The next structural shift in South-East Europe’s power market is no longer coming only from the supply side. For the past several years, the dominant investment narrative has been built around renewable capacity additions, cross-border congestion, battery storage and market coupling. That narrative remains intact, but it is no longer sufficient. A second force is now moving into the centre of price formation and grid planning: large-load demand. Data centres, AI compute infrastructure, telecom and cloud expansion, electrified industrial loads, and digitally anchored campuses are beginning to alter demand growth assumptions across the region. This is not yet a fully mature Balkan version of Northern Virginia or Dublin, but the direction is clear enough that investors, developers and system operators can no longer treat demand as a passive background variable. It is becoming an active pricing force. The International Energy Agency now projects that global data-centre electricity demand will more than double to around 945 TWh by 2030, with data-centre electricity consumption rising around 15% per year from 2024 to 2030. The IEA also expects data centres to account for 10% of EU electricity demand growth to 2030 under current policy settings.
That broader European backdrop matters because South-East Europe is entering this demand cycle from a very different starting point than the mature data-centre hubs of Western and Northern Europe. Power availability remains less saturated in several emerging markets, land is cheaper, and strategic geography is increasingly attractive for cloud, telecom and low-latency applications spanning Europe, the Eastern Mediterranean and the Middle East. CBRE’s 2025 global data-centre assessment notes that power availability is uneven and that some projects are increasingly being pushed into emerging markets where demand is not yet as saturated as in established hubs. At the same time, CBRE’s recent battery and grid analysis points to rising around-the-clock electricity demand from new data centres as a key reason power price volatility is likely to persist in Europe, with both BloombergNEF and S&P expecting European data-centre demand to double by 2030.
In South-East Europe, the most visible signal is not yet a single hyperscale cluster dominating the regional market, but rather a set of linked developments that collectively point to a new demand curve. Greece already has the clearest market narrative. Data Center Dynamics reported that the Greek data-centre market is expected to more than double by 2030, supported by submarine cable landings, geography between Europe, Asia and Africa, and a widening roster of large-scale digital infrastructure projects. One of the most strategically important recent moves was the joint venture between IPTOand Serverfarm to build hyperscale data-centre infrastructure in Greece, effectively tying power-grid expertise directly into digital-load development. This matters because it signals that large-load demand is no longer being treated as an external customer of the grid; it is being incorporated into grid strategy itself.
Romania is moving even more clearly into the large-load category. In December 2025, Data Center Dynamics reported that Accelerated Infrastructure Capital partnered with ClusterPower on an 800 MW data-centre build-out in southwestern Romania. An 800 MW AI-oriented campus is not a marginal load. At a high utilisation rate, it implies annual electricity consumption on the order of 5.5–7.0 TWh, depending on load factor and redundancy design. That is equivalent to a material slice of national electricity demand growth, and it fundamentally changes the economics of nearby generation, transmission reinforcement and storage deployment. Romania is therefore no longer just a renewable-export and balancing market. It is also becoming a potential anchor market for very large, round-the-clock digital demand.
Serbia is earlier in the curve, but the direction is visible. Domestic reporting tied to the state technology and infrastructure agenda indicates that the construction of a new data centre in Niš is planned to begin in 2026, while Serbia has already expanded sovereign compute infrastructure through the operation of a second supercomputer. Those signals do not yet amount to a hyperscale commercial hub on the scale of Greece’s international positioning or Romania’s ClusterPower plan, but they do indicate that Serbia’s digital-load story is moving from public-sector compute capacity into a broader infrastructure phase. In a market already dealing with connection queues, solar clustering and transmission bottlenecks, even a moderate ramp-up in 24/7 digital load changes the pricing and grid-value map.
The financial significance of this is easy to underestimate if one looks only at current aggregate demand. South-East European power markets have so far been dominated by supply-side volatility: hydrology, gas pricing, cross-border constraints and renewable intermittency. But data-centre demand behaves differently from most traditional incremental load. It is usually high load factor, requires very high reliability, tends to grow in concentrated geographic clusters rather than evenly across a country, and often arrives with strict commissioning timelines that force transmission and substation investment decisions. In grid terms, a data-centre campus is not just another customer. It is an anchor node that can justify or accelerate transformer upgrades, new 110 kV or 400 kV reinforcement, firming contracts and co-located storage.
