The electricity that never gets made

France’s 2025 electricity review contains two figures that appear to describe the same problem.

Around 3 TWh of wind and solar output was not produced during periods of negative electricity prices.

In the same year, France’s transmission operator ordered 0.1 TWh of wind and solar generation to reduce output as part of its real-time balancing work.

Three terawatt-hours. One-tenth of a terawatt-hour.

The thirtyfold difference is not a mistake. It marks the boundary between two events often described with the same word.

In the first, generators responded to a market telling them another unit of electricity had no buyer at that hour. In the second, the system operator instructed generation to move downward to keep supply and demand balanced safely.

Both left usable wind or sunlight unconverted into electricity. They did so for different reasons.

RTE’s review of French generation in 2025 separates the two. It estimates that 1.3 TWh of onshore wind and 1.6 TWh of solar output was modulated during negative-price periods, while downward adjustments requested directly by RTE accounted for 0.1 TWh.

Public discussion tends to call all of this curtailment. That is understandable. The equipment was available, but less electricity reached the system.

For anyone trying to understand what went wrong, the distinction matters. Renewable output can fall because a local network cannot carry it, because the wider system has more supply than demand, because operational stability requires a change, or because the market price makes continued generation uneconomic.

The missing megawatt-hour looks the same on the final meter. Its cause determines what might have saved it.

One missing megawatt-hour, several explanations

The International Energy Agency uses technical curtailment for renewable output dispatched downward for network or power-system reasons. It excludes generation reduced because of economic or market conditions. Its Renewables 2025 report sets out that distinction.

A local congestion problem arises when a wind or solar farm can generate but the lines carrying power away have reached their limit. Demand may exist elsewhere; the network cannot safely deliver another megawatt from this point.

A system-wide surplus occurs when total generation exceeds demand. Supply and demand must remain balanced continuously, so generation has to fall if demand cannot rise, exports cannot increase, and storage cannot absorb the difference.

A system-stability instruction may be needed even when annual or national energy balances look comfortable. Operators manage frequency, voltage, reserve, and rapidly changing flows in real time.

Then there is economic modulation. A generator may reduce output because the wholesale price has reached zero or fallen below it. No system operator needs to issue an order; the price signal makes another megawatt-hour unattractive under the generator’s contracts and support arrangements.

France experienced 513 hours of negative spot prices in 2025, up from 352 in 2024. RTE links the increase to rising solar generation, relatively low consumption during parts of the day, and more renewable installations able to respond to prices.

A negative price does not mean electricity has acquired negative physical value. It means the market has more supply than buyers want at that location and moment, after accounting for storage, exports, flexible consumption, and network limits.

Curtailment is therefore not simply a story about the grid refusing renewable electricity. Sometimes the market rejects it first.

The grid has geography

Electricity can be abundant and unavailable at the same time.

One region may have strong wind, low local demand, and limited transmission. Another may have high demand and expensive generation. The physical network decides how much electricity can move between them.

In 2024, European transmission operators carried out remedial actions involving 60 TWh to manage congestion, at a cost of €4.3 billion. ACER’s 2025 monitoring report says delayed reinforcement is widening the gap between the network Europe has and the one its changing power system requires.

That 60 TWh should not be read as 60 TWh thrown away. Remedial action can involve reducing generation on one side of a constraint and increasing it elsewhere, changing cross-border flows, or taking other steps to keep the system safe. The volume measures intervention, not a pile of missing energy.

It does show how much work operators perform after the market schedules electricity along paths the network cannot fully support.

Wind and solar make the geography visible because their best generating locations do not necessarily match concentrated demand. New generation can sometimes be built faster than the lines needed to carry it, while large new loads seek connections in already congested regions.

This is also why data-center power demand is a location problem. Building more generation does not guarantee that a particular load can access it.

Cross-border interconnection, stronger domestic networks, better coordination, and grid-enhancing technologies can release capacity. But grids take time. Permitting, land access, public opposition, equipment shortages, financing, and construction can turn a transmission project into a decade-long undertaking.

The problem is no longer a simple shortage of renewable projects. In many places, it is a shortage of ways to integrate what developers are ready to build.

Some curtailment is rational. Too much is a warning.

Every curtailed megawatt-hour can look like failure. The turbines were available and the wind cost nothing. Why would a sensible system choose less clean generation?

Because capturing the final possible unit also has a cost.

A transmission line built for a rare annual peak may sit partly unused during the rest of the year. A battery large enough to absorb every surplus period may cost more than the energy it saves. Maintaining generation and demand that can respond quickly is part of keeping the system secure.

