Serbia has begun serial production of humanoid robots at a Chinese-owned factory in Šabac, taking an initial €20 million investment into one of the fastest-growing areas of advanced manufacturing and testing whether the country’s foreign-investment model can move beyond assembly into higher-value engineering and physical artificial intelligence.
Chinese automotive-components manufacturer Minth Group, working with robotics developer AGIBOT, launched the production line at Minth’s existing industrial complex in Šabac on Aug. 29.
The first robot produced at the site received serial number 001 during the opening ceremony.
The companies and Serbian authorities describe the operation as Europe’s first mass-production base for humanoid robots.
The immediate investment is relatively modest by the standards of Serbia’s largest industrial projects.
What makes the project potentially significant is the proposed scale of what follows.
Minth plans a wider Robotics Industrial Park in Inđija, which would take total investment in the programme to about €200 million. Once fully developed, the project is targeting annual capacity of as many as 20,000 humanoid robots and robotic dogs, primarily for European and other international markets.
Serbian authorities have said an initial target would be to assemble more than 5,000 robots annually, underscoring an important distinction between what exists today and the eventual industrial ambition.
The Šabac line is operating.
The 20,000-unit capacity, broader product portfolio and Inđija robotics park remain targets rather than current production.
That distinction will determine the project’s economic importance for Serbia.
If the factory remains predominantly an assembly operation using imported Chinese robotics systems and components, its contribution would resemble much of Serbia’s existing manufacturing foreign direct investment: additional exports, employment and industrial output, but relatively limited domestic technological ownership.
If Minth and AGIBOT progressively transfer engineering, software integration, testing, component sourcing and product-development functions into Serbia, the project could represent something substantially different.
It could move Serbia into the European supply chain for embodied or physical AI, an emerging field in which artificial intelligence is integrated with machines capable of sensing, navigating and manipulating objects in real industrial environments.
That is a considerably higher-value proposition than conventional assembly.
Minth says the robots produced through the project use multimodal perception systems, tactile sensing and AI-based control designed for functions including material handling, loading and unloading, industrial inspection and repetitive or hazardous production tasks.
The commercial opportunity is therefore tied primarily to factories rather than consumer robotics.
Manufacturers across Europe are dealing simultaneously with ageing workforces, labour shortages, pressure to automate and the need to maintain industrial competitiveness against China and the United States.
Humanoid and mobile robotic systems are being developed as one possible response.
Unlike conventional industrial robots, which generally operate in fixed positions performing predetermined tasks, new humanoid systems are intended eventually to work inside factories and warehouses designed for humans.
Their economic advantage, if the technology develops as expected, would be flexibility.
A machine able to navigate an existing production environment, recognise different objects and perform several physical tasks could reduce the need to redesign entire factories around dedicated automation equipment.
Whether humanoid robots can achieve that economically at scale remains uncertain.
Hardware costs, reliability, battery endurance, safety, AI performance and the difficulty of manipulating objects in unpredictable environments remain significant constraints.
But investment is accelerating rapidly.
AGIBOT, also known as Zhiyuan Robotics, has emerged as one of China’s major humanoid-robot developers.
According to figures cited by Serbia’s Development Agency, AGIBOT delivered 5,168 humanoid robots in 2025 and about 8,400 units in the first half of 2026, figures which would place it among the world’s largest suppliers. Those numbers are company-related market estimates and should not be treated as independently audited global market-share data.
For Serbia, attracting such a manufacturer provides an opportunity to participate early in a sector that could become materially larger if industrial adoption accelerates.
Minth already gives the project an industrial base.
The group entered Serbia in 2018 and says it has established 10 manufacturing plants in the country, creating an existing production, logistics and workforce platform onto which robot manufacturing can be added.
That makes the Šabac project different from a technology start-up attempting to build manufacturing capacity from scratch.
Minth already operates large industrial facilities, particularly in automotive components, where quality control, supply-chain management and high-volume manufacturing are established disciplines.
The company can therefore provide industrialisation capability while AGIBOT supplies robotics technology.
For Serbia, however, the strategic question is what happens between those two companies.
The country has attracted tens of billions of euros of foreign investment over the past two decades, particularly into automotive components, electrical equipment, metals, tyres and other manufacturing.
That model has transformed exports and employment.
Its weakness has been that a large portion of the highest-value intellectual property, product development and strategic procurement remains outside Serbia.
Humanoid robots provide a test of whether that model can evolve.
