Serbia’s planned Đerdap 3 pumped-storage hydropower plant, with estimated investment requirements of approximately €2.6 billion, is being positioned as a major electricity-system project with potential implications for industrial and digital investment.
- Pumped storage designed to support renewable expansion
- Data centres require broader energy infrastructure
- International demand highlights importance of power availability
- Project financing requires stable revenue framework
- Cross-border and environmental approvals remain key issues
- Preparatory work begins with technical documentation procurement
- EPC structure and engineering preparation will determine execution risks
- Strategic value depends on multiple electricity-market benefits
The proposed facility has gained attention as Serbia seeks to attract energy-intensive technologies, including artificial intelligence infrastructure and large data centres, by improving grid flexibility and electricity management capacity. Đerdap 3 would not increase electricity production in the traditional sense. As a pumped-storage facility, it would use electricity during periods of lower demand and prices to pump water into upper reservoirs at Pesača and Brodica, then generate electricity when demand and market prices increase.
Around 20–25% of electricity used during each storage cycle would be lost, meaning the project’s economic value would come primarily from flexibility, balancing services and market optimisation rather than additional net generation.
Pumped storage designed to support renewable expansion
The role of Đerdap 3 is closely linked to Serbia’s long-term renewable energy plans, which envisage approximately 3.6 GW of wind capacity and more than 7 GW of solar capacity by 2040. A renewable portfolio of that scale would create larger differences between periods of high and low electricity production. Solar generation would increase supply during midday hours, while evening demand peaks and low-renewable periods could continue requiring additional flexibility and imports.
The pumped-storage plant could absorb electricity during periods of strong renewable output and return power during peak demand periods, helping reduce renewable curtailment and limiting exposure to expensive electricity imports. The project’s commercial model would need to rely on several revenue streams rather than electricity price differences alone. Potential sources include wholesale market trading, balancing energy, frequency regulation, reserve capacity, congestion management, black-start services and possible availability payments. The final economic value would depend on the plant’s design, storage volume, turbine configuration, operating cycles, grid connection arrangements and market rules in place when the facility begins operation.
Data centres require broader energy infrastructure
The growing demand for artificial intelligence infrastructure has become one argument supporting the development of large-scale energy storage in Serbia. A major data-centre ecosystem, however, requires more than access to a large power facility. Developers typically need reliable electricity supply, high-capacity transmission links, reserve capacity, predictable pricing, redundant grid infrastructure, cooling solutions, digital connectivity and access to low-carbon energy sources.
Đerdap 3 could contribute to some of these requirements, but it would need to operate as part of a wider investment framework involving Elektroprivreda Srbije (EPS), Elektromreža Srbije (EMS), renewable-energy developers, technology companies and regional electricity markets.
The existing State Data Centre in Kragujevac has reported capacity of approximately 14 MW. Operating continuously at full load would require around 123 GWh annually, equivalent to approximately 0.35% of Serbia’s electricity consumption.
This level of demand would be significant for a local electricity network but would represent less than 0.6% of Đerdap 3’s proposed maximum capacity, meaning Serbia’s current data-centre sector alone would not justify the investment. A 300 MW continuously operating data-centre campus would consume approximately 2.6 TWh annually, more than 20 times the estimated consumption of the Kragujevac facility and around 7.5% of Serbia’s current annual electricity demand. A gigawatt-scale artificial intelligence campus would require substantially greater generation capacity, multiple high-voltage connections and additional reserve infrastructure.
International demand highlights importance of power availability
Global developments in artificial intelligence infrastructure show that electricity availability has become a major factor in selecting locations for large technology investments. Large AI campuses under development in the United States are moving from tens of megawatts toward hundreds of megawatts and, in some cases, several gigawatts.
European markets are also experiencing pressure from data-centre expansion. In Ireland, data centres account for approximately 23% of national electricity consumption, while grid congestion around Dublin has restricted new connections and increased the importance of dedicated generation, storage and renewable supply arrangements. For Serbia, the experience highlights the need to match digital investment plans with transmission capacity and electricity-system development. A pumped-storage plant cannot replace the need for sufficient primary generation. Serbia would still require expanded renewable capacity, stronger transmission networks, reliable existing generation and potentially regional electricity agreements.
