The Last Plant
A wind farm in Germany gets paid the price of natural gas. The wind is free. The gas is not. Germans now pay about thirty-nine and a half cents per kilowatt-hour, the highest household electricity bill in the European Union bar one.1 A third more than the French. More than double what Americans pay.
Germans were told the Energiewende would deliver clean, secure, affordable electricity. They got the most expensive household power in the EU, an industrial base shrinking faster than any other on the continent, and a country that imports the same carbon-free electricity it switched off in April 2023.
The standard postmortem gives you two villains. Russia, for the gas crisis of 2022. The Energiewende’s authors, for switching nuclear off too fast. Both are real. Neither names what actually happened. For twenty-five years Germany ran one experiment, under three governments and six coalitions, in whether political will can replace the price signals of an electricity market. The bill is the result.
Almost every fix currently on the German political table assumes the problem is one of money. Subsidise the grid fee. Cap industrial electricity at six cents. Cut the Stromsteuer. Pay-out, pay-around, pay-down. None of these move gas out of the slot where the auction reaches for the last plant it needs. They split the bill between the household, the industrial buyer and the taxpayer, and hope the next government can find another envelope.
The German electricity bill is not a market failure. It is a policy outcome with an auction attached.
Electricity has no shelf
One fact explains everything that follows. You cannot store electricity at any meaningful scale.
Every other thing you buy sits on a shelf until you want it. Electricity does not. The instant you switch on a kettle, somewhere in the country a generator must produce a little more, right then, to balance it. Supply must equal demand every second of every day or the grid’s frequency drifts and machines start to break. Germany’s entire grid storage, every pumped reservoir and every battery together, could run the country for a matter of minutes. Against the roughly five hundred billion kilowatt-hours it consumes each year, that is a rounding error.
So electricity has to be matched, continuously, in real time. The way you do that for a whole country is an auction. It is run by an exchange called EPEX SPOT, and it clears at noon every day for all twenty-four hours of the following day. It clears in fifteen-minute slices, so ninety-six prices per day instead of twenty-four.
This is the wholesale price. It is the raw, volatile cost of the electricity itself before any cable or any tax. Everything on your bill is built on top of it. And it is set, every day, by a rule I had not heard of until recently.
The staircase
Every plant in the country makes an offer for tomorrow. Each one says, in effect: “At 7pm tomorrow I can deliver this many megawatt-hours, and the least I will accept is this many euros.”
That floor is not the cost of building the plant. That money is long spent. It is the cost of producing one more unit, right now, from a plant that already exists: the fuel, the carbon permit, a little wear on the machine. Economists call it the marginal cost, and it differs enormously by technology.
A solar panel’s fuel is sunlight. Its marginal cost is essentially zero. A wind turbine’s fuel is wind. Also essentially zero. A nuclear plant is low, because uranium is cheap and the machine cannot easily be throttled. Lignite, the brown coal Germany still digs out of the ground, sits in the middle. Hard coal is higher. A combined-cycle gas turbine is higher still, and it lurches up and down with the price of gas. A gas peaker or an oil unit is highest of all.
Line up every plant’s offer cheapest first and you get a rising staircase. Free renewables at the bottom. Nuclear next, then lignite, then hard coal, then gas in a thick band somewhere near the top, with oil and peakers above. Then ask the only question that matters: how much power does Germany need tomorrow at 7pm? Draw that as a vertical line across the staircase. The auction switches plants on from the bottom up until it has enough to meet that line.
The plant where the line lands is the marginal plant. It is the last one needed, the most expensive one running, and it sets the price for every plant to the left of it. It is the only line in German energy policy that was not drafted in a coalition. It is also the only one that does what it is supposed to.
The wind farm gets paid the price of gas
Every plant that runs in that hour is paid the same price. The price of the marginal plant. The most expensive one switched on.
