Why Capital Prefers a Power Plant To a Megawatt Saved

America is entering another power-building cycle. After roughly 15 years of nearly flat electricity use, U.S. demand grew an average of 2.1% a year over the past five years, according to the Energy Information Administration, with data-center servers a major reason growth is expected to continue.

Utilities, technology companies and investors are responding with plans for new generation and transmission. But there is a quieter capital-allocation problem on the other side of the meter: finance is much better at underwriting a megawatt produced than a megawatt avoided.

Put two projects before a credit committee. One generates power and sells it under a long-term contract. The other avoids the same amount of demand through better motors, cooling, insulation or controls.

The power project has an asset, a meter, a buyer, a price and a contract. The efficiency project produces something harder to own: an energy bill lower than it otherwise would have been. That does not make generation safer. It makes contracted generation more legible to capital.

The financing problem has two parts. Payment security determines what a lender can enforce. Performance evidence determines whether promised savings actually appeared and persisted. Solving the first can make an efficiency project lendable without solving the second.

A saving cannot simply be read from a meter. It has to be calculated against what consumption would otherwise have been, adjusting for production, weather, occupancy or operating hours. Even after savings are verified, the money remains inside the customer’s broader cash flow. Unless a contract creates a separate payment obligation, a lender cannot repossess a kilowatt-hour that was never consumed.

A conventional corporate loan sidesteps much of the problem by underwriting the company rather than the saving. That works when the borrower has room on its balance sheet. It works less well when an efficiency project must compete with factories, acquisitions, inventory and every other demand on the same borrowing capacity. Positive NPV is not a financing structure.

I saw the distinction repeatedly in earlier work on more than 200 industrial energy-efficiency projects. Equipment could perform as promised and energy use could fall while investment returns deteriorated because operating constraints, lifecycle costs or later interventions had not been fully priced at the start. Engineering performance and financial performance are related. They are not the same thing.

Commercial Property Assessed Clean Energy, or C-PACE, shows how much financing can change when payment is made easier to enforce. PACENation reports $13.5 billion of cumulative commercial PACE investment through 2025 across 3,837 projects, with 55% classified as energy efficiency. Repayment runs through a property assessment created under state and local programs.

That structure can make payment secure without making the underlying energy saving easier to observe. Berkeley Lab reported in 2024 that relatively few C-PACE programs require robust measurement and verification compared with traditional energy-savings performance contracts or utility programs.

Capital markets already show the distinction. In Calvert Impact Climate’s 2023-1 C-PACE securitization, the Class A notes received an AAA rating. The published rating rationale emphasized credit enhancement, property values, lien priority, legal structure and stressed cash flows. Calvert separately reports project-level utility monitoring of actual consumption and emissions, but whether the financed equipment delivered its projected energy savings was not among the listed rating considerations.

That is not a criticism of the rating. It is the point. Finance has become very good at making repayment claims legible. It has far less standardized information for making the operating performance beneath those claims equally legible.

This can create a capital-allocation bias. The more easily a cash flow can be isolated, contracted and pledged, the more naturally it fits existing financing machinery. An avoided cost can be economically valuable and still remain difficult to separate from the company that produced it.

Two changes would help.

First, companies should preserve a standardized version of the original investment case: expected operating conditions, lifecycle costs, baseline assumptions and the alternative that lost when the project was approved. Second, they should use post-project verification proportionate to the size and complexity of the investment. Measuring a small retrofit should never cost more than the information is worth.

Better evidence will not turn every retrofit into a standalone security, nor should it. It would do something more basic: help lenders and investors distinguish weak projects from good projects whose value is simply harder to isolate, while gradually building the performance record needed to price outcome risk more intelligently.

As America spends heavily to meet rising electricity demand, financing architecture will influence what gets built and what gets ignored. The power plant arrives at the credit committee as a financial asset. The saved megawatt still too often arrives as an argument.

Closing that gap is not only an energy-efficiency problem. It is a capital-allocation problem.