This is where the interaction with power prices becomes strategic. South-East Europe’s price system has been shaped by renewable oversupply in some hours and gas-linked scarcity in others. Large digital load changes that pattern by adding a relatively inelastic demand component during off-peak and shoulder hours. In practical terms, that raises the local price floor. Midday solar oversupply does not disappear, but it is absorbed more readily in zones with data-centre concentration. Evening and night demand becomes structurally firmer, which raises the marginal value of flexible generation, imports and batteries. The result is not necessarily uniformly higher prices, but tighter spreads between zero-like surplus hours and moderate-demand hours, combined with stronger support for peak-price events if the load is not fully hedged with local generation or storage.
For renewable developers, that creates a new commercial layer. A solar plant located in a weak-demand zone may still face severe midday capture-price erosion, particularly in southern corridors exposed to oversupply. The same project located near a growing digital-load cluster may achieve a higher effective capture price simply because the local demand sink is deeper. This is especially important in markets such as Serbia and Romania, where many project models still assume broad national price references without sufficiently incorporating the local effect of emerging load centres. As the region adds more data-centre demand, that simplification becomes less defensible.
Romania offers the clearest example of how this could evolve. An 800 MW data-centre region, if built and energised in stages, would create a load profile large enough to alter regional dispatch assumptions and potentially absorb a significant share of renewable output that might otherwise pressure local prices. It also changes the storage case. A large-load cluster with high uptime requirements is unlikely to rely purely on wholesale market procurement. It typically requires a layered power strategy: firm grid connection, backup generation, storage for resilience and possibly long-term renewable procurement. That opens the door for hybrid structures combining nearby solar or wind, BESS and long-term contracted offtake, which is precisely the type of project architecture that supports stronger DSCR and higher debt capacity in South-East Europe.
A basic numerical illustration shows the scale. A 100 MW data-centre load running at a 90% utilisation rate consumes roughly 788 GWh per year. A 300 MW campus at the same load factor consumes around 2.36 TWh per year. An 800 MW regional cluster runs toward 6.3 TWh per year at high utilisation. That is large enough to underpin multiple gigawatts of renewable PPAs, several hundred megawatts of battery storage and major transmission reinforcement. It is also large enough to change basis risk and PPA pricing in surrounding zones because demand becomes more anchored and less seasonal than traditional industrial consumption.
The same logic applies to Greece, though through a different route. Greece’s attraction is not only domestic demand but its position in international connectivity and cloud architecture. The country’s growing data-centre pipeline sits on top of a power market already shaped by LNG-linked marginal pricing and large solar additions. That is an unusually interesting combination for investors. On one side, Greek power prices remain among the most volatile in the region, supporting battery arbitrage and flexible asset returns. On the other, the growth of digital demand strengthens the case for long-term renewable procurement and 24/7 matched energy strategies. This means data centres do not simply increase demand; they create premium demand for firmed and structured electricity supply.
That premium demand increasingly translates into bankable procurement. Romania already offers a useful signal through Orange Romania’s ten-year virtual PPA with Engie Romania, covering 40 GWh per year of electricity demand. While this is not a hyperscale arrangement, it shows that digital and telecom operators in the region are moving toward long-duration power contracting rather than relying purely on spot or short-term supply. As cloud and AI infrastructure scales, this behaviour is likely to intensify, because the economics of volatile grid procurement become less attractive for operators with large and continuous loads.
