Serbia is spending close to €1bn on new trains, trams and metro rolling stock supplied by companies from Spain, China, France and Turkey, while two factories in Kragujevac belonging to Siemens Mobility continue producing rail vehicles primarily for foreign markets. The contrast has opened a larger argument about procurement transparency, sovereign guarantees and the absence of a domestic-content strategy capable of turning public transport investment into industrial development.
The central question is not whether Serbia should purchase foreign technology. Modern rolling stock is inherently international: propulsion systems, braking equipment, signalling interfaces, control electronics and safety components are drawn from multinational supply chains. Nor does the presence of a Siemens factory automatically mean every Serbian order could be produced there on the required terms. Production allocation is ultimately decided by the company, while public contracts must be awarded under applicable procurement and financing rules.
The harder issue is why a country with an established rolling-stock manufacturing base repeatedly concludes large supplier and financing agreements without an open competitive process, transparent lifecycle-cost comparison or binding local-content obligations. Serbia is assuming repayment risk through state guarantees, yet there is little evidence that these purchases have been integrated into a broader policy for domestic production, maintenance, engineering skills or component localisation.
The latest and largest conventional rail acquisition involves 30 electric multiple-unit trains for the BG Voz urban and suburban network. The contract with Spain’s Construcciones y Auxiliar de Ferrocarriles, better known as CAF, is valued at approximately €310mn.
The price implies an average of around €10.3mn per train, although a simple per-unit calculation does not reveal the complete commercial structure. Train length, passenger capacity, maximum speed, onboard systems, spare-parts packages, depot equipment, staff training, warranties and long-term maintenance obligations can materially alter the cost of otherwise similar rolling stock.
The financing structure is clearer. Serbia has guaranteed a credit facility of up to €263.87mn arranged through Deutsche Bank, with support from the Spanish export-credit agency CESCE. This covers roughly 85 per cent of the contract value. The remaining approximately €46.9mn, or 15 per cent, is financed through a long-term loan from the state-controlled Bank Poštanska štedionica, also backed by a Serbian government guarantee.
The purchase therefore follows a conventional export-credit pattern. A foreign export agency supports the buyer’s financing, the supplier receives a commercially bankable contract, and the borrowing entity gains access to long-tenor funding that may not otherwise be available on equivalent terms. The trade-off is that financing and supplier selection become closely connected, potentially reducing room for open competition or domestic production.
Export-credit finance can lower the apparent funding cost, but it does not eliminate the liability. The ultimate risk remains with Serbia because the government guarantees Srbijavoz’s obligations. The value of the state guarantee should therefore be assessed alongside the supplier price, interest costs, guarantee fees, foreign-exchange exposure, maintenance commitments and the railway operator’s capacity to generate enough cash to service the debt.
That capacity is uncertain. Srbijavoz reportedly increased its net loss from approximately €2.2mn in 2024 to around €16mn in 2025, despite receiving public subsidies of roughly €40mn. Passenger railways rarely cover all infrastructure and public-service costs from ticket revenue, but a state guarantee converts the company’s financial weakness into an explicit contingent liability for the national budget.
The government has classified the BG Voz purchase as a strategically important investment, partly connected with the transport requirements surrounding Expo 2027. Yet the contractual delivery schedule weakens that argument. The first CAF trains are due 43 months after payment of the advance, while the remainder may take as long as 58 months.
Even under the most favourable interpretation, the first units are unlikely to arrive before the second half of 2029, roughly two years after the Expo has ended. The trains may still be justified by the long-term requirements of Belgrade’s suburban railway, but Expo cannot credibly be treated as their primary operational deadline.
Serbia’s urban rail expansion does require new rolling stock. BG Voz is expected to develop beyond its current corridors and eventually connect more closely with the airport, Surčin, the National Stadium area and the wider metropolitan transport system. Existing capacity is insufficient for a high-frequency regional railway serving a city of Belgrade’s scale.
The procurement should therefore have been presented as a 30-year metropolitan transport investment, supported by demand forecasts, service-frequency plans, depot capacity, route electrification and projected operating subsidies. Framing the order around a temporary international exhibition risks substituting political urgency for transport economics.
Another €21.6mn agreement signed with China’s CRRC covers nine electric trains intended for the urban and suburban route between Zemun Polje, Nikola Tesla Airport and the future National Stadium. The trains were expected to arrive by the end of 2026.
