856 GWh outside the monthly series: the unattributed remainder of output

How much electricity in Slovakia is produced by sources that have no line of their own in the monthly statistics? The question comes straight out of the table: the series „Electricity sources in GWh“ gives output and three types of plant, but their sum does not give output. For 2025 the unattributed remainder is 856 GWh and for January to May 2026 already 386 GWh (ŠÚ SR; DATAcube, table en1001ms, updated 31. 7. 2026, retrieved on 23 August 2026).

This text is a calculation from zero out of one public table, not an assessment of energy policy and not advice on reporting. It does not state which technologies the remainder consists of — this series has no line for that — and every assumption used is named.

The question: how much is produced by sources without a line

The table has seven indicators. The first is „Electricity output“, followed by three lines introduced with the words „of which plant“: thermal, hydro and nuclear. Then come „Imports“, „Exports“ and „Sources in total“. The phrase „of which“ suggests a part of a whole, yet nowhere is it stated that the whole listed is complete.

That is exactly the subject of the calculation. If the three named types are the only plants, their sum should equal output. If it does not, output exists that belongs to none of them, and its size can be established by subtraction — without any external input and without an estimate.

The aim is not to name a technology. The aim is to establish the size of the difference, to check whether it is rounding, and to show how the result changes with what it is divided by and over what period it is counted.

The input data and where they come from

One table of one source is used. The year dimension holds 21 values from 2006 to 2026, the month dimension holds 13 values — twelve months and the annual sum from January to December — and the indicator dimension holds seven items. The unit is a gigawatt-hour per calendar month.

For 2025 the annual values are 28 883 GWh of output, 5 271 GWh thermal, 3 493 GWh hydro and 19 263 GWh nuclear; imports were 15 968 GWh, exports 18 320 GWh and sources in total 26 531 GWh. The data come from the machine output of table en1001ms; by hand they open in the DATAcube database.

The choice of this particular table is itself an assumption that has to be named. The annual energy balances of the same office work in terajoules and with a different breakdown of indicators, so their values are not directly comparable with this series. Everything that follows comes solely from the monthly series in gigawatt-hours and no conversion between units is made anywhere.

Step 1: checking whether the annual line matches the sum of months

Before subtracting, it has to be known whether the annual value equals the sum of twelve months or whether it is separately collected. The check is made inside the table and needs nothing else: the monthly values of a given year are added up and compared with the value in the thirteenth position of the month dimension.

For 2025 the sum of twelve months gives 28 883 GWh of output, 5 271 GWh thermal, 3 493 GWh hydro, 19 263 GWh nuclear, 15 968 GWh of imports, 18 320 GWh of exports and 26 531 GWh of sources in total. All seven values equal the annual line exactly. For 2024 the same match comes out at the values 29 248, 5 040, 5 284, 18 200, 14 455, 17 411 and 26 292 GWh.

The annual line is therefore a sum of months and not a separate collection. From that it follows that any difference between output and the sum of the types must exist at the level of the month as well, and not only as an annual rounding effect.

A match on seven values in two consecutive years also rules out the second possibility — that the annual line carries an additional correction that does not feed into the months. If such a correction existed, the sum would differ from the annual value in at least one of the fourteen figures compared; it differs in none.

Step 2: subtracting the three named types

The calculation is one subtraction per month: output minus the sum of thermal, hydro and nuclear plant. For May 2026 that is 2 219 − (331 + 259 + 1 502), that is 2 219 − 2 092, giving 127 GWh. For January 2026 it is 2 825 − (680 + 288 + 1 836), that is 2 825 − 2 804, giving 21 GWh.

The same subtraction across all five months of 2026 gives the series 21, 34, 86, 118 and 127 GWh. The series rises month by month, with 32 GWh added between March and April and 9 GWh between April and May; over five months it comes to 386 GWh.

For 2025 the same subtraction gives twelve values: 19, 49, 73, 92, 94, 120, 107, 114, 79, 59, 31 and 19 GWh, together 856 GWh. For 2024 it is 28, 32, 55, 76, 92, 85, 97, 92, 66, 50, 32 and 19 GWh, together 724 GWh. For 2023 the annual values give 29 672 − (5 807 + 4 872 + 18 343), that is 650 GWh.

The minimum and the maximum of the series lie in the same year. In 2025 the lowest monthly value is 19 GWh, and twice — in January and in December — and the highest is 120 GWh in June; the ratio between them is 6,32-fold. In 2024 the lowest is 19 GWh in December and the highest 97 GWh in July, a ratio of 5,11-fold.

Over five months of 2026 the remainder accounts for 386 GWh out of total output of 12 554 GWh. The three named types give over the same period 2 374 GWh thermal, 1 524 GWh hydro and 8 270 GWh nuclear, together 12 168 GWh; the difference against output is exactly 386 GWh, so the check holds on the five-month sum as well, not only on the individual months.

Step 3: what does not belong in the remainder

Imports do not belong in the remainder. The line „Imports“ stands outside output and enters only the line „Sources in total“, which arises as output plus imports minus exports. The check for May 2026: 2 219 + 1 020 − 1 121 gives 2 118 GWh, the published value; for 2025, 28 883 + 15 968 − 18 320 gives 26 531 GWh, again exactly.