This is where the demand shock begins to feed directly into renewable project finance. In the older South-East European model, a merchant solar or wind project often had to prove bankability through a combination of relatively optimistic price assumptions, moderate leverage and perhaps a partial floor-price mechanism. In the new model, a nearby data-centre or digital-load offtaker can act as a credit anchor. That supports longer-term contracted cash flow, compresses merchant exposure and improves the bankability of both generation and storage. A well-structured renewable-plus-storage project supplying part of a data-centre load profile can support leverage in the 65–75% range and DSCR assumptions closer to 1.30x–1.40x, whereas a purely merchant equivalent in a congested node may be pushed toward 50–60% leverage and 1.45x–1.60x DSCR requirements.
The CAPEX implications are equally important. Large digital loads often force grid reinforcement, but they can also justify it. A new 110 kV or 400 kV connection package, substation reinforcement, reactive power equipment and backup integration can add tens of millions of euros to the effective energisation cost of a campus. At the same time, those upgrades can unlock broader local capacity for renewables, industrial electrification and storage. In that sense, data-centre demand behaves like a catalyst for grid investment, similar to how a major industrial plant historically justified transmission upgrades in the region.
This has direct relevance for Serbia. The country’s broader energy planning already assumes rising electricity demand and significant new grid and generation requirements through 2030 and beyond. The IEA’s latest electricity outlook also underscores that global electricity demand growth through 2030 is being accelerated by industrial electrification, cooling demand and the expansion of data centres and AI. In Serbia’s case, even without a hyperscale commercial boom on day one, a combination of public digital infrastructure, telecom expansion, electrified logistics and private compute capacity can still produce a meaningful incremental load shock over the remainder of the decade.
A realistic Serbia scenario is therefore not a single 800 MW hyperscale cluster tomorrow, but a cumulative digital-load increase that could still reach the low terawatt-hour range by 2030 if the Niš data-centre project progresses, sovereign compute infrastructure expands, telecom and cloud demand rises and international operators begin to test the market. In pricing terms, that matters because Serbia’s current market is still shaped heavily by thermal baseload, hydrology, cross-border spreads and growing solar exposure. New all-hours load would support higher baseload demand, reduce some midday softness in well-connected nodes and strengthen the commercial case for storage and industrial-style PPAs.
Romania is where this becomes investable at scale first. An 800 MW data-centre build-out is large enough to drive dedicated power strategies. Nearby renewable developers will no longer be competing only in a wholesale market shaped by HUPX and OPCOM spreads. They will be competing to become part of a structured supply stack for digital infrastructure. That can include virtual PPAs, physical sleeved contracts, co-located solar and BESS, and even long-term balancing arrangements. Investors who understand this early will see that the value is not only in selling electricity, but in selling reliability, firmness and carbon-adjusted supply to a load that increasingly cannot tolerate price spikes or delivery uncertainty.
Platforms such as Electricity.Trade become strategically relevant in that environment because the monetisation problem changes. The old question was where to build generation into a fragmented and volatile regional grid. The new question is where generation, storage and transmission access intersect with a new class of round-the-clock buyer. That requires granular visibility into node-level prices, congestion patterns, balancing spreads and route-to-market structures. It also requires understanding that the best renewable project in the next South-East European cycle may not be the one in the highest irradiation zone, but the one with the best line of sight to a firm digital-load cluster.
The broader market implication is that South-East Europe is starting to move from a surplus-and-export narrative toward a dual narrative of flexible supply and anchored demand. Renewables still set the direction. Gas still sets the margin in many hours. Storage still monetises volatility. But data centres and digital infrastructure introduce a new load layer that raises the long-term value of firm power, strengthens the economics of hybrid assets and makes certain transmission nodes much more valuable than they looked under a purely supply-led model.
That is why the next demand shock should now be treated as a first-order market variable. It will not eliminate congestion. It will not erase the need for BESS or cross-border arbitrage. It will, however, change the shape of the demand curve and the quality of offtake available to new projects. In a region where many sponsors are still underwriting assets primarily against wholesale market assumptions, that shift is large enough to create mispricing.
The emerging power map of South-East Europe is therefore no longer just north-versus-south, solar-versus-wind or merchant-versus-contracted. It is also compute-versus-constrained-grid, and the winners are likely to be those who understand that digital load is becoming one of the most important demand-side assets in the regional electricity system.