It can therefore be rational to accept some curtailment rather than build enough wires, storage, and flexible capacity to capture every theoretical megawatt-hour under every possible condition.

The IEA describes some curtailment as expected in systems with growing shares of variable renewable generation. Persistent or widespread curtailment is the stronger warning: it can signal gaps in transmission, flexibility, storage, market design, or coordinated planning.

Zero curtailment at any cost is not a sensible target. The useful questions concern frequency, scale, location, cause, and trend.

How much available generation was reduced? What share of potential output did it represent? Was the constraint local or system-wide? Was the reduction ordered for technical reasons or chosen in response to price? How often does the same bottleneck recur? Would another investment recover enough energy or system value to justify its cost?

Curtailment also affects project economics. A wind or solar installation earns nothing from electricity it does not sell unless its contract or market rules provide compensation. Persistent reductions can lower revenue, weaken further investment, or shift costs toward consumers and public support systems.

The technical reason and commercial allocation are separate questions. One asks why output fell. The other asks who pays for the missing megawatt-hour.

The system has four broad responses

There is no single cure because the causes differ.

The first response is to move electricity through space. Transmission, distribution upgrades, and interconnectors allow renewable output to reach a wider pool of demand.

The second is to move electricity through time. Batteries, pumped hydro, thermal storage, hydrogen, and other technologies operate over different durations and with different costs and losses. A two-hour battery cannot solve a week of strong wind and weak demand.

The third is to move demand. Electric vehicles can charge when electricity is plentiful. Heating, cooling, and some industrial processes can shift consumption within operating limits. The IEA argues that demand flexibility can reduce peaks and absorb more renewable generation, while noting that equipment, software, prices, regulation, and incentives all have to align.

The fourth is to make generation more controllable. Wind and solar farms can participate in balancing markets, respond to system instructions, and modulate output. Better forecasting gives operators more time to prepare.

These responses complement one another. The goal is to identify why electricity was unavailable to the system and apply the response that addresses that cause.

Compute can become part of the demand side

Data centers enter this discussion because they can consume substantial electricity while producing information rather than a physical product.

Some workloads need to run immediately, continuously, and close to users or data. Others can tolerate a later start, temporary reduction, or another location. The available flexibility depends on workload, service commitment, hardware, data location, network, and the cost of idle capacity. A flexible data center has to define exactly what can move.

Where that margin exists, computing can respond to electricity conditions rather than remaining perfectly rigid. Work may be scheduled for periods of abundant generation, reduced during constrained hours, or routed to another suitable site.

Physical location can also be reconsidered. A data center at or near an energy site can create demand where power is available. That does not prove it consumes electricity that would otherwise have been curtailed.

The electrical relationship matters: connection structure, metering, timing, market conditions, and a credible account of what the generator would have done without the load.

Reliability adds another condition. Curtailment occurs during surplus periods, while a data center also needs an operating plan for hours when local generation is weak or absent. Networking can become the next constraint. A site with power still needs fiber, suitable land, cooling, access, permits, and an operating model. Those conditions meet in site selection.

Policloud develops and deploys physical, modular data-center infrastructure for defined sites. An energy or industrial location can be evaluated around the power available there, together with its network, land, workload, regulation, ownership, and operating requirements.

The commercial opportunity is not “free electricity.” It is a site where useful computing demand and real power conditions fit.

Ask why the power disappeared

Return to France’s two figures.

Three terawatt-hours of wind and solar output was modulated during negative-price periods in 2025. One-tenth of a terawatt-hour was curtailed at the transmission operator’s request.

Calling both “wasted renewable energy” would hide the market rules, contracts, grid conditions, and operational instructions that produced them. Calling neither a problem would be equally complacent.

The right response starts with diagnosis. Was electricity unwanted because demand was low? Could it not leave the region because a line was full? Did the system need a rapid adjustment? Did market incentives encourage the generator to reduce output? Does the same constraint recur often enough to justify new infrastructure?

Renewable curtailment is sometimes a safety tool, sometimes an economic response, and sometimes a sign that generation has grown faster than the systems around it.

The target should not be a power system that never curtails. It should be one that curtails for clear reasons, measures the result honestly, and recognizes when a temporary operating tool has become a persistent infrastructure failure.

The missing electricity is not stored somewhere, waiting to be collected later. It was an opportunity the system could not, or chose not to, take.

Knowing which one happened is the beginning of using more of it.