Serbia has also acquired five CRRC electric trains capable of operating at up to 200 kilometres per hour. According to public audit information, that purchase was worth approximately €54.5mn, implying an average of €10.9mn per unit. The trains have entered service on the modernised Belgrade–Subotica corridor, reflecting the sharp technological divide that has emerged within the Serbian rail system.
On upgraded sections, new electric trains can operate at European high-speed conventional-rail standards. Elsewhere, degraded track, obsolete signalling and ageing vehicles can limit passenger services to speeds of 30 kilometres per hour or less. A train fire near Jasenovik in July 2026, involving a service between Niš and Svrljig, was a reminder that new flagship corridors coexist with neglected regional lines.
Rolling-stock renewal cannot correct that imbalance by itself. Modern trains generate their full economic value only when track condition, electrification, signalling, maintenance depots and operating schedules can support reliable utilisation. Deploying expensive vehicles on a limited number of prestige corridors while regional networks deteriorate produces a transport system with high capital intensity but uneven public value.
Belgrade’s tram procurement has created similar controversy. In 2024, the city contracted 25 low-floor trams from Turkey’s Bozankaya for approximately €63.7mn, excluding value-added tax. That equates to roughly €2.55mn per tram, although some public comparisons have placed the effective cost closer to €2.7mn per unit, depending on the items included, reports local media Radar.
Critics have argued that the vehicles were significantly more expensive than similar Bozankaya trams supplied to Timișoara, reportedly by as much as €800,000 per unit. Such comparisons require care because vehicle length, specifications, spare parts, warranty conditions, financing and contract timing may differ. Even so, a price gap of that scale requires a detailed public explanation, particularly when the procurement is financed by taxpayers and implemented by a municipal transport company under persistent financial pressure.
Questions have also been raised about operational availability. The Centre for Local Government has claimed that only 12 to 14 of the 25 vehicles have regularly entered service because of technical and compatibility problems. Any conclusion about fleet quality would require verified records covering acceptance testing, defects, daily availability, warranty repairs and temporary withdrawals. The absence of such public performance data leaves the debate dominated by political claims rather than engineering evidence.
Belgrade subsequently launched a tender estimated at €188.6mn for another 85 three-section low-floor trams. The procedure was temporarily suspended following objections that the technical requirements could favour Bozankaya. The size of the proposed purchase makes competition especially important: the order would shape Belgrade’s tram fleet, depot requirements and spare-parts supply for decades.
A fragmented fleet carries hidden costs. Vehicles from different manufacturers may require separate diagnostic systems, specialised tools, component inventories, training programmes and maintenance contracts. These expenses do not always appear in the initial purchase price but become significant over a 25- to 35-year operating life.
Fleet standardisation can reduce maintenance complexity, but it can also create supplier dependence when spare parts, software and proprietary systems remain controlled by a single manufacturer. A properly designed tender should therefore balance commonality with open technical standards, price competition and the availability of alternative service providers.
The planned Belgrade Metro adds another layer. Serbia has selected France’s Alstom to supply 32 metro trains, with the rolling-stock element incorporated into a wider French-backed package for the first metro line. Publicly discussed figures place the combined value of train supply, system design and construction-related elements at approximately €915mn.
Comparisons with metro projects in other cities have suggested that Belgrade may be paying substantially more per train than London paid under a €1.7bn agreement involving 94 Siemens trains and associated line work. Yet dividing total project values by the number of trains can be misleading. Metro contracts may include radically different combinations of signalling, platform systems, depots, power supply, design, civil works and long-term support.
The more defensible criticism is that Serbia did not conduct an open tender in which manufacturers could compete on equivalent technical and financial terms. Without competition, it becomes difficult to demonstrate that the Alstom package represents the best available price, technology or lifecycle cost.
France’s role is not limited to equipment supply. The metro is embedded in a wider government-to-government relationship involving French finance, engineering and project participation. Such arrangements can accelerate implementation and provide access to experienced suppliers, but they shift accountability from a visible procurement process to bilateral negotiation.
This makes contract disclosure more important, not less. The public should be able to see what portion of the €915mn relates to rolling stock, what is allocated to design and systems, which risks remain with the supplier, how price escalation is treated and which performance guarantees protect the buyer.
The missing participant across these procurements is Siemens Mobility’s Kragujevac operation. The company employs approximately 1,050 people at two facilities producing trams, railway coaches and train-end structures. The factories are part of Siemens’ European manufacturing network and currently produce vehicles for cities including Nuremberg and Ulm, with potential work also discussed for the US market.