This was genuinely counterintuitive to me. If gas is the last plant standing at 7pm and it sets the clearing price at ninety-five euros per megawatt-hour, the wind farm that offered its power at zero is also paid ninety-five. So is the nuclear plant. So is the lignite block. Everybody to the left of the demand line gets the same number on the cheque.
It is not a windfall. The gap between what the wind farm needs to operate (nothing) and what the auction pays it (the gas price) is exactly how it pays back the cost of being built. The system pays everyone the same clearing price on purpose, because the alternative, paying each plant whatever it bid, teaches every plant to stop bidding its true cost and start guessing the clearing price instead. You end up paying more, not less.
This design has a name. Pay as clear. Or, more formally, uniform marginal pricing. It is the dominant design in the United States, across most of Europe, and in much of the world. Most of the time it is efficient and quietly fair. The cheap plants get the producer surplus that finances the next plant. The expensive plants only run when they need to.
The rule has one other consequence. It hands the price of every megawatt-hour the country consumed that hour to whichever plant happened to be last in line. It does not matter that wind and solar made up half the supply. It does not matter that their fuel was free. If the last plant needed to keep the lights on was burning gas, then gas sets the price. For everyone.
Why gas
In 2024 gas-fired plants generated 13.2 percent of Germany’s electricity. A little under a seventh. And yet during most trading hours of the year, gas was the marginal plant. Gas set the price.2
Those two numbers do not match because being the price-setter is not about volume. It is about who is last in the merit order during the hour that matters.
Wind and solar flood the bottom of the staircase, but only when the weather agrees. The wind has to blow and the sun has to shine. At 7pm in January, with the sun down and the wind dropped, the staircase below gas is mostly empty, and the auction climbs up to gas to find the last megawatts.
Take total demand and subtract the wind and solar generated in that hour. What is left is the gap the controllable plants must fill. The Germans call it the Residuallast, the residual load. It is the most important number you have never heard of, because it is what the gas plants are sized for.
On the worst windless evening of January 2024 the residual load peaked at around sixty-six gigawatts against a total demand of about seventy-six. The country needed sixty-six gigawatts of controllable plant ready to deliver. In July 2024 the residual load went negative at noon on sunny days, because solar was generating more than the country was consuming and the controllable plants were being asked to disappear. The same month’s evening peak still cleared fifty gigawatts.
Renewables changed the floor. They did not change the ceiling. Germany still needs sixty-something gigawatts of on-call power for the worst hour of the year, and the on-call power is mostly gas. The new fifteen-minute auction even sharpens this: the solar-noon crash is now ninety-six prices instead of twenty-four, and so is the evening climb.
This is the structural reason gas sets the price most hours. It is not a temporary problem. It is what happens when you build a lot of weather-dependent capacity without building, at the same speed, the storage, the demand response, and the firm low-carbon plant that would let you stop reaching for gas in the dark hours.
The night the bill went vertical
If gas sets the price, the price of gas eventually becomes the price of electricity. For most of the 2010s nobody had to think about this, because gas was cheap. The European benchmark contract, the TTF, sat below €20 per megawatt-hour for most of the decade and was trading around €27 in August 2021.
Then Russia turned the taps.
On the 24th of February 2022 Russian troops crossed into Ukraine. Germany suspended the certification of Nord Stream 2 the same week. In the months that followed Russian pipeline gas was throttled, then disrupted, then in September the Nord Stream pipelines were physically sabotaged. By the 26th of August 2022 TTF peaked at €319 per megawatt-hour.3 More than ten times where it had been a year earlier.
The German wholesale power price tracked it almost perfectly, because gas sets the price. The annual average day-ahead price went from €97 per megawatt-hour in 2021 to €235 in 2022. The August 2022 monthly average alone was €465. On the evening of the 29th of August, in the hour from 7 to 8pm, the German price hit €871 per megawatt-hour.4
Under pay as clear, every megawatt-hour produced in that hour, including the zero-fuel wind and the zero-fuel solar that made up nearly half of supply, re-priced upward to the gas number. The country was generating a lot of cheap power. It was paying for all of it as if the cheap power were expensive.
