Serbia has opened a RSD 625 million, approximately €5.3 million, public procurement procedure for the planning and initial technical documentation required to develop the Đerdap 3 pumped-storage hydropower plant, moving the decades-old project into its most substantive preparatory phase.
The Ministry of Mining and Energy is seeking a consultant or consortium to prepare the General Design, Preliminary Feasibility Study, Special-Purpose Spatial Plan and Strategic Environmental Assessment. Bids are due by 20 August 2026.
The procurement remains an early-stage development contract rather than a construction tender. Its purpose is to determine whether Đerdap 3 can be configured, permitted and financed as a technically coherent project while managing its impact on the Danube, the existing Đerdap hydropower system, navigation, protected areas and Serbia’s cross-border water relationship with Romania.
The proposed plant would operate as a large water-based electricity-storage system. During periods of low demand or excess renewable generation, electricity would be used to pump water from the Đerdap 1 reservoir into one or possibly two upper reservoirs. When electricity demand and prices rise, the stored water would be released through reversible pump-turbine units to generate power.
Current concepts indicate an installed capacity of between 1,200 MW and 2,400 MW, with the Ministry now examining an intermediate configuration of approximately 1,800 MW. A hydraulic head of around 400 metres is considered technically possible, with Pesača and Brodica on the Northern Kučaj mountain range being assessed as potential upper-reservoir locations.
The Danube and the existing Đerdap 1 reservoir would serve as the lower reservoir. The project area lies between Golubac and Donji Milanovac, approximately 65 kilometres upstream of Đerdap 1.
The complete scheme would require intake and outlet structures on the Danube, large underground or surface waterways, pressure tunnels and penstocks, upper dams and reservoirs, reversible generating units, a powerhouse, transformers, switchgear and a high-capacity connection to the Serbian transmission network operated by Elektromreža Srbije.
This is considerably more than another hydropower station. Đerdap 3 would consume more electricity during pumping than it later returns to the system because of hydraulic, electrical and mechanical losses. Its value would come from shifting electricity between periods, absorbing surplus wind and solar generation, supplying peak demand and providing rapid-response balancing, reserves, frequency control and system-restoration capability.
The project has an indicative investment value of approximately €2.6 billion, although the final cost cannot be considered reliable until the plant configuration, reservoir volume, geological conditions, tunnelling requirements and grid-connection design are established. At the 1,800 MW configuration currently under discussion, the headline estimate implies capital intensity of around €1.44 million per MW, or €1,440 per kW.
That is plausible for a large pumped-storage scheme using an existing lower reservoir, but the range remains exposed to substantial upward pressure. Complex tunnelling, difficult geology, environmental mitigation, compensation measures, grid reinforcement and imported electromechanical equipment could move the construction envelope towards €3 billion–€3.4 billion. A full 2,400 MW build-out could require €3.2 billion–€4 billion, depending on the storage duration and the number of upper reservoirs.
The €5.3 million documentation tender therefore represents only around 0.2 per cent of the currently indicated construction value. Its importance is not the amount being spent but the decisions that the resulting studies must support. A weak preliminary design could lock the state into an oversized configuration, underestimate geological exposure or fail to establish a defensible environmental baseline. A lender-grade development process needs to settle the plant’s optimum capacity, energy-storage duration and construction sequence before Serbia negotiates an engineering, procurement and construction contract.
A 1,800 MW plant with eight hours of full-load generation would provide approximately 14.4 GWh of usable storage. A ten-hour configuration would raise that to 18 GWh. The maximum 2,400 MW option would provide between 19.2 GWh and 24 GWh over the same duration range.
These are illustrative figures because the reservoir volumes and final operating envelope have not yet been selected. They nevertheless show the scale of the project. Even the intermediate option would be significantly larger than the battery-storage systems presently being developed across Southeast Europe and could balance several gigawatts of variable renewable generation for extended periods.
Đerdap 3 would sit alongside Serbia’s separate Bistrica pumped-storage project, rather than replacing it. Bistrica is smaller and more advanced in its development pathway, while Đerdap 3 is conceived as a strategic regional-scale asset. The two plants could eventually provide complementary services: Bistrica offering system flexibility at a more manageable project scale and Đerdap 3 providing deep storage, peak capacity and cross-border balancing across longer operating cycles.
The economic case will depend on the revenue structure. Pure energy-price arbitrage is unlikely to provide sufficient certainty for several billion euros of non-recourse debt. The plant would need a combination of market revenues, long-term capacity remuneration, ancillary-service payments and potentially a state-backed availability or system-services contract.
