A Scoop of Ice Cream

On 30 July 2004, two days before the reformed Renewable Energy Sources Act (the EEG) came into force, the federal environment minister, Jürgen Trittin, gave Germans a number to hold on to. Es bleibt dabei, his ministry quoted him, it remains the case: supporting renewable energy costs an average household “nur rund 1 Euro im Monat”, only about one euro a month, “so viel wie eine Kugel Eis”, as much as a scoop of ice cream¹. This was the message chosen for the launch, not a slip in a hostile interview, and it stuck. The programme set out on the premise that the transition would cost next to nothing.

The scoop is worth another look, because the government no longer tells the story that way. In September 2025 a monitoring report commissioned by the economics ministry found that the legal target of 80% renewable electricity by 2030 “scheint erreichbar”, seems reachable, for an unhappy reason: the capacity targets for wind will be missed, but the country will use far less electricity than had been assumed². German industry uses about 16% less electricity than in 2015 and has not returned to its level of before the energy crisis², and output per head was lower in 2025 than in 2019³. A country that makes less needs less power. The minister kept the target and made the lower demand the first of her ten measures⁴. In September 2026 the grid regulator held its first auction for new back-up power stations⁵, and the federal government is paying €6.5 billion this year to hold down grid fees⁶. None of this was in the price of a scoop of ice cream.

What follows is the tally after 22 years.

First, the part that is true

The build-out happened, and it is not small. In 2025 renewables generated as much electricity as 55.1% of what Germany used⁷. Not all of it was used in Germany: on Electricity Maps’ hour-by-hour tracing of power flows, about 16 terawatt-hours more wind, solar and biomass power left the country than came in (own calculation)⁸. Wind and solar were the two largest sources of power on the public grid for the first time; solar output grew by 21% in a single year and overtook lignite⁹. Germany has 1,404 watts of solar panels per inhabitant, more than any country except Australia and the Netherlands¹⁰, and it built them while people argued about wind turbines in every county council in the country.

Anyone who says the German renewables numbers are inflated has not read them. The panels and turbines are there, and their output shows up in the grid data. The criticism that nothing was built under the Energiewende, as Germany calls its energy transition, is wrong.

The real criticism is this: the programme was sold on two promises, low cost and clean electricity. After 22 years Germany has one of the world’s largest wind and solar fleets, but it did not come cheap and it has done little for the climate.

The scoop, priced

All of it was paid for through the EEG surcharge, a levy on every kilowatt-hour that funded guaranteed feed-in tariffs. As a description of the bill so far, Trittin’s number was right. On the economics ministry’s own table the levy came to 0.37 cents per kilowatt-hour in 2003¹¹, which is €1.08 a month before VAT for a household using 3,500 kWh a year. In 2004, the year he spoke, it was already 0.54 cents, or €1.58 a month (own calculation). The trouble lay in es bleibt dabei.

Trittin’s scoop, year by year

The EEG surcharge in cents per kilowatt-hour. At the 2017 peak a household using 3,500 kWh paid about €20 a month before VAT, twenty times the one euro Trittin had named and almost thirteen times what the levy cost in 2004. From July 2022 the levy was zero. The cost had not gone; it had moved to the federal budget.

Show the numbers

Year	EEG surcharge (ct/kWh)	A month for 3,500 kWh a year, before VAT
2000	0.19	€0.55
2001	0.25	€0.73
2002	0.36	€1.05
2003	0.37	€1.08
2004	0.54	€1.58
2005	0.70	€2.04
2006	0.89	€2.60
2007	1.03	€3.00
2008	1.16	€3.38
2009	1.32	€3.85
2010	2.05	€5.98
2011	3.53	€10.30
2012	3.59	€10.47
2013	5.28	€15.40
2014	6.24	€18.20
2015	6.17	€18.00
2016	6.35	€18.52
2017	6.88	€20.07
2018	6.79	€19.80
2019	6.41	€18.70
2020	6.76	€19.72
2021	6.50	€18.96
2022, January to June	3.72	€10.85
From July 2022	0.00	paid from the federal budget

Two things in that chart matter more than the peak.

