Thermal power in South-East Europe in 2026 will not be determined by a single “coal versus renewables” narrative. It will be determined by how much hydropower the region actually receives, how high the CO₂ price floor sits across European-linked markets, how reliably coal mining can deliver lignite tonnage to power plants, and how much cross-border trading is required to patch system gaps in tight hours. In this region, thermal generation is still the system’s security backbone in dry periods, but its economic role is increasingly that of a residual, high-volatility marginal unit whose value rises when hydro and wind underdeliver and collapses when they overperform.
The most important anchor for SEE thermal in 2026 is Serbia because Serbia remains one of the last systems in the region where lignite-fired generation still provides a very large share of domestic electricity in most years. EPS is the dominant actor, and the hard physical data defines the envelope. Serbia’s lignite-fired installed capacity is often cited around 4,390 MW, with total EPS installed capacity around 7,662 MW, and hydro capacity around 2,936 MW, which means the country’s security-of-supply logic is still thermal-first when hydro is weak. EPS also commissioned the new Kostolac B3 unit at 350 MW in late 2024, a meaningful addition because it increases both firm capacity and baseload flexibility inside the coal stack. In 2024, EPS generation was reported around 32.9 TWh, down from 35.5 TWh in 2023, and this decline is consistent with a year where either hydro, outages, or demand patterns changed the mix and increased imports. These numbers are not just descriptive; they define what 2026 can plausibly look like in both system and fiscal terms.
The direct link between hydro and thermal is the key forecasting mechanism. Serbia’s hydro fleet is large enough that a hydro swing of 2–3 TWh in a year is not unusual, and that swing almost always shows up as an opposite swing in thermal dispatch and/or imports. When hydro underdelivers, lignite units run harder and longer, which increases lignite burn, raises variable OPEX, increases outage risk later in the year, and pulls forward maintenance. When hydro overdelivers, lignite units cycle more and some units may move closer to minimum stable output, reducing annual lignite burn but increasing cycling wear. This is why 2026 thermal forecasting must be expressed as a hydrology-conditioned band rather than a single number.
A reasonable Serbia thermal generation band for 2026 begins with the system’s typical annual requirement and the implied residual after hydro and renewables. If EPS remains in the ~33–36 TWh annual generation corridor and hydro is near-normal, lignite-fired generation in Serbia typically sits in a broad range of roughly 23–28 TWh depending on outages, import economics, and hydro timing. If hydro is strong, thermal can compress toward the low end of that band, or even below it, because hydro displaces baseload coal in shoulder seasons and reduces peak-hour scarcity. If hydro is weak, thermal can lift toward the high end or exceed it, because Serbia must protect domestic adequacy and avoid high-priced imports in tight hours.
This leads to three quantified 2026 thermal cases for Serbia that map directly to the hydro probability bands already implied by recent volatility. In a base hydrology case, where hydro normalises and EPS avoids major unit outages, Serbia’s thermal generation is likely to stabilise around 24–27 TWh for the year, with Kostolac B3 adding a firmer baseload contribution and improving system adequacy in winter evenings. In an upside hydro case, thermal can compress toward 21–24 TWh because more hydro displaces lignite in spring and early summer and reduces import-hour exposure. In a downside hydro case, particularly if the summer is dry and reservoirs cannot buffer, thermal can rise toward 27–30 TWh, and the marginal import requirement rises sharply if mining or unit availability becomes constrained.
The next constraint is coal mining supply, because Serbia’s lignite is mostly domestic and the thermal fleet is structurally dependent on continuous mine-to-plant delivery. EPS operates the Kolubara and Kostolac basins. A high-level national figure commonly referenced for EPS lignite output is around 37 million tonnes per year, but what matters for operational forecasting is the basin-level deliverability and the recent performance signal from mine operations. Reporting around the Serbian lignite system indicates that even in a difficult year the Kolubara basin delivered around 22 million tonnes in 2023, while the Drmno mine in Kostolac produced 9.2 million tonnes in 2023 and reached a new annual record above 9.9 million tonnes by late 2025 while being expanded toward a 12 million-tonne capability. This mine-level detail matters because it signals whether the system can feed a higher thermal run-rate in a dry year without creating a supply bottleneck.
To link coal tonnage to thermal output in a practical model, the key relationship is that lignite plants in Serbia burn large volumes of low-calorific coal, and a shift of 1 TWh of lignite generation can translate into a swing of multiple million tonnes of lignite depending on plant efficiency and coal quality. The newest unit, Kostolac B3, is reported with net efficiency around 37.3%, meaning it should deliver more electricity per tonne than the oldest lignite units, which slightly improves the system’s coal-to-power economics at the margin. In a 2026 base thermal scenario of 24–27 TWh, Serbia’s lignite burn requirement tends to sit in a broad corridor around 30–37 million tonnes, depending on which units are dispatched, outage patterns, and coal quality. In a downside hydro year where thermal rises toward 27–30 TWh, lignite demand pressure increases, and the system becomes much more sensitive to strip ratio, equipment availability, and overburden logistics in the mines.