This is not a failure of pay as clear. Pay as clear was doing exactly what it was designed to do. It is a failure of the bet that gas would stay cheap. The Energiewende, the German energy transition, had been built on the implicit assumption that Russian gas would be the bridge fuel smoothing the renewables ramp and replacing nuclear. The assumption broke in a single year. The auction did the rest.
Paying to give it away
The same rule that pays a wind farm €95 when gas sets the price pays it negative numbers when gas does not. Negative electricity prices mean the generator pays the buyer to take the power away. It happens when subsidised renewables keep producing regardless of price and dispatchable plants cannot be turned off fast enough. The hours per year when this happens have been compounding: 69 in 2022, 301 in 2023, 457 in 2024, 573 in 2025.5
Germany succeeded at building enormous amounts of weather-dependent generation. It did not build the storage, demand response, or interconnection to use that generation when it shows up. The energy is there. The flexibility is not. In midsummer noon hours the country gives the energy away. In midwinter evenings it pays the gas price to get some back.
The four floors
The wholesale price right now is about 8 cents per kilowatt-hour. The household bill is 40. Pay-as-clear explains the 8. Two more layers explain the other 32.
For an average household using 3,500 kilowatt-hours a year, the 2025 retail bill breaks down into roughly 16 cents of energy (wholesale plus supplier margin), 11 cents of grid fees, and 13 cents of state taxes, levies and VAT.6 Of the four floors, only the wholesale floor is set by the market. The other three are set by people who do not pay them.
The most politically charged line for twenty years was the EEG-Umlage, the surcharge that financed the gap between the feed-in tariffs paid to renewable generators and what they earned on the wholesale market. It started at 0.19 cents in 2000 and peaked at 6.88 cents in 2017, more than the entire wholesale price that year. Then in 2022 the Scholz government zeroed it. The cost did not vanish. It moved. From the power bill to the tax bill, where it still runs about €17 billion a year and where almost no household notices it.7
The other half of the trick is that wholesale prices have fallen much faster than retail. Spot prices dropped about 67% between 2022 and 2024. The retail bill fell about 13%. Suppliers hedge their volumes years ahead, so the cheap power they buy now will only show up on bills in 2026 and 2027. And grid fees rose at the same time, because integrating dispersed renewables into a grid designed for centralised plants is expensive, and the bill is where that expense lands.
How Germany made gas the last plant
If pay-as-clear is the rule, the policy of the last twenty-five years is what put gas in the slot the rule reaches for. Each of the moves below is defensible on its own terms. Each one removed a cheap rung between renewables and gas, or made the gas itself more expensive. The slot the auction reaches for at 7pm in January did not become gas by accident. It became gas by twenty-five years of choices, none of which named gas as the target, all of which ended in the same place.
This is the camel at the scale of an entire energy system. A sequence of half-decisions, each one defensible alone, ground through the overlap of what every coalition in turn could stomach, ending in the position nobody designed.8
2000. Schröder’s SPD-Green government passes the EEG, guaranteeing twenty-year feed-in tariffs to renewables, and signs the nuclear consensus with the four utilities, agreeing to phase out the seventeen German reactors by around 2022. Two decisions in one year, by one coalition, that will define the next quarter century. Renewables are 6.3% of consumption.
September 2010. Merkel’s CDU-FDP government extends reactor lifetimes by eight to fourteen years. The nickname is Ausstieg aus dem Ausstieg, the exit from the exit.
11 March 2011. Fukushima. Within days, Merkel takes seven of the oldest reactors offline. On the 30th of June the Bundestag votes overwhelmingly to shut every German reactor by 2022. Eight close immediately. This is the pivot point of the whole story. It is also a panic decision with no replacement plan of equivalent scale for the roughly 140 terawatt-hours per year of carbon-free baseload that was about to disappear.