The first is the shape. Feed-in tariffs were guaranteed for twenty years, so each year’s installations locked in two decades of payments at that year’s price. The 2004 law promised solar plants 45.7 to 57.4 cents per kilowatt-hour¹²; a new solar park now generates for 4 to 7 cents¹³. Germany is still paying for the solar of those years. Between 2000 and 2020 solar supplied 17% of the electricity the EEG paid for and caused 42% of its cost (own calculation from the ministry’s table). That cost, meaning what plant operators received above the market value of their electricity, came to €248 billion by the end of 2020¹¹. The transmission operators’ accounts add about €71 billion for 2021 to 2025¹⁴, which makes roughly €320 billion in all (own calculation). Much of that money bought early plants at the prices of their day. Its defenders add, fairly, that this early demand helped to make solar panels cheap for everybody who bought later. That is the best that can be said for it.

The second is 2022, where the chart drops to zero. But nothing became cheaper: since then the taxpayer has paid. The levy was cut to zero in July 2022 and abolished by law from 2023; the support is now “Teil des Bundeshaushalts”, part of the federal budget, as the transmission operators put it¹⁵. The federal government paid €18.5 billion into the EEG account in 2024 and €16.5 billion in 2025¹⁴. Its 2026 budget sets aside €16.2 billion under a title called Zuschüsse zur Entlastung beim Strompreis, grants to relieve the electricity price¹⁶. A further €6.5 billion goes from the Climate and Transformation Fund to the transmission operators in 2026 to hold down grid fees⁶, and the draft budget for 2027 has €5.5 billion more¹⁷.

The scoop after it left the bill

What the EEG has cost since 2000, and two sums the federal government pays in 2026 to keep electricity bills down.

€320 bn

net cost of EEG support, 2000 to 2025

payments above the market value of the electricity; own calculation

€16.2 bn

for EEG payments in the 2026 federal budget

€18.5 billion was paid in 2024 and €16.5 billion in 2025

€6.5 bn

from the climate fund in 2026, to hold down grid fees

€5.5 billion is planned for 2027

So when the government reports that electricity for new household customers got 6.7% cheaper in January 2026, it names the main cause itself: lower grid fees¹⁸. That is a price cut bought with tax money. Even so, German households paid 38.69 cents per kilowatt-hour in the second half of 2025, more than in any EU country except Ireland, and 36.18 cents in the first half of 2026¹⁹. The scoop is still being served. It is simply no longer itemised.

What the money bought, measured in grams of CO₂

The test of a decarbonisation programme is how much carbon a kilowatt-hour still carries once the clean capacity is built. The chart shows 2025, on figures that follow electricity across borders, so that imports count against the country that uses them.

Carbon intensity of electricity consumed, 2025

Grams of CO₂ equivalent per kilowatt-hour, traced along the flows of electricity so that imports count against the country that uses them. German electricity carried ten times as much carbon as French electricity and sixteen times as much as Swedish.

Show the numbers

Country	g CO₂eq/kWh, 2025	Change on 2024
Poland	676.6	−25.8
Czechia	467.7	−12.8
Germany	335.4	+0.9
Italy	285.8	+18.3
Netherlands	262.2	−1.0
Great Britain	175.9	+0.9
Spain	132.1	+8.0
France	32.0	−1.5
Sweden	21.3	−1.2

Three qualifications to this chart:

Germany does not have the dirtiest grid in Europe. Poland’s figure is twice as high, and Czechia’s and Bulgaria’s are higher too. Among the five largest economies of western Europe, though, Germany’s is the highest. Italy is not far behind, but British electricity carries about half as much carbon as German electricity and Spanish electricity two-fifths⁸. And because Germany is the largest of the five, its power stations emit more than those of any other EU country: 165 million tonnes of CO₂ equivalent in 2025 on Ember’s estimate, or 28% of the EU total²⁰.

The change from one year to the next says little. Germany’s figure rose by one gram in 2025, a year in which the Federal Environment Agency recorded clearly lower wind speeds than the year before and wind output fell by 4%²¹. A verdict read into one gram is worth as much as one read into a warm February. What counts is the level.

The level has improved. The Federal Environment Agency keeps its own series, which counts only the CO₂ from German power stations and so cannot be set beside the chart. On that series a kilowatt-hour caused 618 grams in 2004, the year of the scoop, and 344 grams in 2025, a fall of 44%²². The complaint is about what the money bought. About €320 billion of subsidy and one of the largest wind and solar build-outs of any industrial country have brought German electricity to 335 grams on the chart’s measure, against 32 in France, which decided in March 1974 to build a fleet of reactors²³.

The scoop is still being served. It is simply no longer itemised.

The order of operations

A large part of that gap has a simple cause, and it has nothing to do with wind or solar. Germany closed its zero-carbon nuclear plants before its coal plants.

