Batteries Solve Hours. Tarnița Could Solve Romania’s Overnight Flexibility Problem.

Romania’s storage market is scaling fast, and short. Tarnița Lăpuștești pumped storage is the first credible answer to the part of the day batteries were never built to cover.

The gap nobody is pricing

Romania built a battery market in roughly eighteen months. Installed capacity moved from about 138 MW and 269 MWh to close to 500 MW and 914 MWh, crossing the 1 GWh threshold on the way, with around 600 MW dispatchable by the end of 2025 and a project pipeline now above 9 GW. That is a genuine success story and it deserves the attention it gets.

It is also, on average, a 1.8 hour fleet.

Look at what the Romanian day actually does. On a recent trading day OPCOM averaged around 109 to 115 euros per MWh, but the average is the least interesting number on the page. Prices fell to 1 euro per MWh in the afternoon and reached 199.10 euros per MWh in the evening. In March, commercial solar output hit a record 2,048 MW around midday, covering roughly a third of instantaneous demand, and the day-ahead market printed its first negative hourly intervals of 2026 shortly afterwards.

A two hour battery monetises the top of that evening peak beautifully. What it does not do is carry the system from sunset through to the morning ramp. That is the window Romania currently fills with imports and with thermal plant that the country intends to retire. It is a duration problem wearing the costume of a capacity problem, and almost nothing being financed in Romania today is sized for it.

What actually changed in January 2026

On 27 January 2026, Hidroelectrica’s extraordinary general meeting approved the creation of a 50-50 joint venture with EDF Power Solutions International to develop the 1,000 MW Tarnița Lăpuștești pumped storage plant on the Someșul Cald river in Cluj County. Shareholders also approved the shareholders’ agreement governing the vehicle.

Read the structure rather than the headline. The collaboration runs in four phases: pre-feasibility, feasibility including the grid connection study and permitting, implementation and construction, then operation. Phase 1 carries a budget of 400,000 euros, split equally between the partners, covering site data collection, an initial land assessment and a pre-feasibility study. There is no final investment decision, and the shareholder resolution was explicit that none is being assumed.

The most commercially important line in the whole file is the condition EDF attached. Hidroelectrica’s supporting documentation states that EDF requested the adoption of a regulatory framework for a revenue support scheme covering the project, naming capacity reservation, a contract for difference or a feed-in tariff, in order to justify a decision to continue. In plain terms: Europe’s largest utility will not underwrite a billion euro merchant pumped storage asset in Romania on today’s market design. It wants the revenue floor written into law first.

The same meeting approved legal due diligence on acquiring, at commissioning, a second pumped storage project: the 300 MW Frasin-Pângărați scheme on the Bistrița river being developed by Hydro Blue Energy, with an optimistic target date of 2030.

The number that reframes the comparison

The design that Tarnița has carried since the 2008 feasibility study, updated in 2014, is four reversible units of 250 MW each. Annual generation was put at 1,625 GWh, against 2,132 GWh consumed in pumping, giving a round trip conversion coefficient of 0.76.

Those two figures do more analytical work than any press release. Derived from them, the plant moves roughly 4.45 GWh on an average day. On a 300 cycle year, a single cycle is somewhere near 5.4 GWh. That is about five hours of discharge at the full 1 GW rating.

Now hold the energy constant and drop the power. Run one 250 MW unit instead of four and the same stored volume delivers for close to twenty hours. That flexibility between power and duration is the structural difference between a reservoir and a battery rack, and it is the reason the word overnight belongs in this conversation at all. Tarnița is not a twelve hour asset in the way a hydrogen cavern is. It is a machine that can choose to be a five hour, 1 GW asset on a tight evening or a twenty hour, 250 MW asset across a windless night.

For scale: one Tarnița cycle moves roughly twice the total energy capacity that Romania’s entire 150 million euro Modernisation Fund battery call is designed to procure, which is a minimum of 2,174 MWh across the whole national programme. Every day.

Two honest caveats. These duration figures are derived from published design parameters last revisited in 2014, not from a current specification, and pre-feasibility exists precisely to revise them. And a plant that can run twenty hours is not the same as a plant that is dispatched for twenty hours, which depends entirely on the price shape it faces.

