Serbia’s electricity system is entering a structurally different operating regime in the 2026–2032 window, driven less by domestic policy shifts than by external market coupling, renewable penetration in neighboring EU markets, and grid-level constraints. The critical question is no longer whether Serbia needs dispatchable capacity, but what form of dispatchability minimizes fiscal risk, curtailment exposure, and balance-of-system cost under declining baseload economics.
At present, Serbia’s dispatchable backbone consists of aging lignite units operated by EPS, complemented by hydropower with growing interannual volatility. Gas capacity remains marginal in volume terms. However, cross-border price convergence has already changed dispatch behavior. Imported electricity from Hungary, Romania, and Croatia increasingly sets marginal prices during solar-heavy hours, while scarcity pricing appears only during evening ramps, winter peaks, and regional stress events. This price shape is decisive for asset economics.
A simplified system stress test for Serbia over 2026–2032 shows that full-load-hour assumptions collapse for all thermal assets. Lignite units that historically ran 6,000–6,500 hours are structurally pushed toward 3,500–4,200 hours, with further erosion as regional solar capacity expands. New combined-cycle gas turbines (CCGTs), if built on a merchant basis, would struggle to exceed 2,000–2,500 hours, insufficient to recover CAPEX without long-term contracts or capacity payments. Battery storage, by contrast, monetizes volatility rather than volume and performs best precisely where thermal assets lose ground.
Quantitatively, the CAPEX and system cost comparison is stark. A modern 400–500 MW CCGT in Serbia would imply €350–450 million in upfront investment, excluding gas infrastructure reinforcement and fuel hedging. Even assuming optimistic spreads, annual EBITDA under merchant conditions would be highly volatile, with downside risk dominant once solar penetration in Hungary and Romania exceeds 30–35% of generation, a threshold already in sight. The risk profile worsens further if indirect carbon pricing continues to leak into Serbian prices through imports.
Coal reserve economics look superficially cheaper but hide structural inefficiencies. Keeping lignite units in cold or strategic reserve avoids immediate CAPEX but shifts costs into maintenance, staffing, and forced outages. A realistic reserve cost for Serbia’s older lignite units converges toward €70–90/kW/year, implying €140–180 million annually for a 2 GW reserve fleet. Crucially, these costs do not buy flexibility; they buy availability with long start-up times and poor ramping capability, precisely the opposite of what the system increasingly needs.
Battery energy storage systems (BESS) change the comparison. A 1 GW / 4 GWh utility-scale battery build-out — roughly equivalent in peak response capability to a mid-sized thermal fleet — would require €700–850 million in CAPEX at current prices, declining toward €550–650 million by the late 2020s. While headline CAPEX appears higher, the system value is materially different. Batteries capture intraday spreads, reduce imports during scarcity hours, lower reserve margins, and defer grid reinforcement. Modeled over a 15-year life, system-level net cost per avoided MWh of scarcity is lower than both gas peakers and coal reserve.
Grid constraints are the multiplier. Serbia’s internal north–south transmission bottlenecks and limited dynamic line rating restrict the ability to absorb cheap imports or export surplus hydro. In such a system, batteries co-located at constraint points outperform central thermal plants by reducing redispatch costs. Gas peakers located away from congestion points lose optionality, while coal reserve adds no congestion relief at all.
From a fiscal standpoint, the ranking is clear. Coal reserve externalizes cost into perpetuity, gas peakers create stranded-asset risk within a decade, while storage front-loads CAPEX but stabilizes OPEX and reduces exposure to imported price shocks. Under conservative assumptions, Serbia’s system cost minimization case favors accelerated storage deployment combined with targeted lignite reserve drawdown, not new baseload gas.
The implication is not ideological. It is arithmetic. Serbia’s grid and price shape reward speed and flexibility, not fuel throughput. Any future capacity remuneration must therefore be technology-neutral but flexibility-weighted, or it will lock public capital into declining assets.
