The government is quietly moving forward with its nuclear ambitions. The political discussion is yet to be had. A lack of expertise, high costs, corruption, as well as environmental conditions make the planned shift to nuclear power a risky bet.
People of a certain age in Serbia remember Chernobyl as a taste. In the summer of 1986, the strawberries came in as usual, red and ripe, and were eaten under a cloud that had drifted west from Soviet Ukraine. Vesna Goldsworthy made the fruit the title of her 2005 memoir, Chernobyl Strawberries, a coming-of-age story set in Belgrade, the capital of then-socialist Yugoslavia. Forty years on, Serbia is preparing to build a nuclear plant of its own.
For decades the question of nuclear power in Serbia was settled. A 1989 moratorium, passed in the shadow of Chernobyl and the final years of Yugoslavia, banned the construction of nuclear plants. Over time it also drained the country of nuclear expertise and hardened a conviction that nuclear power was both dangerous and unnecessary. In November 2024 the government reopened the file. An amendment to the Law on Energy repealed the ban and set out a phased programme for developing nuclear power. Serbia now says it wants a reactor. Whether it can build one, and whether it should, are separate questions. Public debate about this future trajectory was almost entirely absent. The issue has been treated as a technical matter, not unlike many other key strategic energy and climate-related decisions.
Why Serbia thinks it needs nuclear
Nuclear has been pushed by industry circles for years. The argument starts with a supply squeeze. Demand for electricity is rising while the plants meant to meet it are old and mostly coal-fired. About two-thirds of the country's electricity generation comes from lignite plants, with hydropower supplying most of the rest and (non-hydro) renewables only a small share.
Demand for electricity is rising.
Official projections have electricity rising from 26.8 percent to 33.7 percent of final energy consumption by 2040, driven by the projected electrification of heating and transport, by digitisation and by hopes of reindustrialisation. Peak demand could rise sharply by mid-century.
Environmental effects of this brown coal dependency combined with energy poverty are well documented, not least by the Heinrich Böll Foundation’s work in the field, among them some of the worst air pollution on the continent. Despite commitments made as a member of the Energy Community, Serbia has not committed to decarbonising its energy sector. Until now, it seems.
The strain is already showing. In 2022, with two units of the Nikola Tesla coal complex out of service and a poor year for hydropower, Serbia had to import more than 3TWh of electricity, around 8 percent of what it used, at a heavy cost to the state. Officials warn that without new capacity the country could face regular winter shortfalls in the near future. Import dependence brings its own risk of volatile prices on regional exchanges.
There is a second, newer pressure, and it points the other way. Since January 2026 the EU's Carbon Border Adjustment Mechanism (CBAM) has put a carbon price on imports into the bloc, including electricity. It raises the cost of the power Serbia sells. In the meantime, Serbia has hastily introduced its own carbon tax to avoid the EU’s “indirect taxation”. The effects of these measures remain to be seen. Serbian electricity is so carbon-intensive that the charge is steep. That gives Belgrade a fresh commercial reason, on top of the environmental one, to want low-carbon power that can run around the clock.
This is the promise of nuclear: steady, low-carbon electricity that does not depend on the weather. This is referred to as "baseload", the minimum, round-the-clock demand that never goes away. For a coal-dependent economy under decarbonisation pressure, the pitch is attractive. It is also the easiest part of the story.
Serbia Is Still in the Very Early Stages of Implementation
Serbia is following the roadmap set by the International Atomic Energy Agency (IAEA), which its national programme translates into three legally defined phases. Phase one tests whether developing nuclear power is justified at all. It is largely a research exercise: a battery of technical, organisational, legal, economic, financial and staffing studies, meant to establish whether the country has the grounds and the capacity to proceed. Phase two develops the programme proper, and ends when the National Assembly, on the government's proposal, adopts a Strategy for the Peaceful Use of Nuclear Energy. Only phase three involves an actual reactor: the design, construction and operation of a plant, and it would be governed by a separate law passed specifically for the purpose. The Ministry of Mining and Energy runs the first two phases; the third, if it is ever reached, would be a different undertaking altogether.
