Serbia’s CBAM exposure is often discussed as if it were a reporting problem that sits inside customs paperwork and corporate sustainability departments. In reality, from 2026 onward, it behaves more like a competitiveness tax on industrial systems that cannot credibly separate themselves from a coal-heavy electricity baseline. That is the Serbian problem in its simplest form: even if Serbia adds renewables, an exporter’s electricity input remains “coal-linked by default” unless the exporter can demonstrate traceable low-carbon sourcing at the meter. The default is not the country’s production narrative; it is the residual mix that applies when consumption is not sufficiently backed by verifiable attributes.
Serbia’s national residual mix for 2024, published by EMS, shows exactly why the default is so difficult to escape at scale. The corrected residual mix is dominated by brown coal and lignite at 66.60%, with hydropower at 23.81%, natural gas at 5.02%, wind at 0.97%, and solar at 0.36%. That residual mix is the “grid electricity fingerprint” that many CBAM-exposed exporters inherit unless they can prove otherwise through Guarantees of Origin (GOs) or an equivalent instrument architecture that EU counterparties accept as credible. The same EMS report also illustrates why “just buy green” becomes scarce quickly: GOs issued for production in 2024 total 2,405,275, while GO cancellations total 2,447,795—a real market, but one that can be overwhelmed by a small number of heavy industrial buyers seeking to ring-fence their electricity footprint.
This is why the discussion about Serbia’s renewable auctions and grid upgrades matters, but also why it can be misleading if read as an automatic solution for exporters. The second Serbian renewable auction, supported by EBRD, offered a quota of 424.8 MW split into 300 MW wind and 124.8 MW solar PV, supported through 15-year contracts for difference. In energy terms, that is a meaningful supply block. In exporter terms, however, what matters is how much of that block can be made exporter-specific through PPAs and GO allocation—because the residual mix remains the “penalty factor” for everything not ring-fenced.
To treat this properly, the right approach is to map Serbia’s CBAM-exposed exporter landscape at the company level, estimate conservative “green electricity demand” in annual MWh that would need traceable coverage to materially displace the residual mix, then stress-test that demand against the auctioned pipeline and the realities of attribute allocation and grid deliverability.
The Serbian exporter set worth modelling is not infinite; it is concentrated in a handful of large platforms. In steel, the anchor is HBIS Serbia at Smederevo, the single largest CBAM-exposed industrial export complex in the country. HBIS’s corporate materials state an annual production capacity of 2.2 million tons of finished products and employment of more than 5,000 people. Steel sits directly inside CBAM’s initial scope, but electricity becomes a key competitiveness variable for a second reason beyond reporting: electricity is the lever that can move fastest while deeper process changes take longer to finance and execute. If HBIS Serbia can credibly ring-fence a large share of purchased electricity as low-carbon—through a portfolio of renewable PPAs and cancellable GOs—the steelworks can begin shifting its product footprint narrative in EU procurement discussions even before the heaviest decarbonisation capex lands.
In cement, Serbia’s exposure concentrates around three plants operated by international groups. Moravacem (Popovac near Paraćin) states annual capacity of 1,350,000 tons. TITAN Cementara Kosjerić states production capacity of 750,000 tons per year. The national competition authority’s sector analysis states that total annual cement factory capacity in Serbia is 3.4 million tons, which implicitly frames the remaining capacity around the third plant. Cement is in CBAM’s initial scope, and while cement’s core emissions driver is clinker chemistry and kiln fuel, electricity procurement still matters: it is measurable, auditable, and contractable, and it is increasingly used as a “proof of seriousness” signal by lenders and EU-facing customers.
In fertilisers and chemicals, Serbia has export-oriented platforms where electricity intensity and traceability are becoming product strategy rather than optional CSR. The clearest case is Elixir Prahovo and the broader Elixir industrial platform. Elixir’s Prahovo 2027 materials state the new technology would reduce thermal energy consumption by 50% and electricity by 25% per ton of phosphoric acid produced. That statement is important not because it tells you the plant’s MWh today, but because it proves that electricity input is material and is being actively targeted—exactly the behaviour you see when an exporter anticipates tighter EU carbon and traceability requirements.
