Europe’s ambition to reduce its dependence on external artificial intelligence infrastructure is beginning to translate into industrial partnerships, as governments and technology companies seek to establish computing capacity under greater European control.
The latest development comes from French computing company Bull and Czech infrastructure operator České Radiokomunikace (CRA), which have signed a memorandum of understanding to support a proposed AI Gigafactory serving the Czech Republic and wider Central and Eastern Europe.
The agreement strengthens a consortium led by CRA that is seeking support through the European AI Gigafactories programme. It also represents the first industrial agreement announced under a bilateral cooperation arrangement on AI and digital technology between France and the Czech Republic.
Although the project remains at the proposal stage, it illustrates how European AI policy is increasingly focused on the physical infrastructure required to develop and deploy advanced computing systems.
The Czech initiative is intended to serve a broader regional market, reflecting commitments by nine Central and Eastern European countries to coordinate AI policy, computing infrastructure and technical expertise. Alongside the Czech Republic, these include Slovakia, Poland, Croatia, Hungary, Lithuania, Latvia, Slovenia and Romania.
For Europe, the underlying challenge extends beyond building additional data centre capacity. AI sovereignty also depends on where computing systems are manufactured, who operates the infrastructure and whether organisations can retain control over the software and data used to develop their applications.
The proposed consortium aims to maximise European technology across its infrastructure and AI systems. CRA would contribute its infrastructure expertise, while Bull would provide computing hardware, software and deployment capabilities.
Bull’s manufacturing operations in France and the Czech Republic, working with Foxconn, could support production of servers for the Gigafactory in both countries. The consortium also intends to incorporate European processors as suitable technologies become available.
Energy efficiency is another consideration. Bull proposes using direct liquid cooling and workload-level energy monitoring, alongside software designed to direct less demanding AI tasks towards appropriately sized models rather than unnecessarily powerful computing resources.
The company’s BullSequana AI platform, developed by teams in France and the Czech Republic, would support organisations deploying applications on the proposed infrastructure. More than 200 DataSentics AI engineers based in the Czech Republic would provide additional technical expertise.
Bull also intends to reserve computing capacity for its own customers, potentially providing an initial source of demand alongside its role as a technology supplier.
However, the memorandum does not establish a construction timetable, investment commitment or confirmed computing capacity. The project must therefore be distinguished from an operational or fully financed data centre development.
Its significance lies in the industrial model being proposed. Rather than treating AI infrastructure simply as a property and power investment, the consortium is seeking to combine regional data centre operations with European manufacturing and software expertise.
Whether that approach can deliver competitive computing capacity at scale remains to be demonstrated. Nevertheless, the agreement highlights an increasingly important question for European policymakers: how much control over AI can Europe realistically achieve without greater control over the infrastructure on which it depends?
