Serbian data centres could convert domestically consumed electricity into higher-value digital services for European Union customers, but financing the sector requires lenders to assess facilities as power-intensive infrastructure rather than conventional real estate.
- Electricity supply becomes a core credit factor
- Grid resilience and renewable contracts affect bankability
- Customer contracts determine revenue quality
- Carbon exposure extends beyond CBAM
- Hourly renewable matching can support customer demand
- Data regulation creates additional market requirements
- Financial-sector customers face DORA requirements
- Higher-density computing increases technology risk
- Stress tests should combine operating risks
- Operating platform is more important than building value
- Domestic power consumption can support digital-service exports
The central credit issue is whether projects can turn Serbian electricity consumption into durable euro-denominated revenue while maintaining reliable power, competitive carbon intensity, regulatory access to EU customers and sufficient technological flexibility over the loan tenor. For lenders, CBAM is not the primary financing risk. The key considerations are power reliability, customer contracts, carbon exposure, regulatory requirements and the ability of facilities to remain commercially relevant.
Electricity supply becomes a core credit factor
Electricity consumed by a Serbian data centre to provide storage, cloud or computing services to EU customers does not create the same direct CBAM liability as electricity physically exported from Serbia into the EU, because the exported product is a service rather than a CBAM-covered good. This creates a potential value chain in which Serbian power is consumed by a Serbian data centre and converted into digital services sold to EU customers, rather than being exported through an interconnector to an EU electricity importer.
The model could generate higher value per consumed megawatt-hour than wholesale electricity exports, provided the resulting digital-service revenues are sufficiently stable for debt financing. Banks should assess secured MW, annual MWh demand, firm grid capacity, redundancy, PUE, network charges, contracted power prices, backup generation and renewable-supply arrangements.
A facility with lower construction costs but weak electricity security could be less bankable than a more expensive project supported by a robust grid connection and long-term power strategy.
Grid resilience and renewable contracts affect bankability
Continuous electricity supply makes grid resilience a central component of data-centre financing. Lenders should examine dual-feed arrangements, substations, transformer redundancy, N+1 or stronger designs, battery storage, generators, fuel reserves and the capacity to maintain operations during equipment failures or scheduled maintenance. Even a relatively short outage can result in SLA penalties, customer claims and reputational damage that exceed the value of electricity that was not consumed. Reliability therefore needs to be assessed alongside electricity pricing when determining debt capacity.
Renewable power can strengthen the commercial proposition, particularly for European customers seeking lower supply-chain emissions. A renewable PPA should therefore be assessed as part of the financing structure rather than solely as an ESG feature.
Key considerations include PPA tenor, pricing formula, generator creditworthiness, production profile, balancing exposure, curtailment, change-in-law provisions, termination rights, replacement supply and ownership of guarantees of origin. Because data centres operate continuously, annual renewable coverage does not necessarily eliminate hourly supply gaps. Banks should model periods in which wind or solar generation is insufficient and calculate the cost of obtaining residual electricity from the grid.
Customer contracts determine revenue quality
The second major financing consideration is the strength of contracted revenue. Banks should evaluate contracted IT load, customer credit quality, contract duration, minimum commitments, price indexation, termination rights, SLA deductions, renewal assumptions and customer concentration. Long-term agreements with investment-grade European customers can support greater debt capacity than facilities dependent on future occupancy.
The largest customer also requires separate downside analysis, including scenarios in which it leaves, reduces capacity or does not renew. Digital workloads can be moved more easily than many industrial production activities, increasing the importance of contractual strength.
Carbon exposure extends beyond CBAM
A Serbian data centre providing services to EU customers does not need to purchase CBAM certificates for the electricity it consumes. Carbon exposure nevertheless remains commercially relevant. Electricity consumption contributes to the operator’s Scope 2 emissions and can also become relevant to the EU customer’s Scope 3 reporting.
European procurement teams may therefore increasingly assess service providers according to energy efficiency and carbon intensity. For lenders, the issue is whether carbon intensity could affect customer demand, pricing power or refinancing value. The supporting energy documentation should cover hourly and annual electricity consumption, renewable coverage, GoO cancellation, PPA allocation, backup-generator fuel, location-based Scope 2, market-based Scope 2 and emissions intensity per service unit.
