Between January and September 2026, German solar plants fed 76.2 TWh into the public grid. In the same nine months of 2025 the figure was 63.0 TWh. On the simplest reading of how solar surpluses push day-ahead prices below zero, a 21% rise in output should have produced more negative-price hours. It produced fewer. The DE-LU day-ahead price was negative for 463.25 hours in the first nine months of this year, down from 525 hours in 2025, a fall of about 12%.
Those counts are TEI calculations from the day-ahead price and generation series published by the Bundesnetzagentur on SMARD, downloaded through the Fraunhofer ISE Energy-Charts interface on 23 October 2026. The 2025 prices are hourly. From 1 October 2025 the day-ahead auction moved to 15-minute products, so the 2026 series has 26,205 quarter-hour prices; we counted negative quarter-hours and divided by four. Generation is quarter-hourly in both years. Periods run from 1 January to 30 September, local time.
The count, month by month
The fall is uneven, and most of it sits in early summer. Negative hours by month, 2025 then 2026:
- January: 14 and 2.5
- February: 0 and 5.75
- March: 30 and 34.75
- April: 75 and 123
- May: 129 and 78.75
- June: 141 and 49.75
- July: 12 and 76.5
- August: 64 and 56.5
- September: 60 and 35.75
April 2026 was worse than April 2025 by 48 hours, and July 2026 was much worse than a July 2025 that was unusually dull. May and June went the other way, by a combined 141.5 hours. Monthly solar output does not explain that by itself: June 2026 generation was 11.6 TWh against 10.0 TWh a year earlier, and May was 11.5 TWh against 9.8 TWh. More sun went into the grid in exactly the months where negative pricing receded most.
The number of days with at least one negative period fell from 96 to 90. Prices at or below zero, which picks up the many periods that clear at exactly €0/MWh or a fraction under, covered 551 hours this year against 603.
Fewer hours, deeper troughs
The averages point the other way once a price does go negative. The mean of all negative prices was minus €16.83/MWh in 2026, against minus €11.84/MWh in 2025. The single lowest price last year was minus €250.32/MWh, at 13:00 on Sunday 11 May. This year the market cleared at minus €499.99/MWh for five consecutive quarter-hours from 13:15 on 1 May (13:15 to 14:30 local time), a public holiday, close to the harmonised floor of the coupled day-ahead market. Prices below minus €200/MWh also appeared across most of the early afternoon of Sunday 26 April. The longest unbroken stretch of negative prices was 17 hours this year and 14 hours in 2025.
So the distribution has changed shape. Low-demand spring weekends and holidays still produce extreme clearing prices when solar peaks against thin load. Ordinary sunny weekdays in May and June produced fewer negative periods than last year.
Part of the reason is visible in the same SMARD series. We averaged each generation and flow line across the periods when the day-ahead price was negative:
- Solar output averaged 37.5 GW during negative periods in 2026, against 31.7 GW in 2025.
- Net exports (the cross-border line, where negative values mean Germany is sending power out) averaged 9.5 GW, against 7.0 GW.
- Pumped-storage pumping averaged 5.0 GW, against 4.6 GW.
- Lignite output averaged 2.8 GW, against 3.2 GW.
- Grid load averaged 48.3 GW, against 49.6 GW.
Germany is absorbing a bigger solar surplus in those hours mainly by moving it abroad, with some help from pumping and slightly less lignite running underneath. The residual load series, load minus wind and solar, averaged minus 3.6 GW during this year's negative periods; last year it was plus 4.1 GW. SMARD's public series used here do not include home or utility batteries as a separate line, so any battery contribution is hidden inside load and cannot be measured from these data. Neighbouring markets' own solar and prices are outside this calculation too, and a full explanation of the May and June drop would need them.
What a negative quarter-hour costs a new solar plant
The policy side changed while this was happening. The Act amending energy law to avoid temporary generation surpluses, published in the Bundesgesetzblatt on 24 February 2025 (BGBl. 2025 I Nr. 51) and known as the Solarspitzengesetz, rewrote section 51 of the Renewable Energy Sources Act (EEG 2023). Section 51(1) now reads in full: for periods in which the spot market price is negative, the applicable value is reduced to zero. There is no longer a minimum run of consecutive hours before support stops. Section 51(2) exempts plants under 100 kW until the end of the calendar year in which a smart meter is installed, and plants under 2 kW until the Bundesnetzagentur makes a determination under section 85.