Seventeen reactors held operating licences when the Fukushima accident happened in March 2011. Eight never ran again: five were switched off within days, three were already off the grid²⁴, and all eight lost their licences that August. Six more closed between 2015 and the end of 2021, and the last three on 15 April 2023²⁵. Coal was cut too, but more slowly. In 2010 Germany’s reactors generated 141 terawatt-hours and its coal plants 263. In 2025 the reactors were gone and the coal plants still generated 104, a fifth of all German electricity⁷. That is less than the reactors had produced. Between those two years annual renewable output rose by 185 terawatt-hours, and the nuclear output that was switched off equals three-quarters of that rise (own calculation).

What the choice cost has been measured. Stephen Jarvis, Olivier Deschênes and Akshaya Jha used hourly data from individual power plants to work out what replaced the reactors closed between 2011 and 2019. It was mostly coal-fired power and imports. In the Journal of the European Economic Association they put the cost to society at €3 billion to €8 billion a year²⁶. Most of it is the cost of 330 to 800 additional deaths a year from the air pollution of the fossil plants that ran instead. Climate damage is left out of the central estimate, because the plants sit under the EU’s emissions cap. Against this the authors set the avoided risk of accidents and the avoided cost of storing waste, which come to about €0.2 billion a year even on the highest values in the literature. An earlier working-paper version had put the cost at $12 billion a year and the deaths at 1,100²⁷; the refereed version revised both down.

In 2023, the year the last reactors closed, Germany also became a net importer of electricity. It imported a net 19.5 terawatt-hours in 2025⁷, though trade was nearly balanced in the first half of 2026²⁸. Electricity Maps finds that the imports of 2025 were cleaner than German generation, which on its own would have come to 362 grams rather than 335, and that about 4% of the electricity Germans used was nuclear power from abroad⁸.

A country that closed its reactors on safety grounds buys the output of its neighbours’ reactors, and its emissions figures are the better for it.

Where the CO₂ savings are coming from

Sceptics tend to overstate their case here, so the figures need care. German emissions are not rising. They fell again in 2025, to 649 million tonnes of CO₂ equivalent on the Federal Environment Agency’s figures, a decline of 0.9 million tonnes, or 0.1%²⁹.

What matters is which numbers moved. Emissions from buildings rose by 3.4 million tonnes and from transport by 2.1 million, and the energy sector’s barely changed. Industry’s fell by 5.6 million tonnes, which is what kept the total from rising. The agency names one of the main reasons itself: “geringere Produktionsmengen in energieintensiven Branchen”, lower output in the energy-intensive industries²⁹.

How German industry “decarbonised”: change since February 2022

Production, adjusted for season and working days, and employment in the branches that use the most energy. Each row runs to the latest month published. Emissions from a factory that is not running are, technically, avoided emissions.

Show the numbers

Measure	Period	Change
Chemical production	February 2022 to July 2026	−20.8%
Output of the energy-intensive industries	February 2022 to August 2026	−19.2%
All industrial production	February 2022 to August 2026	−9.9%
Employment in the energy-intensive industries	February 2022 to March 2026	−6.3% (about 53,300 jobs)

Chemicals account for 27.9% of the energy German industry uses, more than any other branch³⁰. In July 2026 the chemical industry produced 20.8% less than in February 2022³¹. The energy-intensive industries as a whole produced 19.2% less in August 2026, and in March 2026 they employed about 53,000 fewer people than four years earlier. All German industry produced 14% less in 2025 than in 2018 and has not recovered³². Industry’s emissions fell by a fifth between 2021 and 2025, from 180 million tonnes to 144 million³³.

Lower output does not by itself mean that the factories have moved abroad. Some have. In the 2026 energy survey of the German chambers of commerce, which about 3,100 companies answered, about 60% of large industrial firms said they were planning or carrying out cuts to their activities in Germany or a move abroad, and about a third said a move was already under way³⁴. That is a business lobby’s survey, not a measurement. But for the euro area an analysis by the European Central Bank found as early as 2022 that imports had replaced part of the output of energy-intensive industries³⁵. In Germany’s trade figures the shift shows where energy matters most: in 2021 the country exported slightly more nitrogen fertiliser than it imported, and in 2025 it imported 1.1 million tonnes more than it exported³⁶. There is no general flood of imports, though. Germany’s imports of chemicals, fertilisers and plastics fell from 42 million tonnes in 2021 to 38 million in 2025, and its exports fell further, from 55 million tonnes to 47 million (own calculation). Customers abroad buy elsewhere, and German factories buy less themselves. Neither is a gain for the climate or for the country: what is now made abroad causes its emissions there, and what is no longer made at all is output lost.