This is where “market finance indicators” become real rather than theoretical. For Serbia, the indicator is not whether coal is expensive in commodity terms, but whether the mines can sustain output without forcing emergency imports of power at high prices. A thermal-heavy year raises OPEX through higher mine operating intensity, higher maintenance, and higher failure probability in older units. It also raises capital needs because higher dispatch accelerates wear, which pushes sustaining CAPEX higher. When a utility is already balancing investment in grids and renewables, a thermal-heavy year can crowd out investment, which then increases future risk. The financial story is therefore circular: more thermal today can reduce imports today but increase system fragility tomorrow unless CAPEX discipline is strong.
The regional price environment in 2026 amplifies this. European CO₂ pricing has re-established itself as a firm floor under thermal power costs. Analysts have cited an expected average EUA price for 2026 around €92.02 per tonne, with CO₂ trading near the €90–92 range in mid-winter. This matters even for non-EU SEE markets because the region trades with EU-linked zones and imports price the CO₂ component implicitly. When CO₂ sits near €90+, the variable cost of coal and gas units in interconnected markets rises, which increases the day-ahead price floor in tight hours. For Serbia, this raises the opportunity cost of importing power in dry periods and therefore increases the incentive to run lignite units hard when mining supply allows. In dry conditions, this dynamic can lock the system into high thermal dispatch even if it increases long-run maintenance burden, because the alternative is importing at prices shaped by CO₂ and scarcity.
Trading and balancing value then becomes the second-order driver. Thermal plants are no longer valued only for their annual MWh. They are valued for their ability to provide ramping, reserves, and adequacy when hydro is scarce and wind and solar output is volatile. In 2026, with Eastern Europe’s solar fleet having expanded rapidly and with more variable renewables in the region, balancing becomes more valuable, particularly in evening ramps and winter peaks. Thermal units provide that balancing, but at a rising cost. Coal units are often less flexible than gas units, but they still provide inertia and reserve capability in systems where grid stability constraints are tightening. In practice, the balancing value of thermal assets rises when hydro flexibility is weak, because hydro is the cheapest and fastest balancing resource in the region. This is why a dry hydro year creates a double effect: it reduces energy and reduces flexibility, forcing thermal to do both jobs, and pricing that volatility into ancillary and balancing costs.
When Serbia’s thermal output rises, coal mining and internal coal logistics become the binding micro indicators. The Kolubara and Kostolac supply chain must sustain higher daily output, which increases diesel consumption in mining, equipment wear, and contractor intensity. The signal to watch in early 2026 is therefore not only generation but mine output cadence and mine-to-plant stockpile levels, because those variables determine whether Serbia can sustain a 27–30 TWh thermal year without outages or forced derates. The newest unit, 350 MW Kostolac B3, improves adequacy, but it does not eliminate mine constraints; it increases their importance because higher dispatch capacity requires higher lignite throughput discipline.
In SEE beyond Serbia, thermal 2026 is increasingly shaped by policy and decommissioning schedules, but the direction of travel is not uniform. Romania, for example, has a coal fleet around 2.6 GW that has been subject to phase-out commitments, but the government has sought to extend operation deadlines due to replacement project delays. The economic implication for regional trading is that if Romania retains more coal capacity online through 2026–2027 than previously assumed, regional tight-hour prices can be lower and volatility can be reduced, because the region keeps more dispatchable capacity. If, instead, coal exits faster than replacement arrives, the region becomes more import dependent and more sensitive to LNG-linked gas prices and CO₂. This matters for Serbia because Serbia’s import price exposure is set by regional scarcity, not by domestic costs alone.
A realistic SEE thermal forecast for 2026 therefore looks like this: Serbia remains a thermal-dominant system with lignite generation most likely in the 24–27 TWh base band, with an upside toward 27–30 TWh if hydro underdelivers and imports are expensive, and a downside toward 21–24 TWh if hydro is strong and unit availability is stable. Coal mining supply must remain in a corridor that can support roughly 30–37 million tonnes of lignite burn in the base case, with higher pressure in the dry case, and the key risk is not coal price but operational deliverability in Kolubara and Kostolac. CO₂ pricing near €92 on average in 2026 supports a higher marginal cost for imports and therefore incentivises higher domestic thermal dispatch when physically feasible. Trading and balancing will monetise thermal capacity more in stress hours, but it will also expose the system to higher volatility costs if hydro flexibility is weak.
The macroeconomic transmission into Serbia in 2026 is then straightforward. In the base thermal band, Serbia can keep import exposure manageable and reduce the probability of tariff shock. In the dry hydro/high thermal band, Serbia reduces import bills at the expense of higher internal OPEX and higher future CAPEX needs, while also increasing local air-quality and environmental compliance stress. In the strong hydro/low thermal band, Serbia benefits from lower coal burn and lower maintenance stress, but it must manage coal mining economics and social dynamics in a system where lignite employment and regional economies remain highly dependent on dispatch levels.
The practical conclusion is that thermal output forecasts for 2026 cannot be separated from coal mining performance and hydro uncertainty. Serbia’s system remains physically capable of running lignite hard, but the cost of doing so is increasingly expressed in maintenance cycles, mining logistics, and volatility management rather than in the nominal price of coal. In 2026, the winners in SEE will be the systems that can balance hydro flexibility, thermal adequacy, and cross-border trading without falling into an import-driven price spiral or a maintenance-driven reliability spiral.
Elevated by virtu.energy