On July 30, 2026, the European Union launched the tender process for up to seven so-called "AI Gigafactories" – modular high-performance data centers designed to make Europe competitive in the global AI race against the United States and China. With a total investment volume exceeding €30 billion and a submission deadline of November 12, 2026, this is the EU's largest coordinated digital infrastructure initiative to date. But who actually builds and finances these facilities – and what role do private infrastructure owners play?
The Program at a Glance: Two Classes, Seven Sites
The EU tender through EuroHPC Joint Undertaking is structured into two performance classes:
- Class 1: Up to four projects with EU funding of up to €500 million per site (€100m in Phase 1, €400m in Phase 2)
- Class 2: Up to three large-scale projects with funding of up to €1 billion per site (€200m in Phase 1, €800m in Phase 2)
The overall program combines approximately €10 billion in public funds (EU and member states) with at least €20 billion in private capital. Award decisions are planned for early 2027, with construction expected to begin within the same year.
Technically, the facilities are designed to provide massive GPU clusters for AI training and inference – with power densities that will exceed current European top-tier data centers by a factor of four. This means rack power levels of 80 to 120 kW, direct liquid cooling as standard, and grid connections in the range of 100 to 500 MW per site.
Why Europe Needs This Step
The context is clear: US hyperscalers currently control approximately 70 percent of the European cloud market. The US and China together operate more than 90 percent of specialized AI data centers worldwide. European startups, research institutions, and mid-sized companies are structurally dependent on non-European infrastructure – with all the geopolitical and data protection consequences that entails.
Experts including Monika Schnitzer and Philippe Aghion calculated in a strategy published in September 2026 that achieving a 15 percent share of global AI computing capacity by 2030 (equivalent to 45 GW) would require investments of approximately €1.3 trillion. The EU initiative is an important first step – but not a self-fulfilling prophecy.
The Critical Bottleneck: Energy and Grid Connection
The biggest obstacle for AI Gigafactories is not capital – it is energy. In established European data center hubs (Frankfurt, London, Amsterdam, Paris, Dublin), grid connection waiting times currently span five to ten years. Frankfurt is effectively closed to further large-scale projects until 2030.
For AI Gigafactories requiring 100 to 500 MW of power per site, a secured, fast grid connection is not an option – it is a prerequisite. Three solution patterns dominate the industry:
- Behind-the-Meter Generation: On-site power plants (gas, HVO, biomethane) bypass the grid queue
- Co-located Renewables: Sites developed adjacent to dedicated wind, solar, or hydro generation in Iberia or Scandinavia
- Grid-Asset Model: Data centers as active grid stabilizers providing ancillary services, thereby receiving prioritized grid access
This is precisely where the strategic strength of infrastructure owners lies – those who understand energy not as a cost factor, but as a core competency integrated into their business model.
The Role of Private Infrastructure Owners
The EU tender is deliberately structured as a public-private partnership: public funds mobilize private capital at a ratio of 1:2. This means private infrastructure owners and developers are not merely welcome – they are structurally necessary.
What distinguishes a genuine infrastructure owner from an operator or service provider?
Own-and-hold rather than sale: Infrastructure owners develop data centers, finance them through special purpose vehicles (SPVs), and hold them in their own portfolio long-term. Monetization occurs through long-term NNN (triple net) lease agreements with tenors of 15 to 20 years to hyperscalers or operators.
Energy as core competency: Sites with secured energy supply – direct 380/400 kV grid connection, gas and hydrogen infrastructure – are indispensable for AI Gigafactories. Infrastructure owners who manage energy as part of their portfolio can meet these requirements.
Subsidy advisory and financing structure: Combining EU funding, national programs, and private capital requires specialized expertise. Experienced infrastructure owners can realize subsidy rates of up to 50 percent of construction costs.
European sovereignty: AI Gigafactories must be European-controlled – both legally and operationally. European infrastructure owners fulfill this requirement structurally.
Market Dynamics: Secondary Markets Gaining Importance
Bottlenecks in FLAP-D markets (Frankfurt, London, Amsterdam, Paris, Dublin) are accelerating the shift to secondary locations. For AI Gigafactories, the most promising sites include:
- Brandenburg/Berlin: Germany's fastest-growing secondary market, with capacity expected to triple by 2030 and grid expansion prioritized
- North Rhine-Westphalia: Industrial sites with existing energy infrastructure (Bedburg, Lippetal)
- Iberia (Spain, Portugal): High availability of renewable energy, growing grid capacity
- Poland: Lower land costs, EU member state, growing digital infrastructure
Germany's National Data Center Strategy of March 18, 2026 set the target of doubling IT connection capacity by 2030 and quadrupling AI/HPC capacity. This creates political tailwind for infrastructure owners investing in Germany.
Regulatory Framework: EnEfG and EU Efficiency Standards
AI Gigafactories must meet the strictest efficiency standards from the outset. In Germany, new data centers commissioned from July 1, 2026 must achieve a PUE of no more than 1.2. The Energy Reuse Factor (ERF) – the share of waste heat that must be reused – rises from 10 percent (from July 2026) to 20 percent (from July 2028).
At the EU level, the European Commission proposed a new common rating scheme for energy and water consumption on September 21, 2026. Binding minimum performance standards are expected by end of 2028.
For infrastructure owners, this means: anyone investing today in AI Gigafactory-grade infrastructure must plan liquid cooling, waste heat utilization, and grid integration from the start – not as retrofits, but as design principles.
Investment Volume and Return Expectations
The German hyperscale market is estimated at approximately USD 1.78 billion for 2025, with projections reaching USD 5.5 billion by 2031. Industry experts broadly calculate approximately €10 million per megawatt of installed IT capacity.
For infrastructure investors, return expectations are attractive: long-term NNN leases with investment-grade hyperscalers provide stable, inflation-indexed cash flows. Comparable projects have achieved annual returns exceeding 20 percent in growth portfolios.
The combination of EU funding (up to €1 billion per site), national programs (degressive depreciation of up to 30% in Germany since July 2025), and private capital enables financing structures that are attractive to institutional investors.
Conclusion: The Window Is Open – But Not Indefinitely
The EU AI Gigafactory tender represents a historic window for European infrastructure owners. The combination of public funding, political tailwind, and structural demand from hyperscalers and AI companies creates an investment opportunity that will not recur in this form.
The submission deadline is November 12, 2026. Award decisions follow in early 2027. Those who can demonstrate sites with secured energy supply, planning approval readiness, and financing structure today have a decisive advantage.
European sovereignty in AI infrastructure is not a political slogan – it is a business model. And infrastructure owners who think energy, financing, and location as an integrated system are the actors who can realize this model.
ArcTeron develops, finances, and holds hyperscale data centers in its own portfolio – as an infrastructure owner and energy partner for Europe. Learn more at arcteron.ai