July 2020. Merkel’s CDU-SPD coalition passes the Kohleausstiegsgesetz: coal exits by 2038, with €4.35 billion in operator compensation and €40 billion in structural aid for the coal regions.
31 December 2021. Three more reactors close: Grohnde, Gundremmingen C, Brokdorf.
24 February 2022. Russia invades Ukraine. Nord Stream 2 certification is suspended in Berlin within days. In September the pipelines are sabotaged. Russian pipeline gas, which had been roughly 55% of German gas imports in 2021, effectively ends.
15 April 2023. Isar 2, Emsland and Neckarwestheim 2, the last three German reactors, are disconnected from the grid under the Scholz government. Sixty-two years of German nuclear power end on a Saturday evening. Cumulatively those plants had produced about 5,200 TWh and avoided roughly 4,200 Mt of carbon dioxide.9
You can argue the politics of any one of those moves. The picture they add up to is harder to argue with.
Total electricity generation has shrunk by about a fifth since the 2010 peak (633 TWh in 2010, 497 in 2024). The nuclear band, which produced 170 TWh of carbon-free electricity in 2001, fell off a cliff after 2011 and reached zero in 2023. Wind and solar grew from a slice to most of the chart but did not refill the total. Germany generated less power in 2024 than at any point since the early 1960s, in a country whose population grew and whose stated policy is electrify-everything.
Germany did not gradually replace nuclear with renewables. It switched nuclear off, leaned harder on coal and gas in the intervening years to cover the gap, and built renewables on a parallel track. The two timelines never met. By the time renewables were a serious share of generation, the nuclear plants were already gone.
The flip
The signal that the trouble had already arrived was 2023. For two decades Germany had been a net electricity exporter, typically by 20 to 50 TWh a year. In 2023 it slipped into being a net importer for the first time since the early 2000s. In 2024 it deepened. The net imports were 9 TWh in 2023 and 25 TWh in 2024.10
The mix of the imports is the part that matters. In 2024 about a third came from France, another large chunk from Denmark, then Switzerland and Norway. About a quarter of total imports came from French nuclear plants. About half from renewables on neighbouring grids. The country that switched off its carbon-free baseload in April 2023 spent 2024 buying the same kind of electricity from countries that had not.
The factory floor
The hardest number on the German energy bill is not the household one. It is the industrial one. Germany has an unusual economy by EU standards: a large share of its manufacturing is what the statisticians call energy-intensive, the basic chemicals and metals and glass and paper that sit upstream of almost everything else. When the price of electricity and gas changes, those industries are the first to feel it, and they are not easily moved.
Take the production index for the five energy-intensive branches that Destatis tracks together. Index them to 2021, so the year before the crisis is the 100-mark. By 2023 the index was at 83.7. By 2025 it had drifted further to 82.1. A drop of about 18% in four years, with no recovery visible in the data. Across the whole of German industry the drop over the same period was about half that.11
The single best-documented example is BASF, the largest chemical company in the world and a Verbundsite in Ludwigshafen that has been running since 1865. On the 24th of February 2023 the company announced 2,600 layoffs and the closure of one of two Ludwigshafen ammonia plants, plus several other lines. The reason given, in the company’s own words, was an additional energy bill of €3.2 billion in 2022, of which roughly 84% fell in Europe and most of that in Ludwigshafen.12 In September 2024 the company announced a second, deeper round of Ludwigshafen cuts, targeting one in seven facilities and another €1 billion per year of savings by the end of 2026.
The chemical industry as a whole lost about 23% of its production volume in two years, between 2021 and 2023.13 The Bank of Italy traced the contraction outward through European supply chains and estimated that shocks originating in German manufacturing explained nearly a third of the six-month variance in Italian manufacturing output. The IMF estimated the high-energy-price shock knocked roughly 1.25% off German potential output.