Agora Energiewende, a think tank that campaigns for the transition, says of the falling industrial emissions in its annual review that they are “kein Klimaerfolg”, no success for the climate³⁷.

Some of this is the gas price shock of 2022, and it would be unfair to blame all of it on the Energiewende. But the shock exposed something policy had been building for two decades: an industrial base whose advantage was cheap, reliable process energy, in a country that had deliberately, and at great expense, chosen expensive energy. In the first half of 2026 a medium-sized German industrial customer paid €20.58 per 100 kWh before VAT, against €11.77 in Sweden and €9.28 in Finland³⁸. That was already the subsidised price: six months earlier, before the grid-fee subsidy began, it was €22.64. Swedish and Finnish electricity is also far cleaner than German, at 21 and 66 grams⁸. A factory does not read the press release. It reads its electricity bill.

A country that hits its target by exporting its chemical industry to a jurisdiction with a 600-gram grid has made the climate worse and its statistics better.

A gap that cannot close at this rate

Germany’s statutory target is a 65% cut from 1990 levels by 2030³⁹. The arithmetic of the five years left shows how far behind the programme is.

35 years down, five to go

German greenhouse gas emissions since 1990 and the straight line to the 2030 target. The average cut so far is 17 million tonnes a year; in 2025 it was less than one million. The target needs 42 million tonnes a year from now on.

Show the numbers

Year	Mt CO₂e
1990	1,253.1
1991	1,206.4
1992	1,158.0
1993	1,148.5
1994	1,130.8
1995	1,123.4
1996	1,140.2
1997	1,104.8
1998	1,080.0
1999	1,045.5
2000	1,043.0
2001	1,057.1
2002	1,036.7
2003	1,030.5
2004	1,010.7
2005	988.9
2006	1,002.0
2007	963.5
2008	964.9
2009	901.3
2010	933.2
2011	907.6
2012	916.9
2013	936.6
2014	895.2
2015	898.6
2016	899.0
2017	886.3
2018	850.1
2019	798.0
2020	731.0
2021	762.8
2022	749.3
2023	669.6
2024	649.8
2025 (provisional)	648.9
2030 (target: 65% below 1990; own calculation)	438.6

From 1990 to 2025 Germany cut an average of 17 million tonnes a year. To reach the target it must now cut 42 million tonnes a year, as the Federal Environment Agency says itself²⁹. That is about 2.4 times the average pace of 35 years, for five years in a row, starting from a year in which emissions fell by less than one million tonnes. And the two sectors whose emissions rose in 2025, transport and buildings, have nothing like the EEG behind them.

The agency’s own projection ends at a cut of 62.6% in 2030, yet it calls the target reachable with additional measures⁴⁰. Germany’s Council of Experts on Climate Change, which reviews these figures by law, examined the same projection in May 2026 and expects Germany to emit more between 2021 and 2030 than the law allows⁴¹. The first half of 2026 looks a little better: energy-related CO₂ emissions fell by about 2.5%, on a rough estimate by the statisticians of AG Energiebilanzen²⁸. The target needs all emissions to fall by more than 6% a year (own calculation).

The 1990 baseline has a wrinkle that German officials rarely volunteer. Almost half of the reduction in the first decade came from the collapse and restructuring of the East German economy after reunification, on the estimate of a study for the Federal Environment Agency; policy measures, the same study found, accounted for slightly more than reunification did⁴². Those emissions really are gone, but their disappearance was not a policy achievement. France cut its emissions by 33% between 1990 and 2024, against Germany’s 48%⁴³, but it started from a grid that was already decarbonised. Comparing the rates flatters Germany. Comparing where the two have arrived, 335 grams against 32, does not.

The bill that has not arrived yet

Everything above is cost already incurred. The grid is the cost still to come.

Wind farms are built where the wind blows, and the factories are somewhere else. SuedLink, the main new line from Germany’s north to its south, is 693 kilometres long and is due to carry power from 2028⁴⁴. The grid comes with two bills, one for the high-voltage transmission network and one for the local distribution grids. The transmission operators’ current plan, a second draft of March 2026, puts their network on land and at sea at €365 billion to €392 billion by 2045⁴⁵. That is roughly €50 billion to €75 billion less than the last price put on the previous plan, about €440 billion². The distribution grids need well over €235 billion on their operators’ plans, and the monitoring report expects that figure to rise, because a fifth to a quarter of the low- and medium-voltage network is not yet in those plans and recent price increases are not either². Together that is at least €600 billion over the next two decades (own calculation), nearly twice what the EEG has cost since 2000. Running the grid that exists is getting dearer too: managing its bottlenecks cost €1.3 billion in 2019 and €3.1 billion in 2025⁴⁶.