What 1 GW of long duration does to the merchant stack

This is the part that matters to anyone underwriting Romanian storage revenue past 2030.

• Spread compression. Batteries in Romania monetise the day to night spread, and that spread is widening as midday solar deepens. Tarnița monetises the same spread, at a volume capable of flattening the curve it feeds on. Any model that projects 2030s battery arbitrage revenue from 2026 spreads, without a Tarnița sensitivity, is projecting the wrong curve.
• Ancillary saturation arrives first, and arrives without Tarnița. With more than 9 GW in the pipeline and over 2 GW expected operational by the end of 2026, Romania’s frequency response and balancing pools will crowd long before a pumped storage plant is commissioned. Merchant battery cases built primarily on ancillary revenue face that compression regardless of what happens in Cluj.
• The services batteries sell badly. Synchronous inertia, voltage regulation and black start capability were named in the original Romanian state assessment of the project, alongside system restoration after a blackout. These are structurally under-monetised in the Romanian market today. That is a market design gap, not a technology gap, and closing it moves value toward synchronous machines.
• Duration arbitrage moves upmarket. Published technology comparisons put lithium ion ahead below roughly four hours and at moderate cycling, and pumped hydro ahead above eight hours with the advantage widening as duration grows. If Tarnița is built, the durable battery niche in Romania narrows toward fast, short, high cycle products, co-located solar shifting and distribution level flexibility, which is a perfectly good business, just a different one from the one many decks currently describe.

 

The case against, stated properly

Objectivity here is not decoration. This project has an unusually bad delivery record and pretending otherwise would be a disservice to anyone allocating capital.

• Fifty years of non-delivery. The scheme dates to the 1970s, was studied in 1993 and 1994, approved again in 2019, judged economically unjustified by the Ministry of Energy in 2020, and the dedicated project company Hidro Tarnița entered insolvency in 2023. Two feasibility study tenders failed for want of compliant bids.
• No final investment decision, and the condition for one does not yet exist. Romania has not legislated a capacity reservation, storage contract for difference or availability payment for long duration assets.
• Phase 1 is 400,000 euros against a project last costed above 1 billion euros, with an estimate of 1.15 billion euros attached to the 2019 approval and 1.3 billion euros cited elsewhere. Those numbers predate the construction cost inflation of the last five years.
• Five to seven years of construction after a positive FID. Even on an aggressive path, first commercial energy is a mid-2030s event. Nothing in this project helps the 2027 to 2030 flexibility gap.
• Round trip efficiency of 0.76 against high-80s for new lithium ion. Tarnița gives back about 24 percent of throughput as pumping loss where a modern battery gives back roughly 12. That penalty is severe when charging energy is expensive and close to irrelevant when charging energy is free or negatively priced, which is the direction Romanian midday is travelling.
• Hydrology and water regime. Romania has already seen hydro shortfalls tighten regional supply this year, and the water concession framework for pumped storage perimeters has only recently been clarified.

 

Why Romania is a credible place to build long duration

Set the project’s history aside and assess the location on its merits. The evidence is stronger than the track record suggests.

