The Cannibalization Problem: Why Solar's Success Makes Solar Harder to Finance

The cost story of the last fifteen years is one of the great successes in energy history. Utility-scale solar module costs have fallen by roughly 90 percent since 2010, and in most sunny markets a new solar farm now produces electricity more cheaply than a new gas plant. By the logic of the levelized cost comparisons that dominate energy commentary, the argument should be over. And yet developers in the highest-penetration markets — California, Chile, South Australia, parts of Spain — are finding projects increasingly difficult to finance. The obstacle is not the cost of building. It is the revenue.

The mechanism is straightforward once you look at how wholesale power markets clear. Generators bid into the market roughly in order of their marginal cost, and the last unit needed to meet demand sets the price for everyone. Solar and wind have a marginal cost near zero, so they bid at the bottom and push more expensive plants out of the stack. This is exactly what a well-functioning market should do — but it means that solar suppresses prices precisely during the hours when solar is generating. As penetration rises, the average price a solar farm captures falls below the market average, and keeps falling. In California's midday hours, prices now routinely go negative, meaning generators pay to deliver power. Solar is not competing with gas so much as it is competing with itself.

This is where the standard analytical tool becomes actively misleading. Levelized cost of energy divides lifetime cost by lifetime output and produces a tidy dollars-per-megawatt-hour figure, but it treats every megawatt-hour as interchangeable. Electricity delivered at 1 p.m. in June and electricity delivered at 7 p.m. in January are not the same product, any more than a hotel room in August is the same product as the identical room in February. Comparing solar's LCOE to a gas plant's LCOE compares the per-unit cost of two goods that are not substitutes. The more honest metrics — value-adjusted LCOE, or system LCOE, which incorporate capture prices and the integration costs imposed on the rest of the fleet — tell a considerably less triumphant story, and they are almost never the numbers that make it into press releases.

Storage is the obvious answer, and it is a real one, but it does not escape the same logic. Battery revenue depends on the spread between cheap and expensive hours; deploy enough storage and the spread narrows, eroding the arbitrage that justified the investment. Each successive tranche of batteries is worth less than the one before it, for the same structural reason each successive solar farm is. The practical response in most markets has been to move revenue off the spot market entirely — contracts for difference, capacity payments, long-term corporate power purchase agreements — which stabilizes returns by shifting merchant risk onto ratepayers, taxpayers, or corporate offtakers. That is a defensible policy choice. It is also a redistribution of risk that rarely gets described as one.

The broader point is that the remaining difficulty of decarbonizing electricity is no longer primarily a technology cost problem. It is a market design problem. Energy-only markets were built for a fleet of dispatchable plants whose costs were dominated by fuel, and they reward availability at the moment of scarcity. A fleet dominated by zero-marginal-cost, weather-dependent capital assets inverts that structure: nearly all the cost is incurred up front, and scarcity becomes a matter of hours and seasons rather than fuel prices. Markets that were designed to send the right signals under the old arrangement send perverse ones under the new. Until that redesign is taken as seriously as the cost curves are celebrated, cheap solar will keep arriving faster than the institutions capable of paying for it.