Feed-in compensation 2027: what your solar surplus is worth at midday

Exactly zero. Not a rounded zero and not almost zero: €0.00 per megawatt-hour, published as the clearing price on the Dutch day-ahead market. Across 2021 through 30 August 2026 there are 304 hours at precisely that figure. The second-busiest price bin in the entire neighbourhood reaches 113.75 hours.

In four months you will be selling your solar surplus on exactly this market.

Dutch net metering ends on 1 January 2027. In the comments under our videos, what comes after that is called a penalty. You will pay for exporting power, as punishment for owning panels. Fourteen comments across nine videos say it in those words, between 3 January 2025 and 26 July 2026. The heaviest sits at nine likes.

That penalty is not coming. What is coming is quieter, and it has been sitting in the market's own price list for years.

This article is the dossier: the full measurement with the script that lets you repeat it, the 401 price bins one cent wide, the four suppliers who do or do not publish their own 2027 figures, and the point where my explanation runs out. That last part is here deliberately. I know the peak is there. I do not know why.

Key takeaways

• Net metering stops completely on 1 January 2027. The scheme falls away in one step, including for installations that have been on the roof for years.
• Feed-in costs continue after that date, and they already exist today. Vandebron introduced them in 2023, Eneco in June 2024. Two separate things get fused together: a surcharge from your supplier, and the end of a fiscal offsetting scheme.
• The statutory floor runs from 1 January 2027 through 31 December 2029, and it binds the compensation, not the balance. At least 50% of the bare supply tariff, and the Dutch regulator ACM states in so many words that feed-in costs can come out higher than that compensation.
• Two suppliers publish their own 2027 figures, two do not. Greenchoice works with €0.05434 compensation minus €0.05184 feed-in costs. Eneco with €0.05058 minus €0.03865. Vattenfall and Essent name no amount on their own 2027 page.
• Exactly €0.00 is the most common electricity price on the Dutch day-ahead market, every single year. 304 hours across six years, against a median of 1.50 hours per one-cent-wide price bin. An excess of 203 times.
• Why that peak is there has not been established. The subsidy rule that circulates as the explanation only applies from 200 kW per connection and touches no household at all. The attribution is refuted with two figures, and I have no second explanation I can stand behind.
• The cheapest hour of the day is 13:00, the most expensive 20:00. In 2026 there is €167.1 per megawatt-hour between them, against €98.9 in 2024. That gap is growing, and it decides what storage is worth.

What changes on 1 January, and what does not

Net metering disappears in one step. The Dutch senate voted on it on 17 December 2024, and nothing was put in its place as a phase-out. Through 31 December 2026 you may offset every exported kilowatt-hour against a kilowatt-hour you draw. From 1 January 2027 you cannot.

What does not change are the feed-in costs. Most of the confusion sits on that item, so here are the sources side by side. Every quotation in this article is translated from Dutch; the originals sit on the pages listed under Sources.

ACM ConsuWijzer, the regulator:

"You (still) pay feed-in costs. Suppliers incur extra costs for consumers with solar panels. From 2027 they must charge those costs to that same group."

Essent, answering its own question about whether they will stop:

"No, not yet. If everyone exports power at the same time, that still causes extra costs."

Eneco, under the heading Feed-in costs continue:

"In 2027 too we will charge feed-in costs on the power you export."

Greenchoice, with a nuance the other three do not offer:

"Because energy suppliers still face costs caused by generated power, such as profiling and imbalance costs, feed-in costs remain. Those costs do drop slightly, because the net metering costs fall away."

Five mutually independent sources, four of them commercial parties with opposing interests, and they say the same thing. Feed-in costs survive 1 January.

The second half of the correction weighs more than the first: those costs already exist. Vandebron writes about itself that it was "the first energy supplier in 2023" to introduce them. Eneco followed in June 2024. Anyone opening a 2026 annual statement from one of those suppliers finds the line item sitting there, three years before anyone started talking about 2027.

So what happens on 1 January is one thing, not two: the offsetting scheme falls away. The surcharge you already pay stays, and according to Greenchoice it even drops slightly, because the net metering component comes out of it.

