The country has assembled an unusual bridge between software and manufacturing. Its next test is to retain intellectual property, product mandates and research spending as wages converge.
The cluster is real
Serbia’s strongest innovation claim is not that it is another low-cost software destination. It is that engineers can move between code, control systems, vehicle electronics, power equipment and production lines within one regional ecosystem. Microsoft has operated a development centre in Serbia for two decades. Schneider Electric’s Novi Sad hub employs more than 1,000 people around energy-management technology. Continental develops automotive systems in Novi Sad; Databricks has built a Belgrade development centre; domestic and regional groups such as HTEC, Vega IT and Nordeus add founder-led depth.
Manufacturing gives this software base physical problems to solve. ZF’s e-mobility operations in Pančevo, Continental and Brose, Bosch and other automotive suppliers, Stellantis in Kragujevac and the tyre plants create demand for embedded software, test engineering, automation, tooling and process improvement. Toyo Tire’s new European R&D centre in Inđija is due to operate from January 2027 beside its factory and proving ground, consolidating materials, design and production work in one site.
That combination is scarce in a small economy. Bucharest, Budapest, Kraków and Prague offer larger labour markets; Serbia can differentiate through cross-disciplinary teams and faster access to a working plant. ‘R&T’ — research and technology — is useful only if it connects laboratory or software work to deployment. Serbia’s opportunity is to become the place where European products are engineered for production, not simply where a remote backlog is cleared.
The valuable Serbian engineer is no longer the cheapest coder. It is the person who can turn code into a product and a product into a production process.
The spending base remains too thin
Serbia spent RSD90.58bn on research and development in 2024, equal to 0.94 per cent of GDP. The EU average was 2.24 per cent. Serbia employed 30,810 people in R&D, including 19,323 researchers, and businesses accounted for almost two-thirds of the 487 organisations undertaking research. The institutional map is broader than the GDP ratio suggests, but resources are spread across universities, public institutes and company centres with uneven commercial links.
A sub-1 per cent research intensity is not fatal for an engineering-service hub. It is a constraint on becoming a product and intellectual-property hub. Multinationals can allocate routine development to Serbia while keeping architecture, patents, capital budgets and product ownership at headquarters. Domestic firms can build talented teams but lack patient capital and global distribution. Public institutes may publish without an industrial customer. The missing middle is a contract, laboratory and financing structure that joins them.
The talent pool is another ceiling. Serbia’s universities produce strong engineers, but emigration, demographic decline and rapid wage increases make the old volume model hard to sustain. A subsidy that creates hundreds of jobs can intensify hiring competition without adding teachers, doctoral researchers or experienced managers. Companies then poach from one another and call the wage increase ecosystem growth.
What the corporate map says
The most durable centres have a product mandate. Microsoft Serbia contributes to global engineering rather than serving only local sales. Schneider’s hub works on energy software linked to a global installed base. Continental’s engineers sit near an automotive supply chain. Databricks extends a global data platform. Toyo’s new centre links European product development directly to a Serbian plant and test track. These models embed teams because moving them would disrupt products, not merely reduce headcount.
Domestic value capture depends on spillovers that cannot be assumed. Engineers need permission and capital to launch suppliers or spin-outs without violating intellectual-property rules. Universities need shared equipment and company-defined research problems. Local toolmakers and testing laboratories need certifications that let them serve more than one anchor factory. The state should measure how much R&D procurement reaches Serbian SMEs and how many patents, licences and product revenues arise locally.
M&A will shape the hub. Global technology groups can acquire Serbian teams to secure talent; industrial buyers can buy niche automation and engineering firms; domestic consolidators can combine services with proprietary software. The risk is that every successful product company is absorbed and its strategic decisions migrate. The counter is not to block exits, but to recycle founder capital and management experience into the next generation.
From headcount incentives to mandate incentives
Serbia’s investment regime has been effective at attracting employment and capex. For R&D, the unit of success should change. Incentives should reward multi-year product mandates, business-funded doctoral work, patents used in production, local supplier development and export revenue from owned technology. A company receiving support for an engineering centre should report whether it controls a budget, a product line or merely a staffing target.
The country also needs infrastructure that engineers use: reliable power, cloud and high-performance computing, open research equipment, technology-transfer offices and procurement paths for first customers. Regulation should make employee equity and cross-border licensing straightforward. Schools and universities need more capacity in electronics, materials, mechanical engineering and data science, not a policy assumption that ‘IT’ is one interchangeable skill.
Serbia is already a regional R&D and R&T centre in selected niches. It is not yet a regional research superpower, and it should resist marketing that erases the difference. Its route upward is to connect software to machines, foreign platforms to domestic suppliers, and public research to paying customers. If it does, rising wages will be evidence of productivity. If it does not, the same wage growth will eventually send the next engineering mandate elsewhere.