Under an illustrative 1,800 MW base case, capital expenditure of €2.6 billion–€3.2 billion, round-trip efficiency of 75–80 per cent and annual discharged electricity of approximately 2.5–3.5 TWh could produce an annual energy-arbitrage margin of roughly €150 million–€230 million, depending on the spread between pumping and generation prices.
Ancillary services, reserves, capacity availability and congestion-management value could add another €60 million–€120 million a year. After annual operating costs equivalent to approximately 1.5–2 per cent of capital expenditure, a mature plant could generate EBITDA in a broad range of €170 million–€270 million.
A merchant-only project would remain exposed to volatile price spreads and could produce an unlevered return below the level required for such a long and construction-intensive asset. A bankable structure combining market revenues with contracted capacity and system-service income could support a project return of approximately 6–8 per cent and a leveraged equity internal rate of return in the region of 8–11 per cent.
The 2,400 MW upside case would increase the plant’s ability to monetise scarcity prices and regional balancing demand, but it would not automatically produce better returns. Additional capacity has value only when sufficient low-cost pumping energy and high-priced discharge periods are available. An oversized plant could operate at a lower utilisation rate and require expensive transmission reinforcements that dilute its financial performance.
An upside configuration with capital expenditure of €3.2 billion–€4 billion, storage of at least 19 GWh, annual discharged output of 3.5–5 TWh and a strong regional capacity-payment arrangement could support EBITDA of approximately €280 million–€400 million. The corresponding leveraged equity return could reach 9–12 per cent, but only with disciplined construction costs, adequate storage duration and contracted revenues covering a substantial portion of debt service.
Financing will require a structure closer to major regulated infrastructure than to a conventional merchant renewable project. A plausible envelope would combine 20–30 per cent equity or sovereign-equivalent public funding with 70–80 per cent long-term debt, potentially involving export-credit agencies, US development-finance institutions, international financial institutions and commercial banks.
At a €3 billion base construction cost, this would imply approximately €600 million–€900 million of equity or public capital and €2.1 billion–€2.4 billion of debt. A tenor of at least 20–25 years after completion would be preferable, given the construction duration and the asset’s expected operating life of 60 years or more.
Debt service cannot safely depend entirely on day-ahead electricity-price spreads. Lenders would require a clearly defined revenue floor, a capacity or availability mechanism, regulated cost recovery, or a long-term state-backed agreement for strategic balancing services. The allocation of hydrological, market and dispatch risk will be central to credit approval.
The project’s exposure to schedule risk is equally material. A delay of 12–18 months after financial close could add approximately €120 million–€300 million through construction inflation, extended owner’s costs, contractor claims and interest during construction. Such a delay could reduce the equity internal rate of return by around 0.7–1.5 percentage points, depending on the leverage ratio and whether lost availability revenues are compensated.
The risk becomes more pronounced under a 2,400 MW configuration because larger underground works, additional units and a more demanding grid connection create more interfaces. A phased approach—potentially commissioning an initial 1,200 MW or 1,800 MW block before completing the full plant—could reduce funding pressure and allow the operating strategy to be tested against actual regional price and balancing conditions.
The current political timetable envisages the completion of a first phase by around 2036, with a wider development horizon potentially extending to 2038. A realistic programme would need several years for planning, geological investigation, environmental studies, cross-border coordination, land acquisition and permitting before main construction begins. The initial contracting and front-end engineering phase alone has been indicated at approximately 36 months.
The state has also linked the project to its strategic energy cooperation with the United States. That bilateral framework entered into force in March 2025, following the intergovernmental agreement signed in 2024. A public call inviting US companies to express interest in participating in Đerdap 3 closed on 25 June 2026, attracting submissions from six companies.
The process is intended to identify a strategic supplier or partner capable of supporting front-end engineering and potentially delivering the later EPC contract. The requirements presented to interested companies included experience in hydropower or comparable infrastructure, management of front-end engineering and delivery of projects with investment values exceeding €1 billion.
The American dimension could bring access to project-management expertise, export-credit financing, large rotating-equipment supply chains and construction technology. It could also support Serbia’s wider effort to diversify strategic infrastructure partnerships beyond the Chinese, Russian and European companies that already hold strong positions in its energy and transport sectors.
The selection process must still establish competitive cost discipline. Strategic agreements can accelerate project preparation, but they can also reduce competitive tension when a development partner subsequently moves into an EPC role. Serbia will need transparent open-book pricing, independently verified quantities, benchmarked equipment costs and clear rules governing the conversion of front-end engineering into a construction contract.