Project financing requires stable revenue framework
With estimated capital expenditure of approximately €2.63 billion, Đerdap 3 would require a financing structure capable of supporting a long-term strategic asset. Pumped-storage facilities have long operating lives, but their revenues can depend heavily on market conditions. Financing institutions are likely to place greater value on predictable contractual revenues than on projections based solely on future electricity price volatility.
A potential structure could combine regulated system-service revenues, long-term capacity arrangements, availability agreements with EPS or EMS, market-based income and contracts with major industrial or digital consumers. Data centres could become part of the future customer base, but they would not replace the need for a broader financial model. A possible integrated approach would combine new wind and solar projects with storage capacity, battery systems and grid-backed reserve supply, allowing technology investors to secure more predictable low-carbon electricity.
Such a model could also allow Serbia to capture greater value from renewable production by storing excess solar or wind electricity and using it during higher-value periods.
Cross-border and environmental approvals remain key issues
Đerdap 3 is planned on a cross-border section of the Danube, requiring technical coordination with Romania. A recently established bilateral working structure could support discussions, but issues related to water management, cross-border impacts, operating procedures and downstream responsibilities would require formal resolution. The proposed site also overlaps with the Đerdap National Park, creating additional environmental and spatial-planning requirements.
Although pumped-storage facilities circulate water between reservoirs rather than consuming it permanently, construction could affect biodiversity, landscapes, sediment conditions, hydrology and local infrastructure. The project would involve major works including upper reservoirs, tunnels, access roads, underground structures and construction areas.
Preparatory work begins with technical documentation procurement
Serbia’s Ministry of Mining and Energy has launched procurement for planning activities and parts of the technical documentation, with an estimated value of RSD625 million, approximately €5.3 million. An earlier call for expressions of interest attracted six companies. The project is intended to become the first major undertaking covered by the Serbia–United States intergovernmental energy cooperation framework.
US engineering company Bechtel has been linked to the project after financing earlier feasibility work. The company also has regional experience through the Morava Corridor, built with Turkey’s Enka, and has worked on hydropower studies for Albania’s Skavica project. Bechtel’s cooperation with Nvidia on modular designs for gigawatt-scale artificial-intelligence data centres has added a potential connection between energy infrastructure and future digital investment requirements. Such cooperation would not determine the final project economics and would not replace competitive procurement, independent review or safeguards for Serbia’s commercial interests.
EPC structure and engineering preparation will determine execution risks
The project could be delivered through an engineering, procurement and construction structure covering design, equipment procurement and construction responsibilities. Civil works would likely include reservoirs, tunnels, waterways and an underground or partially underground powerhouse. Specialised equipment suppliers such as Voith, Andritz, GE Vernova or Toshiba could compete to provide reversible pump-turbines, generators and control systems. A single-point EPC approach could simplify project management, but underground construction creates significant risks related to geology, design changes and claims.
Before awarding a major construction package, Serbia would need a detailed reference design, updated geological studies, hydraulic modelling, environmental and social impact assessments, transmission analysis, procurement strategy, financial modelling and clear allocation of cross-border responsibilities. Independent Owner’s Engineer oversight would be important for reviewing designs, monitoring progress, assessing claims and verifying commissioning performance.
The role would also include interface management between civil works and equipment suppliers, cost forecasting, environmental compliance, grid-code testing, reservoir commissioning and verification of efficiency and operational response. For a project of €2.63 billion, weaknesses in early preparation or contract allocation could create significant additional public costs.
Strategic value depends on multiple electricity-market benefits
The economic contribution of Đerdap 3 would extend beyond direct market revenues. The facility could reduce balancing costs, lower renewable curtailment, improve the operation of existing hydropower assets and reduce Serbia’s exposure to winter electricity imports. These benefits would be distributed across different market participants, including EPS, EMS, renewable developers and the wider economy. They would need to be converted into contractual revenues to support project financing.
A potential structure could combine strategic public ownership with commercial agreements involving large industrial consumers and data centres, renewable-energy producers, system operators and international financing institutions.
The project’s investment case depends on maintaining viability even without securing major hyperscale technology investments. Đerdap 3 would need to demonstrate value as national and regional energy infrastructure capable of operating across multiple electricity-market conditions over several decades. Data centres could strengthen demand prospects and create additional contracts, but the project’s core economics would depend on engineering quality, market design, financing structure and integration with Serbia’s wider energy system.