You can argue that fears of broad deindustrialization are overblown, that the damage is concentrated in a handful of sub-sectors, and that some of the drop is permanent efficiency rather than involuntary shutdown. All of those qualifications are real. The numbers are still the numbers. The country that the rest of Europe relies on for its basic chemistry has, on the measure that matters, produced about a fifth less of it for four years running. This is what political incompetence costs when it tries to outrun a price signal.
The strongest defences
The strongest defences of the policy are the ones I have the hardest time dismissing, so let me give each its best version before saying why I don’t buy it.
Emissions fell, and that is the point. True, and it matters. Power-sector carbon emissions went from about 330 Mt in 2000 to 183 Mt in 2024. The single-year drop from 2023 to 2024 was 9%. Total German greenhouse gas emissions in 2024 were 48% below 1990. Renewables share of electricity consumption went from 6.3% in 2000 to 54% in 2024, a sliver to a majority. New renewables clear EEG auctions at 5–7 cents per kilowatt-hour for utility-scale solar and 6–8 cents for onshore wind, which is below the LCOE of new gas or new nuclear in Germany. The case for building more of them is purely economic. You would do it even if you did not care about emissions.
This is the defence that lets German politicians sound serious while avoiding the thing they did. Yes, emissions fell. They also shut down a safe, working, carbon-free technology because a tsunami hit a different reactor design on a different continent, and then pretended the missing power would be supplied by adjectives. Merkel did not discover a new fact about German nuclear physics in March 2011. She discovered a new political fear, and chose to ride it.
That is not a trade-off. A trade-off is when you name the cost before you pay it. Germany was sold clean, secure, affordable electricity. What it got was cleaner electricity, less security, higher prices, and a political class still speaking as if the first word cancels the other three. The honest defence would be: we made electricity expensive, weakened industry, and increased dependence on neighbours because we thought shutting nuclear was worth it. Almost nobody says this, because stated plainly it sounds insane.
Cap industrial electricity to save the factories. This is the policy Germany already chose. The Industriestrompreis came into force on 7 May 2026, applies retroactively from 1 January, runs through 2028, and gives energy-intensive industry electricity at five cents per kilowatt-hour on half its annual consumption. Cost to the federal budget: €3.1 billion over three years, covering roughly ninety sub-sectors from chemicals and steel to glass, cement and semiconductors.14 It is politically attractive because it is fast, it is visible, and it does not require anyone to name what twenty-five years of policy actually cost. It is also, on its own terms, an admission. An Industriestrompreis is the state announcing that the structural problem is permanent, and that the only remaining lever is socialising the bill across a different group of payers. The factories survive on the subsidy. The grid still reaches for gas at 7pm in January. The next subsidy is already coming.
Imports lower system cost. When French nuclear is cheaper than firing a German lignite block plus its carbon permit, importing French nuclear is the rational thing to do. The European single market is built for that. The phrase Agora uses, “lower system cost”, is technically correct on the day. But there is a difference between buying neighbours’ electricity because it is cheaper on a given Tuesday, and buying it because you have structurally less firm capacity than you used to have. The first is trade. The second is dependency. By 2024 the second was quietly also true. A country that switched off its own carbon-free baseload in April 2023 and spent the rest of the year importing carbon-free electricity from neighbours that did not is not paying for trade. It is paying rent on a decision.
The closed door
The cleanest German nuclear policy in 2026 is the one Germany made impossible in 2011 and then finished killing in 2023: keep the last six reactors running. Together they were about eight gigawatts of firm, carbon-free power. Not enough to solve the whole residual-load problem, but enough to move the staircase down in exactly the hours when gas now gets to name the price.
This is what makes the decision so hard to forgive. Nuclear was not a speculative technology Germany had to invent. It was steel and concrete already connected to the grid, staffed by people who knew how to run it, producing the one kind of electricity the Energiewende most needed and least had: clean power that did not care about the weather.
