How Do Power Purchase Agreements Work? PPAs Explained

A power purchase agreement, or PPA, is a long-term contract in which a buyer agrees to pay a fixed or formula price for the electricity, and often the renewable or carbon-free attributes, from a specific power plant, typically for 10 to 25 years. The contract exists mainly so the plant can be financed: lenders will fund a wind farm, solar farm or reactor restart against a creditworthy buyer's promise to pay, but not against the hope of good market prices. Who carries the risk of price, volume and timing depends on whether the PPA is physical or virtual and on how the contract is drafted.

The PPAs that made headlines in 2024–26 were not the renewable deals of the previous decade. In June 2025 Talen Energy and Amazon signed a contract for up to 1.92 GW of output from the Susquehanna nuclear plant in Pennsylvania through 2042, which Talen valued at about $18 billion over the term at full volume (company). The deal had started in March 2024 as something else, a $650 million sale of a data-center campus beside the plant, with power delivered behind the meter. When the Federal Energy Regulatory Commission rejected an expansion of that arrangement on 1 November 2024, the parties rebuilt it as a PPA with power delivered over the grid. The contract that could be signed depended on what the grid's rules allowed.

What is a power purchase agreement?

A PPA has a seller, usually a project company that owns one plant, and a buyer, usually a utility, a retailer or a large corporation. The core terms are few:

• What is sold: energy, capacity, renewable energy certificates, or a bundle.
• Price: fixed, escalating, indexed to something, or a fixed price for a defined shape of delivery.
• Term: how many years, and when it starts, usually at the plant's commercial operation date.
• Volume: "as generated" (whatever the plant produces) or a fixed shape (a set amount in set hours).
• Delivery point: where title passes or where the contract settles, which decides who bears congestion.
• Credit and exit: guarantees, collateral, termination rights, delay damages, force majeure and change-in-law clauses.

Most of the economics hide in the last two lines. The book's chapter on offtake, contracts and pricing argues that a PPA's exit terms decide what it is worth in a crisis.

Why do PPAs exist?

Almost every energy asset costs nearly everything up front and earns it back over decades. A lender to such an asset sizes its loan on predictable cash flow, and a PPA is what makes cash flow predictable.

The book's research works through a stylized 200 MW onshore wind farm costing $320 million, with a 15-year fixed-price PPA at $55 per MWh with a creditworthy utility. Lenders size debt so that cash flow covers debt service with a margin even in a bad wind year. The result is about $176 million of debt, roughly 55% of the cost. Change the PPA and the debt changes with it.

Price is the biggest lever, because a lower price comes straight out of the margin over fixed costs. An escalator can raise debt even at a lower starting price, because more cash arrives in later years. The same research finds that the price this project needs to hit its target return rises from about $42 per MWh when money is cheap to about $52 when it is dear, and to about $66 without the production tax credit. A PPA price is, in large part, a cost of capital. Chapter 14 on project finance explains why.

How does a physical PPA work?

In a physical PPA the buyer takes the power. A utility buying from a solar farm in its territory receives the electricity at an agreed delivery point and pays the contract price for each megawatt-hour produced. If the contract is "as generated," the seller carries volume risk (the wind does not blow) and the buyer carries the shape risk (the power arrives in hours when it may be worth less). Physical PPAs are the traditional route for utilities and for buyers that can take delivery, and they are the standard form in places where a state utility buys from independent producers.

How does a virtual PPA work?

Most US corporate buyers cannot take physical delivery from a wind farm hundreds of miles away. They sign a virtual PPA, a financial contract for differences. The generator sells its power into the wholesale market as usual. Then the two sides settle the difference between the fixed contract price and the market price at an agreed trading hub, for each megawatt-hour the plant produces.

A simple stylized example shows the mechanics. Suppose the strike price is $50 per MWh.

Stylized illustration; assumes the plant's own node price equals the hub price.

The generator earns $50 regardless of the market, which is what its lenders want. The buyer gets the renewable attributes and a hedge against high power prices. The last assumption in the table is the catch.

Who carries which risk in a PPA?

Every energy contract decides who absorbs three kinds of uncertainty: price (the benchmark moves), volume (output differs from the forecast) and shape and location (output arrives in hours, or at a node, where it is worth less). The main contract types divide them differently.

Three risks in that table deserve a closer look.

Basis risk. In a hub-settled virtual PPA, the generator sells at its own node but settles at the hub. If transmission congestion pushes its node price below the hub, the generator earns less than the strike. A busbar PPA settled at the project's own node avoids this.

Shape and capture risk. Solar plants all produce at midday, which pushes midday prices down as more solar is built. A buyer settling a solar virtual PPA on output can end up paying the strike most often in exactly the hours when the hub price is lowest.