Serbia is at the very start of phase one. The institutions to carry it out are only now in place. In February 2026 the government set up an inter-ministerial expert working group, and by the summer the country's Nuclear Energy Programme Implementing Organisation (NEPIO) was running. NEPIO is not a single body created by a single act; it combines that working group with an internal unit inside the Ministry of Mining and Energy. Its members are drawn from across the state: the Vinča Institute of Nuclear Sciences, the Nikola Tesla Electrical Engineering Institute, the Serbian Nuclear Society, the regulator SRBATOM, the power and transmission utilities, and several ministries from environment to defence. Regulatory oversight sits with SRBATOM, set up under the country's radiation- and nuclear-safety law.
A working reactor is treated as an option only after 2040.
The government wants to finish phase one by mid-2027 and phases one and two together by 2032, at which point it hopes to choose a technology. A working reactor — the substance of phase three — is treated as an option only after 2040. France's EDF is the front-runner to be selected as the strategic partner. It signed a memorandum of understanding in April 2024 and prepared the early feasibility study; it has now been asked to write several of the phase-one studies, with the rest put out to tender. But Serbia is keeping its options open. It has signed or discussed similar memoranda with China's atomic-energy institute, Russia's Rosatom, South Korea's KHNP as well as Slovenia. None of these is binding. But the range of suitors, from Paris to Moscow and Beijing, points to the political choice behind the technical one.
Which Type of Nuclear Power Will Serbia Choose?
Suppose Serbia gets as far as choosing. It will face a decision the whole industry is now arguing about: the large conventional reactor, or the small modular reactor (SMR).
The conventional reactor is a known quantity. A standard light-water design produces 1,000-1,600MW from a single, custom-built, on-site complex. It has decades of operating data behind it, and a poor record on cost and schedule. Europe's recent flagship projects, Britain's Hinkley Point C, France's Flamanville and America's Vogtle, all ran far over budget and behind time. A modern 1,000MW reactor is now reckoned to cost between €5bn and €20bn, several times the price of a coal plant, and often takes more than a decade to build.
The SMR is the fashionable alternative and the source of much of the current excitement, not least in the European Commission. The demand, mainly in the USA, is driven to a large extent by the need to power new data centres. It reverses the logic of the big build. Instead of one large machine, it uses a series of smaller units, usually under 300MW and often much less, manufactured in a factory as standard modules, shipped to site and assembled, with capacity added as needed. The claimed advantages are lower upfront cost per unit, faster and more predictable construction, flexible siting (on a retired coal site, for instance, so no need for a very specific location) and passive safety systems that rely on gravity and convection rather than pumps.
It is an uncertain bet.
The economics are less flattering than the pitch. On the measure that matters most – capital cost per kilowatt of capacity – SMRs are currently more expensive than large reactors, not cheaper. The IEA treats cost parity as a target to be reached only around 2040, when it projects SMR construction costs falling to roughly $4,500/kW in Europe and America. The SMR case rests on a bet that factory serial production will bring costs down over time, which needs someone to take on the expensive first-mover risk of building enough units to prove it. It is an uncertain bet. So far only Russia and China have grid-connected SMRs operating, and no Western commercial unit is yet running. But the picture is moving quickly. EDF's own study looks at both options, the conventional and the fashionable.
Human Resources, Costs, and Environmental Protection Are Challenges
Whichever reactor Serbia might choose, the same obstacles stand in the way, and several matter more than the choice itself.
Serbia's shortage of nuclear expertise is severe.
The most pressing is human resources. Serbia's shortage of nuclear expertise is severe. The 1989 moratorium did not ban teaching or research, but the effect was similar: specialists emigrated, secondary school and university courses in nuclear engineering closed, and the older generation retired with practically no one trained to replace them. Nuclear physics at the University of Belgrade is taught by physicists trained in other fields, and the electrical-engineering faculty has produced, by one unofficial account, no nuclear engineers in 20 years. This is the one point on which every side agrees, and rebuilding the expertise is thought to take 10 to 20 years, about the same time it would take to build the plant.