By 2026 the restart case is no longer clean. Some industry groups say several reactors could return in three to five years for one to three billion euros each. The operators and regulators are much colder. Dismantling permits have been issued. Fuel is gone or going. Components are being cut. Isar 2’s operator says plainly that the plant is no longer reactivable. Germany did not merely close the plants. It set machinery in motion to make regret expensive.15
That is the real indictment. The error was not failing to restart nuclear in 2026. The error was creating a country in which restarting safe carbon-free plants has to be discussed like archaeology.
Germany did not fail to build renewables. It built more than almost anyone. It failed to build the things renewables need to be useful: the wires that carry them, the storage that time-shifts them, the firm low-carbon plant that covers the windless winter evening. Until those catch up, gas keeps being the last plant in the staircase, and the auction keeps paying everyone the gas price.
Cheap clean electricity is permitted by physics. It is also permitted by markets. What Germany has spent twenty-five years proving is that it is not permitted by coalitions.
Eurostat, Electricity prices for household consumers, series
nrg_pc_204, second half 2024 and 2025. Germany was the highest in the EU at €39.43 per 100 kWh in H2 2024 and €38.69 in H2 2025, when Ireland slightly overtook at €40.42. EU‑27 average around €28.7 in both halves. ↩︎Volume share from AG Energiebilanzen (AGEB), Bruttostromerzeugung in Deutschland, 2024 annual: gas-fired plants produced 56.9 TWh of 497.3 TWh gross generation. Price-setting frequency is qualitative, summarised by Montel and EPEX SPOT as “gas is the marginal technology in most trading hours in Europe.” ↩︎
TTF month-ahead settlement, Argus Media and ICE Endex, 26 August 2022. ↩︎
Bundesnetzagentur SMARD platform, day-ahead price series 2021–2024, hourly resolution. ↩︎
Negative-price hour counts from Bundesnetzagentur (SMARD) and Montel. ↩︎
BDEW Strompreisanalyse 2025, household reference case of 3,500 kWh/year, all components including VAT. ↩︎
Netztransparenz.de (the four German TSOs), EEG-Umlage history; Bundesregierung press release of 5 July 2024 on the shift from the Klima- und Transformationsfonds to the general federal budget. ↩︎
Compressed from the standard German energy-policy chronologies maintained by Clean Energy Wire and Agora Energiewende. ↩︎
World Nuclear Association country profile, Germany, 2024 update. ↩︎
SMARD net cross-border flows, annual 2023 and 2024. ↩︎
Destatis, Produktionsindex Produzierendes Gewerbe, energy-intensive branches, table 42153; index reconstructed from year-on-year rates 2018–2025. ↩︎
BASF Annual Report 2022, capital markets release of 24 February 2023; September 2024 cost programme communicated under CEO Markus Kamieth. ↩︎
VCI (Verband der Chemischen Industrie) annual industry reports 2022 and 2023, cross-referenced with Destatis chemical industry production index. ↩︎
Industriestrompreis funding directive published in the Bundesanzeiger on 7 May 2026 following European Commission state aid clearance in April 2026. Retroactive coverage from 1 January 2026 through 31 December 2028; 5 ct/kWh on 50% of annual consumption for sectors on the KUEBLL carbon-leakage list (approximately 90 sub-sectors). Budget €3.1 billion, split €1.5 / €0.8 / €0.8 billion across payouts in 2027, 2028, 2029. Sources: BMWE press release of 16 April 2026; Gleiss Lutz, Germany cuts costs for electricity-intensive companies from 1 January 2026; Clean Energy Wire. ↩︎
Operator and ministry positions on reactivation: BMUKN FAQ on the Atomausstieg; n-tv/dpa reporting on PreussenElektra’s April 2026 statement that Isar 2 is no longer reactivable; restart estimates of three to five years and €1–3 billion per plant from Kerntechnik Deutschland via World Nuclear News. The six post-2021 shutdowns (Brokdorf, Grohnde, Gundremmingen C, Isar 2, Emsland, Neckarwestheim 2) together comprised about 8.1 GW net per the World Nuclear Association reactor database. ↩︎