On 9 December 2025 RTE, the company that runs the French high-voltage grid, published its ten-year outlook and did something forecasters rarely do in public: it lowered its own number. The electricity France would need in 2035 was cut by about 35 TWh against the analysis it had published in 2023. In the same document it kept a path on which consumption rises by 135 TWh in ten years, and it said that path was still within reach.
Six weeks later the French Treasury published what the meters had recorded. Consumption in 2025, adjusted for weather and the calendar, stood at 449 TWh, the same level as in 2024 and about 6% below the average of 2017 to 2019. Nobody in these two documents is wrong. They are answering different questions, and a reader who does not see that will misread both.
The loudest arguments inside the energy transition are not disputes about who is right but disputes about what is being measured: a forecast set against a meter, one word covering two schemes, and a heat decision that turns on a ratio no official body publishes for France, so the professional who asks which quantity a number is, before using it, is the one worth hiring.
This is the fourth piece of the energy chapter of the CareerOn Industry Atlas. The first asked where the money is made, the second which positions the power system cannot replace, the third what is actually being built. This one asks where informed, official voices disagree, and it follows one rule throughout. A disagreement is published, never reconciled. Each value is printed as its publisher printed it, with its date and its scope. Nothing is averaged, no winner is chosen, and no motive is inferred.
The piece reads eight documents: the RTE outlook, the Treasury note, two reports of the French energy regulator, a report of the International Energy Agency and three price publications of the statistics service of the French ecology ministry. A fourth argument, on hydrogen targets, was planned and dropped, because the documents it needed could not be opened and checked. It is not discussed here, and its absence is recorded in our research notes rather than filled with memory.
The forecast and the meter
A system operator has to plan lines and substations years before anyone plugs into them. Its outlook is therefore a set of scenarios, not a prediction, and RTE writes it that way. The growth path assumes industry electrifies and heating and transport switch from fuels to the grid. It is the path the operator must be ready for, not the path it promises.
RTE is candid about the timing. It judges a significant upturn in consumption unlikely before 2028. It also reports that about 30 GW of capacity is already reserved for connection to the transmission grid, which would correspond in theory to about 180 TWh of additional consumption, and it adds, in its own words, that not all of these projects will happen.
That sentence is the whole argument in miniature. A reservation is a queue, not a load. A company that books a connection is buying an option on the future; it is not yet drawing power. The Treasury note measures the load. It reports what the country actually consumed, corrected for a mild or harsh winter, and finds it flat.
Demand: a forecast that still climbs, a meter that has not moved
| Kind of number | What the document prints | Publisher, date |
|---|---|---|
| Scenario to 2035 | Consumption up 135 TWh in ten years on the growth path | RTE, 9 December 2025 |
| Revision | 2035 need cut by about 35 TWh against the 2023 analysis | RTE, 9 December 2025 |
| Queue, not load | About 30 GW reserved on the transmission grid; not all projects will happen | RTE, 9 December 2025 |
| Measured 2025 | 449 TWh, weather adjusted, level with 2024 | French Treasury, 23 January 2026 |
Sources 1 RTE (Réseau de Transport d'Électricité) · 2 Direction générale du Trésor
Read side by side, the two documents do not contradict each other. The forecast has been revised downward and still climbs; the meter has not moved. Both statements are true on the dates they carry. The mistake is to put them on one scale and draw a line between them, and the exhibit above refuses to do so.
For anyone advising an industrial site, this is not an academic distinction. A plan built on the growth path assumes grid capacity and prices that depend on demand actually arriving. A plan built on the meter assumes a system with room to spare. The honest brief says which document it relies on, and why.
There is a quieter lesson in the way the operator wrote its revision. It did not bury the lower number in an annex. It stated it plainly and explained it, which is what a public forecaster owes the people who plan around it. A reader who treats a revision as an admission of failure learns to distrust the forecasters who are most candid, and that is the wrong lesson to learn.
One word, two schemes
Flexibility means using less power, or using it later, when the system is under strain. It is the word every energy strategy now reaches for, and it hides at least two different things.
The regulator’s foresight report of February 2026 notes that volumes of demand response actually activated on the markets remain limited outside the exceptional crisis of 2022, despite 3 GW certified in 2023 under the capacity mechanism. The capacity mechanism pays for being available at peak. Being certified is not the same as being called.
The regulator’s smart-grid report of 11 December 2025 counts another scheme, NEBCO, through which demand response is sold into the energy markets. There, certified capacity nearly doubled, from 5.7 GW in 2022 to 10.9 GW in 2024, for a volume shifted of 56 GWh in 2024. The same report adds that the flexibility actually contracted and activated today remains very small.
It is tempting to set the smaller figure beside the larger one and announce that flexibility tripled. The documents do not allow it. The two figures certify different things under different rules, and the second report measures energy shifted in GWh, a third quantity that is not capacity at all.
