The Essentials
- Load shedding involves temporarily reducing or suspending electricity consumption by customers (manufacturers, service sector businesses, and, in some cases, residential customers) during periods of peak demand.
- Companies are compensated for this action under specific contracts, while individuals derive only an indirect benefit—for example, through the installation of connected devices that help reduce their overall bill.
- To encourage more customers to take the plunge, the government has injected additional public funds into the debt forgiveness program. How effective has it been, and at what cost? The Court of Auditors recently published a scathing assessment of the program, which was in place through 2026.
The electrification of energy uses (electric mobility, heat pumps instead of gas furnaces, etc.) and the decarbonization of electricity generation through technologies with limited flexibility (nuclear power and non-dispatchable renewable energy sources) have a very tangible consequence: electricity consumption is increasing, but it is more difficult for the grid to adapt to this in real time.
Developing flexibility in electricity consumption can address this challenge. Customers are encouraged to shift their electricity use to times when power is available and to avoid using electricity during peak demand periods.Load shedding—which involves temporarily reducing or suspending a customer’s electricity consumption during peak demand periods—is part of this approach.
France has set ambitious goals in this area and, starting in 2018, has implemented an annual bidding process that guarantees the winning bidders—demand response aggregators (intermediaries between consumers and RTE)— a fixed payment (in euros per megawatt of load reduction capacity per year) in exchange for their ability to “reduce” electricity consumption at industrial sites, service sector businesses, and residential homes. In residential homes, residents install a connected device that allows them to temporarily shut off their hot water heater or electric heating system.
In its August 2026 report, the Court of Auditors explains that between 2018 and 2025, public support for debt forgiveness cost nearly 430 million euros and that the government has committed to spending 1.18 billion euros in the coming years. The Court estimates that more than two-thirds of this expenditure could have been avoided while achieving the same result.
How did we get here? A series of technical, political, and administrative decisions—each of which could be justified on its own—has led to this situation. As part of a multidisciplinary research project on electricity flexibility, we interviewed experts and industry stakeholders to better understand the decisions that led to this situation.
A single prize for very different efforts
The demand response call for bids (AOE), which was in effect through 2026, operated as a single-price auction:
-
We select the lowest-priced bids up to the desired volume,
-
Everyone is thus paid the price of the last necessary bid, known as the “clearing price.” Participants who submitted the lowest bids receive the difference between their price and this final price;
-
There is a maximum price for these auctions: bids above that amount will not be accepted.
The economic rationale for this price is that, in a competitive market, this surplus can finance investments or reward the most efficient players. But there is no evidence that this reasoning applies to load shedding.
In fact, going without electricity temporarily doesn't cost everyone the same amount.
-
Some factories can postpone or relocate their production at a relatively low cost relative to their energy consumption.
-
For other consumers, this capability requires larger investments.
By applying a single price, the system provides a significant margin to participants for whom the effort is less costly.
A market with no real competition
According to many auction experts, the “clearing price” is intended to encourage economic agents to disclose their actual production costs and to limit strategic behavior.
-
However, every year from 2018 to 2024 (see the appendices to the Court of Auditors’ report), the demand response tender (AOE) was undersubscribed. In other words, there were fewer bidders for demand response than there was demand for it. Without competitive pressure, any bid submitted below the price cap was therefore certain to be accepted. Under these circumstances, a bidder had no reason to disclose its actual cost: it was in the bidder’s interest to inflate that cost and offer a price close to the price cap.
-
The problem was exacerbated in 2021 when this cap was doubled, rising from €30,000 to €60,000 per MW. Originally intended to encourage new players to enter the market, this increase benefited everyone, including existing facilities.The estimate of potential resources, initially conducted by ADEME, was adopted without scrutiny by the designer of the solicitations. Furthermore, the 2023 and 2024 solicitations supported requests for multi-year financing guaranteeing this level of revenue over 4 to 10 years.
-
In 2025, the trend reversed: the call for bids was oversubscribed, meaning there were more bids submitted than the demand for load reduction. Competition finally kicked in, and the price dropped to 28,863 €/MW, proving that the 60,000 €/MW cap was far too high.
These flaws in the mechanism therefore stem from a fragmentation of decision-making between, on the one hand, the identification of resources and, on the other, the design of the calls for bids. Why weren’t they corrected quickly? The Court indirectly answers this question by pointing out that, in the case of industrial load shedding, the beneficiaries were energy-intensive manufacturers (and not the aggregators).
According to several industry stakeholders we spoke with, the mechanism was not improved because it allowed for “disguised subsidies” to the industry. This rationale is said to be an “open secret.”
Does the erasure of individuals enrich aggregators?
Was there a similar windfall effect for the removal of individual records? If so, it would have primarily benefited only the aggregator, since individuals are not compensated by existing aggregators for their contributions.
Initially, calls for bids from business and residential customers were kept separate.
-
However, starting in 2022, a single price has been systematically applied, based on the latest bid for the industrial lot.
-
In 2021, no bid in the residential category exceeded 16,000 euros per megawatt, while the highest accepted bid in the industrial category reached 55,800 euros.