The €5.3 million public procurement and the US strategic-partner process are related but legally and commercially distinct. The first is intended to produce Serbia’s planning and preliminary technical basis. The second concerns the identification of a potential international partner. Keeping those functions separated is important so that the General Design and Preliminary Feasibility Study remain owner-led documents rather than solutions shaped around the commercial interests of a future contractor.
Romania is the other indispensable party. Although the power station and proposed upper reservoirs would be located on Serbian territory, Đerdap 3 would draw water from a reservoir that forms part of the jointly managed Iron Gates hydropower and navigation system. Pumping and generation cycles could influence Danube water levels, navigation conditions and the operation of Đerdap 1 and Đerdap 2, which Serbia and Romania jointly operate.
The two countries signed a memorandum in Bucharest on 16 July 2026 to facilitate information exchange and technical assessment. Romania has made clear that any development must remain compatible with electricity production at Iron Gates I and II, protection of riverside areas and navigation on the Danube.
This creates an unusual interface risk. Đerdap 3 may be a Serbian asset, but its operating rules cannot be designed solely around Serbia’s electricity-market interests. Reservoir-level limits, pumping schedules and high-output generation periods will need to be coordinated with Romanian authorities and the bilateral bodies managing the existing Đerdap system.
Those operating restrictions could affect revenues. A plant prevented from pumping during low-price hours or generating during the most valuable scarcity periods would capture a smaller spread. The Preliminary Feasibility Study therefore needs to model not only engineering performance but also the commercial consequences of bilateral water-management constraints.
Grid integration presents another major challenge. Injecting up to 2,400 MW at one point would represent a very large change for the Serbian transmission system. The connection studies will need to examine the capacity of nearby 400 kV infrastructure, internal Serbian bottlenecks and cross-border corridors towards Romania, Bulgaria and Hungary.
The plant could reduce renewable curtailment by absorbing surplus electricity, particularly during high-wind and high-solar periods. Its economic value would rise as Serbia adds more variable renewable capacity, but that benefit depends on sufficient transmission capacity to move surplus electricity to Đerdap and return stored energy to demand centres during peak periods.
A base renewable-integration case could assume that Đerdap 3 helps Serbia accommodate an additional 3–5 GW of wind and solar capacity during the 2030s. A 2,400 MW plant with longer storage could support a wider 5–7 GW regional renewable envelope, including electricity imported for pumping from Romania, Bulgaria and Hungary. These figures represent system-planning potential rather than dedicated generation directly assigned to the project.
Curtailment savings will form only part of the value. At avoided curtailment of 500–1,000 GWh a year and an average captured electricity value of €40–€70/MWh, the direct annual benefit could reach approximately €20 million–€70 million. The larger system value comes from avoiding emergency imports, reducing fossil-fuel peaking, supplying reserves and improving the ability to manage sharp changes in renewable output.
Environmental permitting may ultimately determine the feasible project size. The proposed development sits within the wider Đerdap Gorge landscape and affects an area of high ecological, cultural and archaeological value, including parts of Đerdap National Park. Upper-reservoir construction would require dams, excavation, access roads, transmission infrastructure and disposal areas for substantial quantities of rock and soil.
The Strategic Environmental Assessment must compare alternatives rather than merely justify a preferred configuration. The location and number of upper reservoirs, surface versus underground infrastructure, construction traffic, biodiversity effects, water quality, sediment movement, landscape impact and cumulative effects with existing hydropower assets all require early assessment.
A technically attractive 2,400 MW scheme could prove less bankable than a smaller alternative once environmental mitigation, permitting time and cross-border constraints are incorporated. The optimum solution may therefore be the configuration offering the strongest risk-adjusted system value rather than the maximum installed capacity.
The new tender moves Đerdap 3 beyond political announcements, but it does not yet make the project construction-ready. The next stage must convert a 50-year-old concept into a modern storage asset with a confirmed reservoir design, defensible environmental pathway, coordinated Serbian-Romanian operating regime, lender-grade revenue model and competitive contracting strategy.
The 1,800 MW configuration presently under consideration appears large enough to transform Serbia’s balancing capability while remaining more manageable than an immediate 2,400 MW build-out. The General Design and Preliminary Feasibility Study will now determine whether that middle option can hold its estimated €2.6 billion–€3.2 billion capital envelope and deliver bankable storage economics without transferring excessive geological, environmental and market risk to the Serbian state.