The misconception is measured, not assumed

I wrote above that people expect a penalty. That comes from a count over our own comments: 14 comments across 9 videos, from 3 January 2025 to 26 July 2026, calling the feed-in costs a penalty or a punishment. The heaviest sits at nine likes, against a median of zero across the whole corpus, a mean of 1.16 and a p95 of 5.

"I'm afraid we'll soon have to throttle what the panels produce, if exporting with a penalty starts costing money."

That comment comes from someone who knows the rules. It puts two events on one pile because they appeared in the same news coverage, and that is an understandable mistake.

The mirror form is in there too, once, at zero likes: "this is going away next year so the so-called export penalty stops." One comment is not a cluster and I do not use it as evidence for anything. It is here because it explains the first fourteen. If you read the surcharge as a penalty for net metering, then it follows that you expect it to leave when net metering leaves.

It does not leave.

The measurement: 401 price bins, six years of day-ahead

The method, so you can follow it. The series comes from the ENTSO-E Transparency Platform, documentType=A44 (day-ahead prices), bidding zone NL (10YNL----------L). Every period is weighted by its own length rather than counted as an occurrence, because since 1 October 2025 the market publishes quarter-hourly prices and before that hourly ones. Missing positions in curveType A03 are carried forward, as the specification prescribes. All prices then go into 401 bins one cent wide, from −€2.00 to +€2.00 per kilowatt-hour.

The measurement script lives in the repo as tools/meten/entsoe-nulpiek.mjs. Run it again and every figure below comes back.

The calibration. A private measurement nobody can check is an opinion with decimals. This one is therefore pinned to two figures other people have already published: 458.00 negative hours in 2024 (pv-magazine and ANWB) and 581.25 in 2025 (De Datadame). Both come out to the decimal. The series is also cross-checked against the energy-charts.info API, which pulls the same ENTSO-E data along a different route, and that matches to within 0.01 hours across 2021, 2022, 2024, 2025 and 2026. The annual totals add up: 8,760 / 8,760 / 8,760 / 8,784 / 8,760 / 5,807 hours, with no gaps.

The two figures in bold are the calibration points.

Across the whole series, 304.00 hours sit at exactly €0.00. The median one-cent-wide price bin in that same series reaches 1.50 hours. That difference is a factor of 203.

Year by year the picture is the same. In 2024, 84.00 hours clear at €0.00, against a median of 0.400 hours per cent tick for prices sitting between 20 cents and 2 euros away from zero: an excess of 210 times. In 2025 that is 145 times, in 2026 through August 137 times. And to put it against a price you would recognise: the busiest price bin outside the zero zone in 2024 is €100.00 per MWh, with 33.00 hours. Zero gets 84.00.

The neighbourhood around zero is not flat, and that is a correction on myself

When I first wrote this up, I thought the peak rose out of a smooth, flat band. One needle in a flat desert. That is how it was drawn in the graphic we had built for the video, too.

It is wrong, and the graphic was withdrawn on 30 August 2026.

Of the 400 neighbouring bins, 369 are filled, and they span two orders of magnitude. These are the twelve busiest price bins in the whole window:

The nearest neighbour, −€0.01, reaches well over a third of the peak. And −€2.00 sits high in the list, which is not a price phenomenon but a bidding one: a round bid placed exactly on the edge of the measurement window.

Bundled per ten cents the pattern gets clearer:

Zero stays the mode by a wide margin. Against its nearest neighbour the ratio is 2.7 times, against the median bin 203 times. But anyone drawing a flat band with a single needle is drawing something other than what was measured.

Three checks that keep the peak standing

A point mass on a round number is precisely the kind of thing a measurement error would produce. Three checks.

It is not a data gap. If a market data provider filled missing values with zero, you would see long contiguous blocks spread randomly across the day. What is there is the opposite: 65 separate episodes in 2024, 58 in 2025 and 129 in 2026, of which 58, 50 and 86 respectively last exactly one period. The longest run is 8, 4 and 9 periods. They cluster between 10:00 and 15:00, and in the months March through September. That is the profile of solar production, not of a data outage.

It is not a rounding artefact. ENTSO-E publishes prices with zero, one or two decimals, exactly as the exchange delivers them. In this series 5,504 raw values carry no decimal point at all. So a 0 in that file is a genuinely published clearing price, not a display of something small rounded down.

