in the same dataset. But a direct retail buyer can sign a virtual PPA with a generator that sells its physical power on the wholesale market. Bartlett counted those projects as merchant. And for about 11% of the 2025 capacity, the offtaker was undisclosed. The figures show a shift in disclosed routes to market; they cannot tell us how much price, volume or basis risk is still open.

Under the 2025 federal law, wind and solar facilities that both begin construction after July 4, 2026, and enter service after December 31, 2027, generally lose access to the technology-neutral production and investment credits, sections 45Y and 48E. Projects that met the construction deadline still have to satisfy continuity and other eligibility rules. Storage has its own credit treatment and is excluded from this accelerated wind and solar cutoff. Eligible developers can sell credits for cash under section 6418; that financing cash does not settle against the project’s hourly electricity price. IRS construction guidance · Section 48E · IRS transferability rules.

1. Where does it settle? Ask for the node, hub or index and the market interval.
2. What quantity settles? Metered generation, a fixed shape and modeled production behave differently in a bad hour.
3. Who pays when output falls? Read the rules for outages, curtailment, negative prices and force majeure.
4. What can the project owe before it gets paid? Check collateral, settlement timing and the cash needed during a price spike.
5. What is still exposed over time? Count the uncovered output, the years after the contract ends, and any battery dispatch rights the project retains.

The hub comparison matches 6,552 hourly day-ahead prices per hub and year from ERCOT, CAISO and SPP to EIA Form 930 generation for the associated balancing authority. Both windows run January 1 through September 30 in UTC. Negative net generation receives zero weight, and EIA’s separately reported hybrid output is excluded. SPP’s market expanded west in April 2026; the generation series here is for SPP East in both years. The comparison measures energy only, without capacity, renewable energy certificates, congestion to individual plants or contract cash flows.

This story updates the U.S. risk framework in Jay Bartlett’s Reducing Risk in Merchant Wind and Solar Projects through Financial Hedges, Resources for the Future, February 2019 (CC BY 4.0). The reporting, contract examples, market analysis and illustrative settlement model are new. The evidence cutoff is November 11, 2026. National generation totals for 2026 cover January through August. Disclosed terms come from company announcements and filings. The worked settlements represent a hypothetical hour and exclude contract-specific adjustments, fees and taxes.