Fixed-shape exposure. A fixed-shape hedge obliges the seller to deliver set volumes. In Winter Storm Uri in February 2021, wind farms with such hedges that were not running had to buy power at ERCOT's $9,000 per MWh cap. Our case study of the 2021 Texas power crisis shows how forward sales turned into short positions.

The rule that follows is that a risk moved by contract is not removed; it becomes the other side's credit risk. The party best able to carry a risk should hold it, and will charge for doing so.

What can go wrong with a PPA?

A fixed price meets rising costs

A long contract lowers the cost of capital only if the price is high enough to survive the costs that arrive after signing. Ørsted's Ocean Wind 1, about 1,100 MW off New Jersey, won its offtake in June 2019 at a levelized price reported at about $98 per MWh, when the ten-year Treasury yield was about 1.5%. By October 2023 the yield was about 4.5–5%, and turbines, vessels and steel had become dearer. The price was fixed. Ørsted cancelled Ocean Wind 1 and 2 on 31 October 2023 and reported about DKK 28.4 billion of impairments for the year (company).

The seller treats the start date as an option

Most PPAs start deliveries at commercial operation. If the contract does not define completion objectively, with a deadline and damages, it hands the seller an option. Venture Global's Calcasieu Pass LNG terminal produced its first cargo in March 2022 and sold "commissioning" cargoes at spot prices for more than three years before declaring commercial operation in April 2025, while buyers whose 20-year contracts had financed the plant received no contract cargoes (reported). The arbitrations split: Venture Global won against Shell, while a tribunal found it had breached its obligations to BP (reported). The same lesson applies to power: read the exit before the price.

The buyer's credit is the contract's value

A 20-year price is worth the balance sheet of whoever pays it. A regulated utility whose costs pass to customers, or a technology company with a large balance sheet, is a strong buyer. A start-up offtaker is not.

How are data-center PPAs different?

The renewable PPAs of the 2010s bought the cheapest energy available whenever the wind blew or the sun shone. The data-center contracts of 2024–26 bought firmness and speed: power that runs every hour, from a plant that already exists or can be brought back quickly.

None of the prices was disclosed. Two cautions apply. Many announcements were frameworks, "up to" several gigawatts, which become offtake only when a PPA is signed for a named plant at a named price. And a PPA from an existing plant reallocates output rather than adding supply; the grid still needs new capacity for the new load, which is why PJM's capacity price rose so far, as our explainer on the PJM capacity auction describes. The book's study of data-center and AI power demand tracks these deals in detail, and company detail is in the Constellation Energy and Brookfield Renewable profiles.

PPAs versus contracts for difference

Europe built much of its renewable capacity under a variant in which the state is the buyer. Britain's contract for difference pays a generator the gap between a fixed strike price and the market price when the market is below the strike, and takes money back when it is above. In the round announced on 14 January 2026, about 8.4 GW of fixed-bottom offshore wind won contracts at £91.20 per MWh in 2024 prices (reported). The structure moves price risk to a government-owned counterparty and, through it, to consumers, which lowers the developer's cost of capital. A US virtual PPA is the private-sector version of the same instrument.

What to check in a PPA term sheet

• Is it physical or virtual, and where does it settle: the busbar, the node or a hub?
• Is the price fixed, escalating or indexed, and who bears the difference between the plant's profile and the buyer's load?
• Who bears curtailment, negative prices and congestion?
• What is the commercial operation date, what are the delay damages, and can the buyer walk away if the plant is late?
• What availability or production guarantees does the seller give?
• What credit support stands behind each side, and what triggers collateral calls?
• What happens on a change in law or the loss of a tax credit?

How PPAs fit the power equation

The Power Equation holds that an energy company creates durable value only when four factors hold at once: supply at a cost the market will pay, a route to market, capital matched to the asset, and a licence to operate. A PPA is the purest form of the second factor, the route to market. It turns a plant's output into revenue before the plant is built, and in doing so it unlocks the third factor, because lenders lend against it. A well-drafted PPA with a strong buyer can be worth as much as a technology improvement; the research's nuclear-restart example finds that moving from a 10-year to a 20-year contract cuts the required price by about a fifth. A badly priced or loosely drafted one can sink the project it was meant to protect. For how PPAs fit among the industry's other revenue engines, see our guide to how the energy industry makes money, and for the market structures they sit in, our comparison of regulated vs deregulated electricity markets.

The takeaway

A power purchase agreement fixes the price of a plant's output for long enough that lenders will finance it. Physical PPAs deliver the power; virtual PPAs settle the difference against a hub price and leave basis and shape risk with someone. The contract's value depends less on the headline price than on the buyer's credit, the risk allocation and the exit terms, as Ørsted's fixed-price wind contracts and Venture Global's commissioning cargoes both showed. Read the exit before the price, and credit the buyer, not the contract. Chapter 13 of The Power Equation sets out how an expert reads an energy contract.