-
Starting in 2022, operators serving residential customers began offering prices close to the cap: as a result, they received a rate three and a half times higher than what they had initially requested (€55,800 received versus €16,000 requested), according to the Court of Auditors.
According to the experts interviewed, standardizing auctions with a single price was in line with the goal of developing distributed load response and with the requirements of non-discrimination among economic actors, which the European Commission has consistently advocated in the context of state aid control and electricity market regulation. Load response is considered a service that must be compensated at a uniform price.
But the Court of Auditors points out an additional problem: the widespread write-off (that is, for private individuals) also benefited from a series of government subsidies.
Launched as part of the energy conservation plan following the 2022 crisis, the “Coup de pouce pilotage connecté du chauffage pièce par pièce” program subsidized the installation of smart thermostats (approximately €500 per home for 715,000 households) between December 2023 and November 2024. However, a demand response aggregator can install a device that combines both functions—control and demand response—and can therefore be compensated through both mechanisms.
The Court assessed the cost-effectiveness of widespread data erasure for economic actors, based on the administration’s estimate of the cost of an installed device billed at 300 euros (plus 30 euros per year in operating costs).
The “Coup de pouce” grant (€500) reimbursed the entire investment, even though the request for proposals already promised a return on investment within three years and a 45% return after five years. By adding the “Coup de pouce” grant, the return reaches 90% in five years—and, she points out, even under unfavorable cost assumptions.
The “Coup de pouce” program was discontinued due to abuse and fraud, but it is possible that the administration was concerned about a windfall effect resulting from the combination of the two programs. In fact, the administration did not seek to single out fraudsters and simply chose not to renew the program. This unilateral decision could confirm the hypothesis that this stacking of benefits was not identified before the program was implemented.
"Phantom" erasure capabilities?
One more troubling question remains: Did these paid load-shedding capacities actually exist? In fact, the Court of Auditors points out that the actual load shedding fell far short of the reported capacities. RTE tests conducted before 2018 showed a reliability rate of less than 75% of the tested capacity. The report refers to “phantom” capacities.
The report, which draws on data from RTE, reveals a tactic used by many aggregators: since they are required to declare the price at which they agree to be activated, they set that price at the wholesale market cap minus one euro—which makes their activation virtually impossible. As a result, they are paid to maintain the capacity to act while remaining inactive.
Those involved in load shedding dispute this accusation and point out that the recent period (since 2023) has also been marked by economic conditions that are less conducive to load shedding.
According to these stakeholders, demand response capacity should be viewed as a last-resort measure, to be used only during exceptional consumption peaks. Indeed, distributed demand response aggregators have no incentive to activate demand response on a large scale—up to the levels of capacity valued in the bidding process—as this would result in significant losses in consumer comfort.
This strategy is partly inconsistent with one of the political motivations for supporting load shedding: to avoid having to rely on additional generation capacity.
Does free competition eliminate the need for profitability oversight?
However, the Energy Code requires the government to ensure that the return on capital invested by the successful bidders “does not exceed a normal return on capital, taking into account the risks inherent in these activities.” Yet no such review was conducted, either before or after.
The administration concluded that the competitive bidding process was sufficient to rule out the risk of overcompensation. This position is consistent with European legal doctrine, which considers competitive bidding to be a sufficient guarantee of proportionality and does not require verification of actual costs.
The Court of Auditors does not share this view; it advocates for more systematic oversight. It therefore criticizes the doctrine that competition precludes the need for oversight. This doctrine is not universally accepted within the government: ADEME, for its own calls for proposals, asks bidders to provide their business models and reserves the right to reduce the subsidy retroactively if economic data warrants it.
In its responses to our questions, Voltalis, the leading demand response aggregator, disputes these arguments. It explains that the subsidies made it possible to develop a new service for consumers, which entailed its own investment and operating costs. If the profitability of its operations had been in line with the Court’s calculations, competitors would likely have invested to take advantage of the windfall effects.
In our interviews, experts emphasize that the lack of appeal of this business is due more to uncertainties regarding financing—and, above all, the business model—which resembles that of“Uber-style” platforms, where significant fixed costs must be invested to design and ensure the reliability of the digital platform, which must then be recouped through a large number of connected sites.
Voltalis has been in business since 2010, but it benefited from the 2022 energy crisis, during which the dramatic surge in electricity prices allowed it to reap extraordinary profits on the wholesale market, followed by particularly generous multi-year funding in 2023 and 2024. It enjoys a first-mover advantage that no competitor can match.
What Does the Future Hold for Data Erasure?
The demand response auction (AOE) was phased out in 2026 and replaced by a revamped capacity mechanism, funded no longer by the state budget but by a tax on electricity consumption.
This new framework groups energy storage, batteries, and dams into a single category of decarbonized capacity. Multi-year contracts can support new projects, but no ex post (retrospective) monitoring of companies’ actual profits is planned: the government continues to rely entirely on market competition to limit excess profits.
At the same time, suppliers are working on incentive offers to encourage customers to spread out their consumption over time (shifting the charging of their electric car or heating during off-peak hours), which is known as explicit flexibility. However, the business model for their offerings remains fragile and could be improved by integrating the distributed load response service.![]()