Hourly renewable matching can support customer demand
Hourly renewable matching is not a CBAM requirement for Serbian digital-service exports, but it could provide a commercial differentiator for facilities serving European customers. A data centre able to demonstrate the relationship between hourly demand, renewable production, contractual allocation, certificate retirement and customer workloads could provide stronger low-carbon evidence than a facility relying solely on annual certificate matching.
For lenders, the potential value is linked to revenue protection. Better carbon data could support customer retention, access to premium clients and exit value, making energy traceability part of the broader credit assessment.
Data regulation creates additional market requirements
For many EU customers, data regulation may present a more direct commercial issue than carbon. Serbia is outside the EU GDPR adequacy framework, so transfers of personal data to Serbian facilities require an appropriate legal mechanism and safeguards.
Lenders should examine Standard Contractual Clause architecture, data-processing agreements, encryption, key ownership, access controls, customer audit rights, data-location commitments and subcontractors. Cybersecurity requirements add another layer. NIS2 covers data-centre and cloud providers and can apply to certain non-EU providers offering services within the Union. A Serbian operator targeting EU customers therefore needs governance, incident management, business continuity, disaster recovery and cybersecurity controls capable of supporting that customer base. Failure to meet these requirements could reduce the number of EU customers able to use the facility.
Financial-sector customers face DORA requirements
Projects targeting European banks and insurers face additional requirements under DORA. EU financial institutions must assess ICT providers in areas including operational resilience, subcontracting, concentration risk, data location, auditability and exit arrangements.
For a Serbian data-centre project targeting financial-sector customers, lenders therefore need to assess whether an EU bank’s procurement and risk committee would approve the provider. Uncertainty on that point should be reflected in revenue assumptions. A DORA-ready contractual framework, documented resilience controls and transparent subcontracting arrangements can affect the bankability of a project before its first customer is contracted.
Higher-density computing increases technology risk
Data-centre facilities can become commercially outdated before their buildings reach the end of their useful lives. The growth of AI and high-performance computing is increasing rack densities and changing cooling requirements. Lenders should consequently review maximum rack density, liquid-cooling readiness, spare transformer capacity, available grid headroom, fibre scalability and expansion space.
Financial models should incorporate recurring upgrade CAPEX instead of assuming that the initial technical configuration will remain competitive throughout the debt tenor. Facilities that cannot accommodate higher-density computing could require substantial additional investment or face weaker occupancy.
Stress tests should combine operating risks
Lenders should test multiple adverse conditions simultaneously rather than relying on individual downside scenarios. The minimum stress framework should include electricity prices 25%-50% above the base case, PPA generation below forecast, slower customer ramp-up, loss of the largest customer, weaker-than-designed PUE, CAPEX overruns of 10%-20%, grid-connection delays, a major SLA outage and higher cybersecurity and compliance expenditure.
Additional cooling or electrical CAPEX should also be tested. Debt sizing should incorporate a combined scenario involving higher power costs, weaker occupancy and lower efficiency. Maintaining acceptable DSCR under such conditions would strengthen the infrastructure characteristics of the project.
Operating platform is more important than building value
Collateral analysis should not focus primarily on the physical data-centre building. Recoverable value can also be linked to secured grid capacity, substations, fibre connectivity, customer contracts, renewable procurement arrangements and expansion rights.
A facility with 50 MW of secure power, strong connectivity and long-term EU customers can attract specialist operators, infrastructure funds and private-equity buyers. A technically comparable building without secured power or customers could have substantially lower recovery value. Exit and refinancing assessments should therefore consider the operating platform alongside the underlying real estate.
Domestic power consumption can support digital-service exports
Serbia has several potential ways to monetise electricity: direct electricity exports, domestic use for physical manufacturing or domestic consumption by data centres serving foreign customers. The data-centre model retains electricity consumption within Serbia while generating foreign-currency service revenue, alongside infrastructure, employment and other domestic value added. Instead of monetising 1 MWh at a wholesale power price, the model converts electricity into compute or storage capacity and recurring euro-denominated service revenue.
Elevated by CBAM.Clarion.Engineer