And the check that argues against me. Round numbers are broadly over-represented in this market. Between 10 and 12 percent of all hours clear on an exact whole euro per MWh, where an even spread across cents would put you around 1 percent. On that roundness measure, €0.00 is not an outlier but an ordinary member of a larger family. That does not make the peak smaller, and it does make it less remarkable than it looks at first glance.

The honest floor: the explanation I had does not hold

There is an explanation circulating for this peak, and I believed it myself. Since the 2023 application round, the Dutch SDE++ subsidy does not pay out during hours with a negative price. A producer running on that subsidy then has a reason to place a bid at exactly zero instead of below it. Sounds airtight.

It has three holes.

The first is the threshold. That SDE++ rule applies from 200 kW per connection. A household with solar panels falls under none of the three regimes, and even a sizeable commercial array on a barn roof usually sits below it. The rule therefore only touches large-scale generation.

The second is the timeline, and it points the wrong way. If the 2023 tightening caused the peak, the share of zero hours should have risen afterwards. It fell. In 2022, 29.4 percent of all negative hours were hours at exactly zero. In 2025 that is 11.5 percent. Both percentages share the same denominator: the negative hours of that year.

The third is the neighbour. Bidding zone DE-LU shows very nearly the same picture in 2025. There, 86.75 hours sit at exactly €0.00, or 0.990 percent of all hours in that year, against 0.765 percent in the Netherlands. Germany runs a different subsidy system. If the Dutch rule made the Dutch peak, the German peak should not be there.

The rule exists. The peak exists. The link between them has not been measured, and with this data it cannot be: price data alone does not separate a bidding convention from a subsidy kink. What would settle it are the aggregated bid and ask curves per auction, which EPEX SPOT publishes. I did not retrieve those for this dossier.

So: I know that it is the case, and not why.

Why your surplus is worth so little at exactly that hour

The zero hours are the sharpest form of a broader pattern. The hour your panels work hardest is the hour the market pays least.

The average price level rises, the cheapest hour gets cheaper, and the most expensive hour gets more expensive. The distance between those two has grown by two thirds in two years.

The Netherlands does not consume less at midday. In June 2026, demand at 13:00 averaged 13,873 MW, higher than the night-time floor of 11,241 MW at 05:00. There is no midday dip in consumption.

The dip sits in the residual demand: the part conventional power stations still have to deliver after solar and wind have made their contribution. In that same month of June, gas-fired capacity ran from an average of 1,347 MW at 14:00 to 5,650 MW at 21:00, a factor of 4.2. Coal went from 632 to 1,479 MW. At midday the Netherlands exports roughly 3.5 GW because the surplus has nowhere to go here.

The evening peak in the price is therefore not a peak in what you and your neighbours switch on. It is the moment the sun is gone and the gas has to come up, while consumption barely moves.

And a claim you will run into often that does not hold: that the number of hours with negative prices keeps rising. Over the equal period from 1 January through 28 August, it went from 390 hours in 2024 to 472 in 2025 to 351.3 in 2026. That is a drop of 25 percent against last year.

What does rise is the spread within the day. Fewer hours below zero, a larger gap between midday and evening. For anyone who can store something, the second movement matters more than the first.

One month of June, seen per quarter-hour

Since 1 October 2025 the day-ahead market trades in quarter-hours instead of hours. That changes how you have to read every report about a "number of hours at zero": four quarter-hours at €0.00 add up to one hour, not four. Anyone mixing those two up multiplies their own conclusion by four.

To show what such a day looks like up close, here are three days from June 2026, pulled per quarter-hour:

On Tuesday 9 June the price runs from almost nothing to €164.12 per MWh within a single day. Same day, same meter, same panels. What differs is the hour the kilowatt-hour arrives.

And one correction on myself: an earlier version said you see this on any Wednesday in June. In the whole of June 2026, no Wednesday has a single quarter-hour at exactly €0.00. That has been checked and fixed.

The 50% floor binds the compensation, not the balance

Here sits the part of the law that gets misread most often, and it is the reason the previous sections matter.

