Summary

  • Google and RWE signed a 15-year power purchase agreement for the entire output of the planned 155MWac Crooked Creek Solar project in McCurtain County, Oklahoma.
  • RWE describes the project as its first solar development in Oklahoma and says it plans to begin onsite construction later in 2026, with commercial operation expected in 2028.
  • The project is intended to support Google’s operations in the Southwest Power Pool, but the release does not identify a data centre, disclose Google load or prove a dedicated physical delivery path.
  • RWE estimates up to 250 jobs at peak construction, $24.3 million of local economic activity and $1.4 million of state and local tax during construction.
  • The company expects about $25 million over the operating life for McCurtain County programmes, emergency services, Broken Bow School District and regional technology centres; these are estimates, not realised payments.
  • No PPA price, annual MWh, hourly production profile, interconnection milestone, financing, curtailment allocation, renewable-certificate treatment or 24/7 carbon-free matching result was disclosed.

The PPA connects two future cash flows

A solar developer needs confidence that somebody will buy output after the plant is built. A large power user needs a defined commercial claim on generation that can support energy-cost planning and environmental procurement. The 15-year agreement links those two needs before Crooked Creek reaches operation.

For RWE, a long-duration buyer can reduce merchant-price exposure and improve the project’s financing case. That does not prove financing has closed; lenders still assess interconnection, construction, equipment, counterparty and operating assumptions. For Google, the agreement can secure the economic attributes of a renewable project for a long period. It does not reveal the price, settlement structure, minimum generation guarantee or who bears shortfall and curtailment risk.

The contract is therefore a risk-transfer instrument, not a delivered-energy certificate. Its value lies in changing who absorbs future market and production outcomes. Until the commercial terms are disclosed, outsiders can identify the duration and project size but cannot calculate the hedge value, cost per megawatt-hour or revenue certainty.

155MWac is plant capacity, not an energy volume

The announced 155MWac figure describes the alternating-current capacity of the project. It is a power rating: a maximum instantaneous output under relevant conditions at the AC boundary. It is not the electricity generated over a year. Annual energy would be measured in megawatt-hours and depends on sunlight, design, losses, outages, degradation and curtailment.

Nor is 155MWac the size of Google’s load. The release says the project will support Google operations in the Southwest Power Pool but does not identify a named data centre or publish its demand. A data centre draws power continuously and may have a relatively flat load; a solar plant produces according to daylight and weather. Equal-looking megawatt figures would still not establish equal hourly supply.

“Entire output” defines the contracted share of Crooked Creek generation. It does not mean all electrons travel exclusively to one Google building. In an interconnected grid, physical flows follow network conditions. A PPA can be settled financially or through market arrangements while renewable-energy attributes are assigned contractually. The release does not specify the node, delivery arrangement or certificates, so a stronger physical claim would be invented.

The 2028 date is an execution clock

RWE says it plans to begin onsite construction later this year and expects commercial operation in 2028. One later sentence in the English announcement calls the project “under construction”, creating internal shorthand that conflicts with the explicit future-start statement. The conservative boundary is the more specific one: construction is planned, not established as commenced by the captured release.

Signature sits at the start of a conversion chain. The project needs final development work, equipment, contractors, civil construction, an interconnection path, testing, energisation and commercial acceptance. A target year compresses those dependencies into one number, but each can move independently.

The next credible milestones are physical and regulatory: a notice to proceed, mobilisation, major equipment delivery, interconnection progress, first energisation and commercial-operation notice. Treating the 2028 expectation as guaranteed would move every remaining risk into the past before any source has done so.

Grid connection decides whether contracted generation can become market output

Crooked Creek will participate in the Southwest Power Pool context cited by RWE. The transmission system has to accept the project’s output at the relevant point, and network constraints can limit when generation reaches the market. The announcement does not disclose queue position, upgrades, study status, network costs or who pays them.

Those omissions matter to both parties. An interconnection delay can postpone commercial operation even if panels are installed. A required transmission upgrade can change capital cost. Curtailment can reduce delivered energy when the plant is technically available but the grid cannot take all production. The PPA may allocate some of these risks, but its terms are private.

Google’s demand and the project’s settlement can also sit at different locations inside SPP. Price differences between nodes create basis risk: project revenue and load cost may not move together. A long-duration contract can manage that risk only to the extent its settlement design addresses it. Project capacity alone says nothing about the basis.

Annual renewable procurement is not hourly carbon-free operation

Solar output is concentrated in daylight. Google infrastructure can run every hour. A contract for all annual output may support a renewable-energy accounting position or offset a quantity of consumption over a period, yet leave nights and low-sun hours dependent on the wider grid mix.

This is the distinction between volumetric and temporal matching. Annual megawatt-hours answer whether a volume of renewable generation corresponds to a volume of consumption over the year. Hourly matching asks whether carbon-free supply aligns with demand in each hour and location. The release makes no hourly claim and supplies no load profile, storage, complementary wind, dispatchable resource or matching result.

The PPA can still be consequential. It can enable additional generation and create a durable demand signal. But readers should not translate “supports Google’s operations” into “powers them 24/7 with solar”. That would require a portfolio and measurement framework beyond one daytime resource.

Local benefits arrive on different schedules

RWE estimates as many as 250 jobs at peak construction. Peak is not the average number of workers, the total number of unique workers or permanent employment after operation. The company also estimates $24.3 million of local economic activity and $1.4 million in state and local tax during construction.

Over the operating life, RWE expects about $25 million to support McCurtain County programmes, emergency services, Broken Bow School District and regional technology centres. That figure spans an undisclosed annual schedule and depends on the project reaching operation and remaining in service. It is not cash already received by public bodies.

The distinctions help local accountability. Construction employment, construction-period spending and long-term fiscal contribution have different denominators. Reporting them together as a single immediate benefit would make a future project look like a completed local revenue stream.

RWE’s Oklahoma footprint provides context, not delivery proof

Crooked Creek would be RWE’s first solar project in Oklahoma. The company already operates the 148MW Boiling Springs wind farm and says six additional Oklahoma projects totalling 1.6GW are in development. That experience can indicate a local development platform and familiarity with the market.

A pipeline is not operating capacity. The six projects are described as development assets, and their combined 1.6GW cannot be added to Crooked Creek as if all were producing. Boiling Springs is the cited operating asset; Crooked Creek is contracted and planned; the other projects remain in development. The status labels are more informative than the sum.

Google’s role similarly establishes a buyer, not a builder or grid operator. RWE remains the project developer in the release. The PPA can improve bankability, but RWE and its contractors must execute the physical project and the grid must accommodate it.

Evidence will accumulate before the first commercial megawatt-hour

The agreement has crossed a meaningful gate: a long-term buyer has contracted the project’s output. That lowers one category of uncertainty and makes the project more legible to capital providers. Several others remain.

Watch for construction start, financing, equipment delivery, interconnection status, any schedule revision and commercial operation. After operation, annual MWh, curtailment, availability and settlement performance can show whether the contracted output materialised. Google would need separate hourly and location-matched data to make a stronger 24/7 claim.

Until then, the narrow conclusion is also the useful one. Google and RWE have connected future demand to a planned 155MWac solar asset for 15 years. They have not disclosed the price, delivered energy or hourly match, and the plant is not established as operating. The PPA is a bridge from development to possible generation; it is not the far bank.

Sources