Section 51a is the compensation. A plant whose payment is cut under section 51 has its 20-year support period extended. For solar, section 51a(2) halves the number of affected quarter-hours, rounds up, and converts the result into "full-load quarter-hours" using a fixed monthly table: 87 for January, 189 for February, 340 for March, 442 for April, 490 for May, 508 for June, then 498, 453, 371, 231, 118 and 73 for July to December. The support period runs on past its original end until that allowance is used up, and then to the end of that month.
Applying that to this year's data gives a rough sense of scale. There were 1,853 negative quarter-hours in the DE-LU day-ahead market from January to September 2026. If every one of them counted for a plant under section 51, halving gives 926.5, rounded up to 927 full-load quarter-hours of extension owed for nine months of 2026 alone. How long that extension lasts depends on when the original period ends. If it ends on 31 December, the allowance runs through January (87), February (189), March (340) and part of April, so the extension reaches the end of April: four months. If it ends on 30 April, May (490) and June (508) cover it in two. The table is weighted towards summer, which appears designed so that the extended months deliver roughly the solar output the lost quarter-hours would have.
That calculation is an upper bound for any specific plant. The extension counts negative quarter-hours whether or not the plant was generating in them, but it is only owed to plants that fall under the current section 51, and section 51(2) keeps most small rooftop systems outside it until they get a smart meter. Older plants that were commissioned under earlier versions of the EEG keep their own rules unless they opt in. Section 100(47) lets an operator declare in writing that sections 51 and 51a should apply, no earlier than the end of the year in which a smart meter is fitted, and in return raises the plant's applicable value by 0.6 cents per kWh.
The same law also capped feed-in for new small plants without control equipment. Section 9 of EEG 2023, as amended, requires plants between 25 kW and 100 kW on a feed-in tariff to limit grid injection to 60% of installed capacity unless the network operator can curtail them remotely, with parallel limits below 25 kW. Section 12(2a) of the Energy Industry Act, inserted by the same Act, obliges every network operator to be able to curtail plants of 100 kW and above and read their actual output, and from 1 January 2026 the annual tests under section 12(2b) must include plants below 100 kW as well.
The capture rate moved too
A volume-weighted average of the day-ahead price over each quarter-hour of solar output gives the price solar actually earned on the spot market. On our calculation that was €55.70/MWh for January to September 2026, against a simple average baseload price of €107.72/MWh, a capture rate of 51.7%. In 2025 the figures were €41.96/MWh against €88.03/MWh, or 47.7%. The share of all solar output that was generated during negative-price periods fell from 26.4% to 22.8%.
The higher baseload level most likely reflects this year's gas price shock, which lifts evening and night prices far more than the midday hours where solar sits, and that alone would compress the capture rate. The rate rose anyway, by four percentage points. The cannibalisation of solar's market value, which has been the main commercial worry for unsubsidised German solar, did not get worse in 2026 even though 13 TWh more solar was sold into the same hours.
A 51.7% capture rate is still low, and the spring weekend troughs show the market can clear near minus €500/MWh when load is thin. Under section 51a(3) the power exchanges must report each year's negative quarter-hours to the transmission system operators by 15 January, and section 51a(4) requires the operators to publish the count by 31 January. The first full calendar year counted in quarter-hours is 2026, so that figure is due in January 2027.
Method notes
Prices: DE-LU day-ahead, SMARD (Bundesnetzagentur), via the Energy-Charts price API, 1 January to 1 October in each year, Europe/Berlin time. 2025 has 6,552 hourly values and 2026 has 26,205 quarter-hour values; no gaps were found. Negative means strictly below zero. Hours for 2026 are quarter-hours divided by four. If 2026 is instead aggregated to hourly averages, 472 hours were negative; 558 hours contained at least one negative quarter-hour.
Generation: SMARD public net generation by source via the Energy-Charts public power API, quarter-hourly, same windows. Solar capture price is the sum of solar output times price over the sum of solar output, with 2025 hourly prices applied to each of the four quarter-hours. Negative-period averages are simple means of each series over the negative periods. The cross-border series sign follows Energy-Charts, with exports negative.