From 1 January 2027 through 31 December 2029, your supplier must offer at least 50 percent of the bare supply tariff as feed-in compensation. Bare means excluding energy tax and VAT. Vattenfall states it about its own contract in exactly those terms: from 1 January 2027 through 31 December 2029.

Two things about that floor.

It does not exist before 1 January 2027. That sounds like a detail and is not. Under net metering there is nothing for it to protect, because your surplus is offset one for one. The floor is new policy that starts at the moment net metering ends. Anyone reading "until 2030" without the start date attached easily concludes there is already a floor under their compensation today.

It binds what you receive, not what you keep. ACM writes directly beneath the sentence about the floor:

"Note: these feed-in costs can sometimes come out higher than the compensation for the electricity you export."

That is the regulator itself, and it means exporting power can cost you money on balance. That is not prohibited and it does not conflict with the floor, because the floor covers the compensation and the costs sit next to it.

ACM adds a worked example: 1,000 kWh × €0.15 = €150 compensation, minus €70 feed-in costs, leaves €80.

What two suppliers publish for 2027 themselves

On 29 August 2026 I went through the 2027 pages of four large suppliers. Two name an amount, two do not.

Status: proposal, not fact. These are supplier estimates on a supplier's own site, not established tariffs. Greenchoice attaches the disclaimer verbatim: "No rights can be derived from this example. It is intended purely as an indication." Eneco names its basis just as explicitly: "We calculate using the tariffs of a 1-year fixed contract and an estimate of feed-in compensation and feed-in costs as of 1 January 2027 (reference date 25 February 2026)."

So read these figures as "what they are working with", never as "what you will get".

Eneco pays exactly the statutory floor, and you can verify it. Their bare supply tariff sits at €0.12240 per kWh on a 1-year fixed contract, reference date 25 February 2026. Divide that by 1.21 to strip the VAT and you get €0.101157. Half of that is €0.050579. To five decimals, that is the amount they state: €0.05058. So the 50% floor is not a floor they sit above; it is the tariff itself.

What the other two suppliers do say. Vattenfall names only the rule: "Until 2030 you get at least 50% of the bare supply tariff back for the power you export." Essent writes "We will inform customers in good time about the new feed-in compensation and costs", and separately announces a change in form that makes comparison harder: "From then on we calculate feed-in costs per kWh instead of in tiers." That last point is worth remembering for next year, because a tiered structure and a per-kWh tariff cannot be compared directly.

The worked example, and the four things it does not say

Take an installation generating 6,000 kWh a year, of which 4,000 is used directly in the house and 2,000 goes onto the grid. All three rows below use the tariffs of one supplier, Greenchoice, because they are the only one publishing both the supply tariff and both 2027 figures on the same page.

Between the first and the second row sits a factor of 99. At Eneco the net margin is roughly four times better (€0.01193 against €0.00250), and that is still a factor of 21 below what net metering delivers today.

What is already settled and comes from my own measurement. After 2027, exporting at midday earns you roughly the midday spot price, which in 2026 averaged €36.5 per MWh or 3.65 cents per kWh. Anyone keeping that same kilowatt-hour and using it in the evening avoids buying on an hour that averages €162.3 per MWh, or 16.23 cents. Both are wholesale prices without tax, VAT and grid charges, so neither is an end price for you.

Feed-in costs are attached to the contract type

There is one layer underneath this whole story that rarely gets mentioned: not every supplier charges feed-in costs, and it is tied to the type of contract. On 29 August 2026 I checked four suppliers of dynamic and market-following contracts, each on their own page.

Zonneplan names the mechanism itself, and that part is usable: the surcharge is attached to the contract type, and was introduced "on fixed and variable contracts".

This does shift who this article actually concerns. The fear of a penalty is a story for people on a fixed or variable contract, because that is where the surcharge sits. The inversion below concerns exactly the group that does not have it. No contradiction: you fear a cost item attached to your contract type, while the real cost item is the hour your surplus arrives.

The inversion: at the zero hour, the earner is whoever can absorb

Put the two halves side by side.

On a sunny June afternoon at 14:00, the clearing price on the day-ahead market is sometimes exactly zero. Whoever pushes power onto the grid at that moment gets nothing for it. Whoever draws power from the grid at that same moment pays nothing for it at wholesale.

Same chart, two sides.