In aluminium, Serbia’s profile is primarily processing rather than primary smelting. The key distinction is that an aluminium processor’s electricity footprint is significant but not at the multi-TWh scale of an electrolytic smelter. That makes green electricity procurement a “high leverage” move: a mid-scale PPA can cover a large share of consumption and reposition the footprint of exported rolled or processed products in EU value chains faster than many other interventions.
With the company set defined, the next step is a conservative translation from production scale into “green power demand,” defined here as the annual MWh that would need to be covered by traceable low-carbon attributes (GOs cancelled against consumption, or equivalent) to materially displace Serbia’s coal-heavy residual mix for those exporters.
For HBIS Serbia, the starting point is its stated production capacity of 2.2 million tons/year. Without claiming proprietary meter data, the responsible way is to use a conservative electricity intensity envelope for integrated steelworks’ purchased electricity and finishing loads—recognising that the biggest CBAM driver remains process emissions from the traditional route, but that purchased electricity is still large and becomes larger under any electrification pathway. An envelope of 0.30–0.45 MWh per ton yields annual electricity demand in the range of 660–990 GWh for a plant at 2.2 million tons/year. The strategic interpretation is straightforward: if HBIS Serbia can ring-fence and prove something like 700–900 GWh/year of renewable electricity, it can credibly say that its purchased electricity input is largely decarbonised, even if its process emissions still require longer-cycle capex solutions.
For cement, producer-stated capacities allow a cleaner calculation. Moravacem at 1.35 million tons, TITAN Kosjerić at 0.75 million tons, and a national total capacity of 3.4 million tons imply that the remaining plant accounts for roughly 1.3 million tons of capacity scale. Cement electricity intensity for grinding and plant utilities (excluding kiln thermal energy) is commonly modelled conservatively around 90–120 kWh per ton. Applying that to 3.4 million tons yields annual electricity demand of roughly 306–408 GWh for the national cement platform envelope. Cement’s “green electricity demand,” therefore, is a few hundred GWh/year problem—large enough to matter, but solvable through a handful of PPAs and explicit GO allocation if the market architecture permits it.
For Elixir Prahovo, the company itself discloses current output scale for phosphoric acid and NPK fertilisers: 165,000 tons of phosphoric acid annually, and an NPK fertiliser plant with 300,000 tons/year capacity. Combined with Elixir’s stated plan to reduce electricity consumption by 25% per ton under Prahovo 2027, a conservative procurement envelope for ring-fencing electricity—without overstating—can be framed at 100–250 GWh/year for the cluster. The goal here is not precision; it is scale. The scale is “hundreds of GWh,” not “multiple TWh.”
For aluminium processing, a conservative annual envelope of 50–150 GWh/year is reasonable for a rolled or processed product plant scale—again, not a primary smelter, but a significant industrial load where electricity-backed product claims can be commercially powerful in EU supply chains.
When you aggregate these conservative envelopes—HBIS Serbia 660–990 GWh, cement 306–408 GWh, fertiliser/chemicals 100–250 GWh, aluminium processing 50–150 GWh—you reach a likely exporter “green electricity demand” range of approximately 1.12–1.80 TWh/year. Within that range, the single dominant driver is HBIS Serbia, which is why Serbia’s green electricity debate cannot be solved by small symbolic projects alone. It needs portfolio-scale procurement and explicit attribute allocation.
The stress test now turns to supply. The second auction’s quota is 424.8 MW, split 300 MW wind and 124.8 MW solar, supported via 15-year CfDs. Converting quota MW into annual energy requires conservative capacity factor assumptions that reflect Serbia’s operating reality rather than optimistic developer presentations. A conservative wind capacity factor range of 30–35% produces annual output of 788–920 GWh/year for 300 MW wind. A conservative solar capacity factor of 15–18% produces 164–197 GWh/year for 124.8 MW solar. Combined, the quota corresponds to roughly 952–1,117 GWh/year of incremental renewable energy in an average year.
