Summary
- Brinell Compute is preparing construction of a €3bn AI datacentre in Maritsa, Bulgaria.
- The site is planned for the Rakovski Industrial Zone north of Plovdiv.
- Customer, capacity, and financing details remain limited, but the project shows how AI infrastructure interest is spreading east.
Brinell Compute is preparing construction of a €3bn AI datacentre in Maritsa, Bulgaria, adding Eastern Europe to the geography of large-scale European compute ambitions.
The German technology company plans to build the facility in the Rakovski Industrial Zone north of Plovdiv. The site is expected to occupy around 2,000 acres, although no details have yet been released on the project’s estimated compute capacity.
Local officials have pushed back against interpretations of the scheme as a physical robotics plant. The site will not be “a conveyor for the physical production of robots,” they said, describing it instead as “a specialised datacentre with enormous computing power and supercomputers, where complex algorithms are trained, and data sets are processed.”
Details of Brinell’s investment emerged earlier this year through Bulgarian government records, when the country’s Council of Ministers signed a memorandum of understanding with the company to develop a datacentre in the region. The ultimate corporate users of the facility remain unclear, though negotiations with US technology companies have been discussed locally.
The proposal is striking because Bulgaria is not usually placed alongside Frankfurt, Dublin, Amsterdam, Paris, or London in discussions about major European datacentre expansion. Its market is relatively large by Eastern European standards, but it has not attracted the same density of foreign investment as the continent’s largest hubs.
AI workloads may change that geography. Developers searching for land, power, political support, and faster approval routes are looking beyond established markets where grids are crowded and planning risks have grown. Smaller European countries can become more competitive if they can offer credible infrastructure, energy access, and government backing.
The uncertainties around the Brinell project still need careful handling. No confirmed customer list, detailed financing structure, or compute specification has been disclosed. In the current AI infrastructure market, early-stage announcements can move faster than chips, grid connections, and tenants.
Even so, the project reflects a wider pattern. AI infrastructure is no longer confined to the places where cloud demand first concentrated. Governments are courting datacentre investment as industrial policy, developers are chasing power wherever it can be secured, and technology companies are weighing whether distributed European capacity can reduce dependence on a small number of mature markets.
Bulgaria’s opportunity will depend on execution. If Brinell can turn political support and a large industrial site into a powered, connected, customer-backed facility, the country could become a more visible node in Europe’s compute system. If the project slips, it will still show how widely the AI infrastructure race is now being marketed across the continent.