That makes exporting an act with a price you do not choose, and absorbing an act whose timing you do choose. A home battery, a heat pump running ahead of schedule, a car sitting on the driveway: all of them are ways to do the second instead of the first. What they are worth does not depend on the compensation your supplier offers, but on the distance between the midday hour and the evening hour. And that distance grew from €98.9 in 2024 to €167.1 per MWh in 2026.

An important limitation, because this is where good stories about home batteries fall over: at your meter those amounts are smaller. Energy tax, VAT and grid charges sit on top and do not move with the hour. The shape of the day is real; the spread at the meter is smaller than the spread on the wholesale market.

Net metering was storage, and where that comparison ends

There is a way to understand what disappears on 1 January, and in accounting terms it is exact.

Net metering was storage. You put a kilowatt-hour away in June and took it back in December at the same price. The storage location was not a battery but a line on your supplier's annual statement. ACM ConsuWijzer writes that it considers it "desirable" that your supplier offsets across your whole annual consumption, and that there are "multiple rulings" that offsetting must be annual, with the caveat that the law does not fix the period and that it is set in your contract.

On 1 January that storage moves from their ledger to your meter cupboard. Or nowhere at all.

Four points where the comparison ends, and every one of them favours the paper version over the hardware:

• On paper there is no round-trip loss. A real battery loses a slice on every cycle. An annual statement does not. Getting one for one back is better than any accumulator can deliver.
• The accounting store was infinitely large and cost nothing to buy. A battery has a capacity and a price.
• No home battery bridges half a year. The summer-to-winter shift that net metering performed goes away on 1 January and does not come back. What hardware replaces is the daily cycle, not the seasonal one.
• It was not free any more anyway. Feed-in costs have existed since 2023, so the years in which exporting cost you nothing are already behind us.

Anyone who does nothing on 1 January keeps roughly what the market pays at the midday hour. What that is, has been measured above.

What is deliberately not in this article

This section belongs here, because the questions I could not close say as much as the answers I have.

• Why the peak at €0.00 is there. The SDE++ attribution is refuted with two figures, and I have no second explanation I can stand behind. Price data alone does not separate a bidding convention from a subsidy kink. What would settle it: the aggregated bid and ask curves per auction at EPEX SPOT. I did not retrieve those for this dossier.
• How many households hold a fixed, variable or dynamic contract. No source found. The ACM page I tried returned a 404. The route to an answer: the ACM Monitor consumentenmarkt energie, or CBS. As long as that figure is missing, nothing in this piece says dynamic is the norm or that fixed contracts are on their way out.
• Whether ACM's "must" is a legal obligation. ACM writes that from 2027 suppliers must charge the extra costs to the group with solar panels. Whether that describes a statutory duty that also binds dynamic contracts, or an economic necessity, has not been established. ACM cites no article of law, and wetten.overheid.nl returned only the table of contents when fetched.
• The feed-in costs at Tibber and Frank Energie. Three URLs were tried at Tibber without a definitive answer, and at Frank Energie the pages searched say nothing about it. That means "I did not find it", and it is not a statement about their tariff. If you want certainty, ask their customer service and get it in writing.
• How many kWh a kWp yields per year in the Netherlands, and how much self-consumption a battery adds. Not checked for this piece, and therefore not built into the worked example either. That example runs on an assumed split of 6,000 and 4,000 kWh and presents itself as such.

Frequently asked questions

Related articles

• The end of Dutch net metering in 2027: what to do now for the timeline, the legal background and the worked household scenarios
• Is a home battery worth it in 2026? for the payback calculation deliberately left out of this piece
• Trading with a plug-in battery for what you can do with the same price spread on a dynamic contract

Transparency

No sponsorship, no affiliate links, no brand preference. Nothing was lent, given or paid for this article. No product I sell appears in it, and no supplier saw this piece in advance.

The price series is my own measurement, and the measurement script sits in this project's public repo as tools/meten/entsoe-nulpiek.mjs. The outcome is calibrated against two figures others published earlier and cross-checked against a second data source. Where a figure comes from a supplier, the supplier is named along with the date I retrieved the page. Where I had to correct an earlier version of this dossier, the correction is in the text rather than underneath it.

Sources