Open data · CC BY 4.0 · a lean pilot, growing as data matures
AI

AI Infrastructure Monitor — Europe

Named European AI data-center campuses & investments · source-linked

Updated Pilot v0.1

Europe is racing to build sovereign AI capacity — one campus and gigawatt deal at a time.

This is a lean, source-linked pilot: an open ledger of Europe's largest AI / hyperscaler data-center campuses and investment programs. Every record is verified against primary sources. We start with what we can stand behind and add tabs (grid, capital, compute) as verified European data matures — never faking parity with the more mature US dashboard.

18
tracked records
10 + EU
countries + EU-wide initiative
11
operators & JVs
100%
records source-verified
How to read this pilot. The ledger mixes three record types: campuses (a sited facility, sometimes with disclosed MW), capex programs (country-level investment, no single sited MW — most hyperscaler entries), and one policy program (the EU-wide InvestAI gigafactories initiative). Because of that, MW figures are not directly comparable and should not be summed — see the caveats in the open dataset. Known gap: Ireland/Dublin, the #1 addition for the next version.

See the Campuses tab for the full ledger.

European AI data-center campuses & investments

Named builds and power/investment deals across Europe. Each row links to a verified primary source; per-record provenance, capacity type, and confidence travel with the open dataset.

18 source-verified records — campuses, country capex programs, and the EU's InvestAI gigafactories.

The ledger

Sorted by disclosed capacity. Hyperscalers rarely publish site MW, so many rows show a dash (—) for MW with the investment figure in the note.

Loading ledger…

Do not sum the MW column. Disclosed figures mix ultimate campus targets (e.g. Sines 1.2 GW vs ~26 MW live today), a wind-supply figure (Meta Odense), and one IT-critical-load number — and most records disclose no MW at all. Always read capacity alongside status and capacity type. Full caveats & known gaps are in the dataset.