• Half the civil works already exist. Tarnița lake is built and operating. The pre-feasibility scope explicitly covers the lower reservoir plus possible upper reservoirs, which is a materially different risk profile from a two reservoir greenfield.
• The operator is already at scale. Hidroelectrica runs 188 hydropower plants and more than 6.3 GW, supplies about 27 percent of national production, and is the second largest Romanian company by market capitalisation. Romania already operates five pumping units totalling 91.5 MW, so reversible operation is not an unfamiliar discipline, just an unfamiliar scale. The company has separately committed more than 500 million euros to storage projects.
• The price signal is sharpening on schedule. Grid connected solar reached about 3.34 GW by April 2026, prosumers passed 359,000 with more than 4 GW installed by the end of June, and the country is expected to move past 7 GW of total solar. Midday floors of 1 euro per MWh and evening peaks near 199 euros per MWh are not an anomaly to be managed, they are the revenue stack for anything that can move energy across the sunset.
• The demand signal is officially quantified and unmet. Transelectrica estimates that integrating planned renewable capacity requires between 10 and 20 GWh of storage by 2030. The national battery call procures a minimum of 2,174 MWh. Even a fully delivered 2 GW battery fleet does not close a 10 to 20 GWh gap by itself. The arithmetic points at long duration whether or not policy does.
• The policy window is open right now. EU member states were required to submit flexibility needs assessments by 1 July 2026, with storage deployment targets due in early 2027 and state aid schemes expected to be built on those assessments from 2027 onward. That is precisely the legitimate route through which EDF’s requested revenue framework could be created. Romania has also already moved storage into primary legislation, into ANRE licensing, and into the auction based grid capacity allocation regime that started on 1 January 2026.
• European money is visibly flowing to this asset class. In January 2026 the Connecting Europe Facility awarded 243 million euros to two pumped storage projects, close to 40 percent of that call’s total energy allocation, with 180 million euros going to the 1 GW Aguayo II project in Spain. A further CEF energy call was planned for the second quarter of 2026. Europe’s hydropower sector and Eurelectric have been lobbying hard for a dedicated long duration framework, and the Paris Pledge identifies a pipeline of around 35 GW across the continent.
• The node is a regional asset, not a national one. Cluj sits in the northwest, near the Transylvanian renewables build-out, inside a bidding zone synchronously coupled to Hungary, Bulgaria, Serbia and Ukraine. Romanian stored energy is sellable across a region that is short at exactly the hours Tarnița would discharge.
• It is not a monument, it is a category. Frasin-Pângărați at 300 MW, privately developed with a state offtake route, gives Romania a second pumped storage project and the beginnings of a domestic supply chain argument.

 

Romania will not be Europe’s cheapest place to build pumped storage. It may be one of the few places where the reservoir, the operator, the price shape, the interconnection and the policy window line up inside the same decade.

What to watch between here and FID

1. Whether the pre-feasibility study lands on schedule or slips. A 400,000 euro study running late is the cheapest possible early warning signal.
2. Whether Romania’s flexibility needs assessment names long duration explicitly, and whether the early 2027 storage target is expressed in hours rather than only in MW and MWh.
3. Whether a capacity reservation, storage contract for difference or availability payment enters Romanian law. Without it there is no FID, and everything else is process.
4. Whether Tarnița re-enters Transelectrica’s network development plan scenarios. It was excluded from the 2022 to 2031 plan for lack of project information, which is a quiet but precise measure of institutional confidence.
5. Whether the joint venture bids into the next Connecting Europe Facility energy call, and whether the project pursues Project of Common Interest status.
6. What pre-feasibility concludes about duration. If the design moves above six hours, the system value case strengthens materially and the battery cannibalisation question gets sharper.

 

Momentum Energy’s View

The framing of pumped hydro versus batteries is the wrong argument, and it is the argument the Romanian market keeps having. The two technologies are not competing for the same hours. They are competing for the same balance sheets, and that is a very different problem.

Our position is straightforward. Batteries remain the correct instrument for the next five years in Romania, because they are the only asset class that can be permitted, financed and energised inside the current price window, and because the Modernisation Fund call opening on 1 September 2026 makes the next twelve months unusually favourable. Nothing about Tarnița changes that.

But anyone underwriting merchant battery revenue past 2032 on an extrapolation of 2026 spreads is modelling a market that a single approved project could reshape. That is not a reason to avoid batteries. It is a reason to build a Tarnița scenario into the downside case, to weight contracted and ancillary revenue more heavily in the back end of the model, and to stop assuming that duration scarcity is permanent.

On location, the evidence supports Romania more than the project’s history does. The lower reservoir exists, the operator runs 6.3 GW, the interconnection makes the output regionally sellable, and the country has a quantified 10 to 20 GWh gap that batteries alone will not close. What Romania lacks is not geology, engineering or demand. It lacks a long duration revenue framework, and the EU flexibility assessment cycle has just handed it a legitimate window to write one.

That is the whole decision. If Romania legislates a revenue floor for long duration storage during the 2027 state aid window, Tarnița becomes one of the more investable pumped storage projects in Central and Eastern Europe and the merchant assumptions underneath every Romanian battery model need revisiting. If Romania does not, this remains what it has been for fifty years: an excellent site with no bankable buyer for what it produces.

We would rather Romanian and regional investors were early to that question than surprised by it.