The economics are no kinder. Beyond the high capital cost, nuclear's levelised cost of electricity is several times that of new solar or wind, around $140–220 per MWh against well under $80 for utility-scale solar and onshore wind, and the gap has widened as renewables have got cheaper, though raw comparisons understate the cost of the storage and back-up that variable power requires. Decommissioning a plant costs €0.8bn-1.3bn, and a deep geological waste store of the kind Finland has built can cost €3bn and must be monitored for centuries.
Nuclear's levelised cost of electricity is several times that of new solar or wind.
Private insurers will not cover the full cost of a serious accident, so the state becomes the insurer of last resort; the Fukushima clean-up alone has passed $180bn, most of it public money. Commercial banks rarely fund such projects without state guarantees. Successful reactors tend to be built by heavily state-backed companies such as France's EDF or Korea's KEPCO. Renewables, by comparison, create about 2.7 times as many jobs per dollar invested and can be dismantled within a year.
Then there is the physical aspect, most of it still unstudied in Serbia. Firstly, no seismic study has been developed specifically for the purpose of analysing nuclear-plant construction in Serbia, even though Serbia sits on a moderately active seismic belt and records small tremors most weeks. Secondly, a traditional reactor needs large amounts of cooling water. The Danube, with an average flow of 5,500-6,000 cubic metres a second, could supply it easily, which is why the plausible sites lie along the river. But cooling brings thermal pollution: water returned 5-10°C warmer, lowering oxygen levels and harming river biodiversity. In addition, climate change pressure on freshwater resources, as demonstrated this year, means that low water levels in major rivers could force reactors to close temporarily during prolonged summer droughts, as witnessed in neighbouring countries this summer.
The country has no store even for low-level material.
On waste, the country has no store even for low-level material, let alone the high-level spent fuel a reactor would produce, which must be kept isolated for hundreds of thousands of years. And there is the accident risk. Nuclear power is genuinely low-carbon, at 10-15 grams of CO2-equivalent per kilowatt-hour over its life, close to wind. But only if you leave out the accidents.
Germany's experience cuts both ways. Since closing its last reactors in 2023, Germany has kept the lights on with wind and solar, which together now supply close to half its power, backed by coal, gas and imports, much of the imported power being French nuclear. Anke Weidlich, an energy-systems researcher, argues in her recent interview with Die Zeit that this shows a modern grid does not need new nuclear: the services reactors once provided, such as frequency stability and inertia, can increasingly come from batteries, grid-forming inverters and cross-border links. But the same story is a warning. A country that gives up nuclear without enough clean, firm power to replace it ends up leaning on coal and on its neighbours, which is the position Serbia, largely dependent on brown coal, is trying to get out of.
Weak Rule of Law and Corruption Make Implementation Unlikely
There was hardly any public debate, and this is not a mere procedural objection. It is an essential concern which relates not only to public participation and transparency but also to institutional capacities of the state, not just in terms of expertise but in terms of its capacity to govern in a democratic way. With the weak rule of law and failure to tackle corruption as well as a terrible track record of environmental protection, it is almost unimaginable that the project of this magnitude and complexity could ever be implemented or even assured by existing institutions. The Ministry of Mining and Energy's handling of the EDF study is a case in point: it mistakenly published the preliminary version with the tracked changes still unresolved.
A reactor tolerates neither error in construction nor incompetence in management.
It is hard, in Serbia, not to think of the Novi Sad railway station canopy that collapsed in November 2024, killing sixteen people. A reconstructed piece of infrastructure built with foreign partners, whose failure, and the evasive institutional response that followed, laid bare the state's inability to manage a project far smaller and simpler than a nuclear plant. That is the uncomfortable backdrop to Serbia's atomic ambitions. A reactor tolerates neither error in construction nor incompetence in management. Before any of it, the country must rebuild a lost profession, find tens of billions of euros, write a legal and regulatory system from scratch, satisfy the IAEA now and Euratom later, and choose whose technology, and whose sphere of influence, it will depend on for the next fifty years. Lifting the ban was the simple part.