On paper, ~1.0 TWh/year is close to the low end of the exporter demand envelope. The problem is that exporters do not consume MWh; they consume MWh with a proof layer. The binding constraint is how much of that auctioned output becomes exporter-allocable with cancellable GOs. Serbia’s EMS residual mix and GO cancellation scale in 2024 imply that most consumption remains outside the GO-backed segment and falls into the coal-heavy residual pool. If auctioned output is absorbed into general supplier portfolios without systematic GO allocation to industrial customers, exporters will still be forced to account most of their electricity as residual-mix electricity.
A conservative, policy-realistic assumption for stress testing is that only 40–60% of the auctioned quota’s annual output could be made exporter-allocable as traceable green electricity in the near term, because other customers—households, public supply obligations, other industrial buyers—will also compete for green attributes, and because administrative allocation is not automatic. Under that allocation range, the exporter-allocable green supply from the quota becomes 381–670 GWh/year (that is 952–1,117 GWh/year multiplied by 40–60%).
Now the quantification becomes clear. Exporter demand sits at ~1.12–1.80 TWh/year. Exporter-allocable supply from the quota, under conservative allocation assumptions, is ~0.38–0.67 TWh/year. The residual gap is therefore approximately 0.45–1.42 TWh/year, which is consistent with the 0.4–1.4 TWh/year range previously used as the planning gap.
This gap is not an abstract accounting mismatch; it is the measurable quantity of annual renewable attributes that Serbia’s major CBAM-exposed exporters would need—but cannot reliably obtain—if they rely only on the auctioned quota and the current attribute market structure. In the absence of dedicated projects and explicit GO allocation, exporters remain tethered to the residual mix of 66.60% lignite/brown coal for uncovered consumption.
The geography of this mismatch matters because it determines what is deliverable and what is expensive. Serbia’s largest exporter load sits in the Belgrade–Danube basin, while many of the strongest wind corridors sit in Vojvodina and South Banat. That is why transmission reinforcement has become strategically aligned with renewable growth. The BeoGrid 2025 project is publicly described as worth €205 million, and official communications describe new high-voltage lines connecting Belgrade and Novi Sad and an additional line toward the Čibuk connection substation, explicitly to stabilise transfer of renewable energy from South Banat and relieve network congestion. This grid architecture is not just a power system project; it is the physical enabler of exporter-anchored renewable procurement. A steelworks like HBIS Serbia can only scale green procurement credibly if the grid can deliver incremental renewable output to the load basin reliably and if the attribute registry can allocate the proof layer.
At this point, Serbia’s CBAM electricity strategy splits into two realistic pathways. The first is “supplier allocation,” where exporters obtain renewable attributes from EPS or suppliers through GO-backed supply contracts, relying on the system’s pooled renewable output. This pathway can work for mid-scale loads like cement and aluminium processing because their green electricity needs are in the hundreds of GWh range and can be met by a supplier portfolio if the GO cancellation mechanism is explicit and credible. The second pathway is “exporter-anchored build,” where large exporters either directly sponsor new renewable projects or contract them through long-term PPAs with GOs assigned and cancelled against their consumption. This pathway is the only one that can solve the scale problem for HBIS Serbia and, by extension, reduce the national green electricity gap that CBAM-exposed exporters face.
The strategic conclusion is uncomfortable but operationally useful. Serbia does not only need more renewables. Serbia needs a deliberate carve-out of renewables—projects and attributes specifically earmarked for CBAM-exposed exporters—because otherwise the residual mix remains the default and the auctioned pipeline’s MWh do not translate into exporter competitiveness. This is why the next question is not “how many MW are auctioned,” but “how many additional projects Serbia must build specifically for exporter PPAs and GO allocation to close the 0.4–1.4 TWh/year gap.”