Flexibility: one word, two schemes, three quantities
| Scheme or source | What the document prints | Quantity |
|---|---|---|
| Capacity mechanism, 2023 | 3 GW certified; activated volumes still limited | Capacity (GW) |
| NEBCO, 2022 to 2024 | Certified demand response from 5.7 GW to 10.9 GW | Capacity (GW) |
| NEBCO, 2024 | 56 GWh shifted | Energy (GWh) |
| Voltalis estimate, 2033 | 11 to 12 GW of heating demand response possible | Operator estimate (GW) |
Sources 3 Commission de régulation de l’énergie (CRE), Prospective · 4 Commission de régulation de l’énergie (CRE)
The foresight report also carries the largest number in this section. At the 2033 horizon, Voltalis, a demand-response operator, estimates that there would be a possibility of 11 to 12 GW of heating demand response. That estimate belongs to Voltalis. The regulator’s report records it; it does not adopt it, and the report itself was prepared by a group co-chaired by outside experts rather than decided by the regulator. We print the figure with its owner’s name attached, which is the only way it can be printed honestly.
The two reports agree on the one thing that matters most for people who build this market: the gap between what is certified and what is used. Capacity exists on paper. The work lies in making it answer when called.
This is also why the reports read so differently depending on who quotes them. An aggregator selling heating flexibility can cite the certified figure; a sceptic can cite the activated volumes; both are quoting the regulator accurately. The disagreement is not between the documents. It is between readers who stop at the headline number and readers who ask which scheme it belongs to, what it certifies and whether it has ever been called.
A ratio nobody publishes for France
The third argument concerns industrial heat, the steam and hot water that factories use to cook, dry and clean. The technology to make that heat with electricity, through industrial heat pumps and electric boilers, is commercially available. The question is whether it pays.
The International Energy Agency’s report on electrifying low-temperature industrial heat, dated December 2025, gives a direct answer for the countries it studies. Deployment has remained limited, it writes, because of unfavourable electricity-to-gas price ratios, long grid connection lead times and the absence of clear policy frameworks. Including distribution costs and taxes, it puts the ratio at 2.7 for Czechia, 2.6 for Germany and 3.1 for Ireland.
France is not in that analysis. The obvious move is to build the French number from French statistics, and the statistics exist. The ecology ministry’s statistics service published both on 23 July 2026. Businesses paid on average 148.8 €/MWh excluding VAT for electricity in 2025, down 9.6% on the year. For gas they paid 65.7 €/MWh excluding VAT, measured on a gross calorific basis, up 1.5%.
Heat: the ratio that decides is printed for three countries, not for France
| Scope | What the document prints | Publisher, date |
|---|---|---|
| Czechia, Germany, Ireland | Electricity-to-gas ratio with taxes: 2.7, 2.6 and 3.1 | IEA, December 2025 |
| France, electricity, 2025 | 148.8 €/MWh excluding VAT, down 9.6% | SDES, 23 July 2026 |
| France, gas, 2025 | 65.7 €/MWh PCS excluding VAT, up 1.5% | SDES, 23 July 2026 |
| France, ratio | Not published in any admitted document | None |
Sources 5 International Energy Agency · 6 SDES (ministère de la Transition écologique) · 7 SDES (ministère de la Transition écologique) · 8 SDES (ministère de la Transition écologique)
We do not divide one by the other. The two averages cover every business, and neither describes the price any single site pays. In its release for the first half of 2025, the statistician itself observes that companies using a lot of electricity pay a price three times lower than those using little. The gas figure is stated on a gross calorific basis, and the agency’s ratio follows its own method of taxes and network costs. A number made by dividing them would look official and would not be. What the two publications do show, without arithmetic, is direction: in 2025 the business electricity price fell and the business gas price rose.
That is a smaller claim than a ratio, and a truer one. For a consultant, it defines the job. The decisive number for a French plant is not published anywhere a reader can cite; it has to be built from that plant’s own contracts, tariffs and load, and the method has to be shown.
The counter-case
The strongest objection is that this piece mistakes caution for insight. Forecasts are always revised, schemes always multiply, and every ratio depends on its method; saying so, the objection runs, is not an argument but a disclaimer.
The objection is fair in part. None of the documents here is confused about its own scope, and the operator, the regulator and the statistician each state their limits clearly. The confusion arises downstream, when their numbers are lifted into a slide or a headline without those limits. That is where decisions go wrong, and it is the reason this piece prints the limits beside the values.
A second objection is that one year of flat demand proves little. That is also correct. The Treasury note measures one year and projects nothing, and RTE’s own caution about the period before 2028 points the same way. Flat demand today is consistent with growth later. It is not consistent with treating growth as already here.
What would change the reading
The reading would change if an official body published a French electricity-to-gas price ratio for industry, if a second official projection for 2035 appeared that could be set against RTE’s, or if the regulator reported activated flexibility volumes that closed the gap with certified capacity. Each would be a dated document, and we will update the piece and record the change when one appears.
What this means for your career
The three arguments share a skill: reading what a number is before using it. Two roles practise that skill every day.
An industrial energy-transition consultant is paid to tell a client whether to electrify a process, and when. The work is exactly the gap this piece describes: separating a scenario from a measurement, and building the price comparison a plant actually faces instead of borrowing one made for another country.
A local smart-grid flexibility project lead is paid to turn certified capacity into capacity that answers. The work sits between the scheme rules, the sites that shift their consumption and the network operator that calls on them, and it is judged by what is activated, not by what is on paper.
If you want to practise the client-side decision, start with the industrial energy-transition consultant simulation. If you want to practise making flexibility real on a local network, start with the smart-grid flexibility project lead simulation. None of these documents tells you how many jobs there are, and we do not pretend otherwise. Each simulation lets you do the work before you choose it.