Nor does the difference between output and consumption belong in it. The table does not track consumption at all and network losses do not appear in it; anything about offtake would come from another source and does not appear in this calculation.

And finally rounding does not belong in it. The values are given in whole gigawatt-hours and the remainder reaches 127 GWh in May 2026, that is two orders of magnitude more than any rounding error on four added whole numbers.

The result: an unattributed remainder of 856 GWh for 2025

The base result of this calculation is 856 GWh for 2025. For 2024 it was 724 GWh and for 2023 it was 650 GWh; the year-on-year rise is therefore 18,23 % from 2024 to 2025 and 11,38 % from 2023 to 2024.

For the incomplete year 2026 five months are available and their sum is 386 GWh. The same period of 2025 gives 327 GWh, so the year-on-year rise is 18,04 % — almost identical to the pace of the whole preceding year.

Within a year the remainder has a clear seasonal course. In 2025 it rose from 19 GWh in January to a maximum of 120 GWh in June and fell back to 19 GWh in December; in 2024 the maximum was 97 GWh in July. The May value of 2026 at 127 GWh is meanwhile higher than any monthly value of 2024 and higher than the May value of 2025 by 33 GWh.

The year 2023 is a weaker point of this calculation than the two that follow. The value of 650 GWh is derived from the annual lines, not from the sum of twelve months, because the monthly check was carried out for this text only for 2025 and 2024. If the annual line of 2023 carried a correction, the result for that year would change; for 2025 and 2024 that possibility is ruled out.

The sensitivity table: the share of the remainder at different denominators

The absolute value of the remainder is unambiguous, its share is not. The overview below shows what happens if the same value is divided by output or by the line „Sources in total“, and how the result changes with the period chosen. All rows use exclusively the values given above.

The spread between the outer values is 3,53 percentage points at the denominator „output“ and 3,52 at the denominator „sources in total“. The share calculated from sources in total is systematically higher, because exports exceed imports across the whole period observed and the denominator is therefore smaller than output.

Sensitivity to the period is meanwhile greater than sensitivity to the denominator. One month — May 2026 — gives 5,72 %, while the whole of 2023 gives 2,19 %; the difference between them is 2,61-fold. Anyone stating a „share of the remainder“ without a period and without a denominator is not stating a verifiable figure.

The May value therefore cannot be scaled to a year. If 127 GWh were multiplied by twelve, the result would be 1 524 GWh, which is 78,04 % more than the actual annual result of 2025 — precisely because the series has a pronounced seasonal course with a winter minimum. An annual figure can only be obtained by summing twelve months.

What this calculation is not

It is not a comparison with the annual energy balance. That one works in terajoules, has 48 indicators instead of seven and a different definition of transformation; its values are not used in this text and none of the figures given is compared with it. Both series are issued by the same office, yet they answer different questions and follow a different publication cycle.

It is not a measurement of solar or wind output. The list of indicators in the description of the table dimension holds seven items and none of them concerns sun, wind or biomass; the remainder is therefore a remainder, not a technology.

Nor is it a statement about installed capacity. The series works with energy produced per month, not with capacity in megawatts, and the table gives no conversion between the two quantities.

And finally it is not a figure for an individual building. The series is national and monthly; none of the values given can be carried over to a specific hall or to a metering point.

The remainder is, in the end, a quantity that moves with what the table names. If a separate line for a further type of plant appeared in some later edition, the remainder would fall by the corresponding value without anything changing in actual output. It is therefore a property of the structure of the table as much as a property of the power system.

The practical consequence is that values from different editions of the table have to be compared with care. If the number of named types changes over time, the series of remainders before and after that change will not be comparable, even at unchanged output values. In the series from 2006 to May 2026 the number of indicators stays at seven.

Conclusions

The base result is 856 GWh for 2025. It arises by subtracting thermal, hydro and nuclear plant from electricity output and it exists in every single month of the whole series, not only in the annual total. For 2024 it was 724 GWh and for 2023 it was 650 GWh, the last of those values being derived from the annual lines alone.

The annual line is a sum of months, so the remainder is not rounding. For 2025 and 2024 all seven annual values equal the sum of twelve monthly ones exactly; the check ran inside the table and needed no external figure.

The remainder is growing and it is seasonal. Over January to May 2026 it reached 386 GWh against 327 GWh a year earlier, a rise of 18,04 %. The May value of 127 GWh exceeds any monthly value of 2024.

A share must be stated with a denominator and with a period. The same quantity gives 5,72 % for May 2026 and 2,19 % for 2023, that is 2,61 times less. The share of sources in total is always higher than the share of output, because exports exceed imports.

All of it can be checked from two machine outputs. The date of the last update of the table is given by the DATAcube collection index and it is 31 July 2026; the last filled month is May 2026, so no value in this calculation is any newer.

Analysis, not personalised advice. Before a transaction decision, verify the tariff value or the wording of the regulation directly with the network operator or the regulator.