The monitoring report is most revealing on the 2030 target. It found the target of 80% renewable electricity still within reach although the wind targets will be missed, because Germany will use 600 to 700 terawatt-hours in 2030 and not the 750 that official supply-security assessments had assumed². The government now says so in its own words. Its bill to reform the EEG, sent to the Bundestag in September 2026, says that lower demand and lower renewable output roughly cancel each other out. It deletes the law’s year-by-year path for renewable electricity, because the path had feigned a precision, a Scheingenauigkeit, that events did not bear out⁴⁷. The target survives because the denominator shrank.

The denominator shrank because the machines that were meant to run on the new electricity are arriving far more slowly than planned, and because German factories are making less. Germany used 529 terawatt-hours in 2025, as much as in 2024⁷; reaching even 600 by 2030 means growth of 13% in five years (own calculation). Heat-pump sales recovered to 299,000 in 2025 but stayed below the 356,000 of 2023⁴⁸. There were 2.4 million battery-electric cars on the road in July 2026⁴⁹, against a goal of 15 million by 2030². Electrolysers with a capacity of 181 megawatts had been installed by January 2026, against a goal of 10 gigawatts⁵⁰. A target that is met because the country electrified less than planned is not the success the target was meant to measure.

Time for a change, but the right one

Nothing here argues for tearing out solar panels. Solar parks and onshore wind are now the cheapest power stations Germany can build, at 4 to 9 cents per kilowatt-hour¹³, and having once paid too much for them is no reason to stop building them now that they are cheap. What needs to change is the accounting.

Publish euros per tonne. Every Energiewende instrument should carry a published abatement cost. Solar plants installed under the EEG in the late 2000s avoided CO₂ at more than €500 a tonne on two separate estimates⁵¹⁵², when an allowance in the EU’s emissions trading system (the ETS) cost about €11. Wind was a different story: at about €50 a tonne it cost less than an allowance does today⁵³. A programme that does not publish its cost per tonne leaves others to work it out.

Fix the order of operations. The sequence should have been coal first, then gas, then nuclear, if at all. It was run the other way round, and a good part of the 335 grams is the price. Restarting anything is another matter. In March 2026 the chancellor, Friedrich Merz, said of the phase-out: “Der Beschluss ist irreversibel. Ich bedauere das, aber es ist so.” The decision is irreversible; he regrets it, but that is how it is⁵⁴. The six reactors that closed in 2021 and 2023 all hold decommissioning licences⁵⁵. The least Germany can do is not repeat the error by closing firm capacity before there is something to replace it.

Stop counting deindustrialisation as progress. Emissions should be reported per unit of output as well as in absolute terms. A country that hits its target by exporting its chemical industry to a jurisdiction with a 600-gram grid has made the climate worse and its statistics better.

Count the storage. The gap between a grid that is 55% renewable and one that is 90% renewable cannot be closed with more of the same. A plan that does not say where the power comes from in a windless winter week is a plan to burn gas, whatever else it says. The regulator’s auction for plants that can run through long spells without wind and sun acknowledges as much⁵.

Price carbon and stop picking winners. The ETS does with one instrument what twenty years of technology-specific subsidy did expensively and unevenly, and unlike a feed-in tariff it does not care which technology wins.

The sentence, put right

As a description of what households had paid up to then, Trittin’s line was accurate. Asked about it in 2013, he wrote that his ministry’s statements had referred to “den Istzustand”, the state of things at the time, or to the few years then foreseeable, and that he had expected renewables to supply 12.5% of electricity by 2012, not the almost 22% they did supply⁵⁶. That is a fair defence of the arithmetic. But the line came with the words es bleibt dabei, and it was the line, more than the arithmetic, that sold a twenty-year commitment.

Written with hindsight, the 2004 line would have read something like this:

Supporting renewable energy cost the average household about a euro a month last year and costs a euro and a half this year. Within thirteen years it will cost twenty euros, and after that it will disappear from the electricity bill. But it will still be paid, out of taxes, through the federal budget. Over 22 years it will buy one of the largest wind and solar fleets in the world, the second-highest household electricity price in the European Union, and electricity that still carries ten times as much carbon as France’s, because we will close the reactors first and the coal plants last.

That sentence would not have sold the programme, which is why it belongs on the record now, before the next twenty-year commitment is priced in confectionery.