A robot assembled from imported motors, sensors, computing hardware, battery packs and software generates considerably less domestic value added than a system for which Serbian engineers participate in control software, mechanical integration, testing, training data, maintenance and component design.
Local sourcing will therefore be more informative than headline output.
The planned capacity of 20,000 units sounds significant, but production volume alone will not reveal how much of the value chain Serbia actually captures.
Employment provides another test.
Current plans point to roughly 200-250 jobs at the new Šabac operation.
That is relatively limited compared with traditional labour-intensive manufacturing investments, but advanced automation factories are not primarily attractive because of the number of workers they employ.
Their potential importance lies in productivity, salaries, engineering content, exports and the supplier ecosystem they create.
A plant employing several hundred engineers, technicians and specialised manufacturing staff can generate greater economic value per employee than a conventional assembly factory employing several thousand lower-skilled workers.
Serbia’s policy challenge is therefore to maximise the technological spillover.
That could mean connecting the project with engineering faculties, AI researchers, software companies, automation specialists and domestic industrial suppliers.
The country already has a sizeable software and engineering workforce.
Linking that talent base with physical manufacturing has long been one of Serbia’s unresolved industrial-development opportunities.
Robotics could provide such a bridge.
Software accounts for an increasingly important share of a robot’s economic value.
Computer vision, motion planning, machine learning, fleet management and human-machine interaction all require capabilities that Serbia’s technology sector can potentially provide.
The hardware opportunity is also broader than humanoid robots alone.
Minth has indicated that its Serbian expansion could eventually include unmanned aerial vehicles and advanced batteries, suggesting that the group sees the country as a potential European manufacturing platform for several related technologies.
That would make the planned Inđija industrial park strategically more important than the initial Šabac production line.
A genuine robotics cluster incorporating robots, batteries, drones, component suppliers, testing and engineering would have a different economic profile from a single assembly factory.
But the investment remains at an early stage.
The announced €200 million programme should therefore be separated from the €20 million already committed to the first phase.
Serbia has frequently attracted ambitious investment announcements whose ultimate scale depended on market conditions and subsequent phases.
Robot demand will determine whether this project follows the same pattern.
Europe is a potentially large market, but it also presents regulatory and geopolitical risks.
Chinese technology companies are facing increasing scrutiny in Western markets over cybersecurity, data security, industrial dependence and strategic technology supply chains.
That scrutiny has already reached advanced robotics.
The United States has tightened restrictions around some Chinese robotics technologies on national-security grounds, illustrating the potential geopolitical complications facing Chinese-designed physical-AI systems seeking international markets.
For a Serbia-based factory aimed partly at Europe, those issues could eventually matter as much as production cost.
European customers will need confidence in cybersecurity, data handling, machine safety and the integrity of software updates.
Humanoid robots operating inside industrial plants could potentially interact with sensitive production data, cameras, factory networks and operational systems.
That makes certification and data governance commercially important.
Serbia’s position outside the EU but closely connected to its industrial market creates both opportunity and risk.
It offers Chinese manufacturers a relatively competitive manufacturing location with free-trade access and established logistics links toward Europe.
But production in Serbia does not automatically remove European concerns over Chinese technology or origin.
The eventual localisation of software, components and corporate control could therefore affect the project’s ability to penetrate sensitive European industries.
For Serbia, the best outcome would be to become more than a low-cost manufacturing bridge.
A domestic robotics ecosystem would require local engineers to understand and modify the technology, domestic suppliers to enter the bill of materials and Serbian institutions to build expertise in testing, certification and industrial AI.
That takes substantially longer than installing a production line.
The first robot carrying serial number 001 therefore represents an industrial milestone, but not yet proof that Serbia has entered the high-value end of robotics manufacturing.
The economic test begins after production starts.
The relevant indicators will be the share of local components, number and quality of engineering jobs, value of exports, domestic R&D activity, Serbian supplier participation and whether the planned €200 million Inđija expansion is executed.
Those measures will determine whether the Minth-AGIBOT project becomes another successful foreign-owned assembly operation or the beginning of a more sophisticated industrial cluster.
Serbia has spent much of the past decade building its manufacturing strategy around attracting global factories.
Humanoid robotics offers it an opportunity to move one step further.
The opportunity is not simply to manufacture machines powered by artificial intelligence.
It is to capture enough of the engineering, software and intellectual value surrounding those machines to change the position Serbia occupies in the supply chain.
That is the real significance of the first robot rolling off the Šabac line.
I kept the €200 million figure clearly separated as planned investment, rather than treating it as already committed capital, and made localisation versus assembly the central investment-risk angle.