In 2025, the Kragujevac business generated revenue of approximately RSD 21.25bn, equivalent to around €181mn, and net profit of RSD 446.6mn, or about €3.8mn. Those results demonstrate that Serbia already hosts a commercially functioning rail-manufacturing operation capable of meeting international customer requirements.
Representatives of the workforce have claimed that Siemens could have supplied trams to Belgrade for €450,000–€500,000 less per vehicle than the price paid for the Bozankaya units. Siemens itself has declined to comment on individual tenders or third-party procurement arrangements, so the claim cannot be treated as a formal supplier offer. It nevertheless highlights the lack of a documented market test.
There is also no certainty that an order awarded to Siemens Mobility would automatically be manufactured in Kragujevac. Large industrial groups allocate production according to plant capacity, product platforms, workforce skills and existing order books. A Serbian contract would need explicit local-production or industrial-participation provisions to ensure that domestic facilities received the work.
This is where Serbia’s policy gap becomes most visible. The state has approached rolling-stock purchases as separate transport transactions rather than as components of a national rail-industry strategy. Supplier selection, export-credit financing, local manufacturing, maintenance capacity and workforce development have not been connected through a single commercial framework.
A serious local-content model would not require every component to be produced in Serbia. That would be unrealistic for orders of this size. It could, however, require final assembly, body fabrication, interior installation, testing, maintenance training, spare-parts warehousing and the gradual qualification of Serbian component suppliers.
The economic value would extend beyond the initial factory contract. Domestic production retains part of the procurement expenditure through wages, payroll taxes, pension and health contributions, local services, logistics and supplier purchases. It also creates technical knowledge that can support future exports and maintenance activity.
Claims that locally produced trains would be made entirely from Serbian raw materials are too broad. Modern rolling stock depends on imported propulsion electronics, signalling equipment, braking systems and specialised materials. The relevant measure is not whether every input is domestic but the proportion of total contract value retained within Serbia and the level of technology transferred to the local economy.
Even a domestic bid priced 5–10 per cent above the cheapest imported offer could be economically competitive when tax receipts, employment, lower service costs and export potential are included. That does not justify automatic protection of a local factory, particularly one owned by a global group. It does justify evaluating bids through total economic value rather than headline purchase price alone.
The domestic industrial risk is becoming more immediate. Siemens had been expected to begin producing complete aluminium trains in Kragujevac, but the project has reportedly been postponed for three to six months. Between 20 and 30 employees may temporarily take paid leave. The delay reflects competition within Siemens’ own European production network, where German factories are also seeking new work.
A large Serbian order would not resolve every capacity decision, but it could have strengthened Kragujevac’s case for a larger role inside the group. Instead, Serbian public borrowing is helping support industrial activity in Spain, France, China and Turkey while the country’s own rail-production platform remains dependent on export orders secured elsewhere.
Supplier diversity has some strategic advantages. Serbia is not dependent on a single manufacturer and has gained access to technologies from Stadler, CRRC, CAF, Alstom, Bozankaya and Russian producers. The downside is a growing collection of vehicle platforms that may prove expensive to maintain and difficult to integrate.
The value at risk is no longer limited to the initial contract prices. Across the CAF, CRRC, Alstom and Bozankaya purchases, Serbia and Belgrade are creating long-term obligations for maintenance, software support, spare parts, depot modification and staff training. The financing horizon may last longer than the political mandates under which the contracts were approved.
State guarantees make the issue relevant to Serbia’s sovereign risk profile. Each guarantee may appear manageable in isolation, but repeated guarantees for financially weak public companies accumulate as contingent liabilities. They can move onto the budget when an operator cannot service its debt, turning procurement decisions into fiscal costs.
Serbia needs new trains and trams. The old fleet, unreliable regional services and expanding Belgrade metropolitan area make continued investment unavoidable. The weakness lies in the procurement architecture: limited competition, incomplete contract transparency, politically compressed justifications and little integration with the country’s manufacturing base.
Kragujevac already provides Serbia with something many smaller European economies do not possess—a functioning rail-vehicle industry connected to a global technology group. Public procurement could have been used to deepen that base while renewing the transport system. Instead, Serbia is financing modernisation abroad and leaving its most credible domestic rail-manufacturing asset to compete for orders from Nuremberg, Ulm and other foreign cities.
