Germany’s Deutsche Telekom and the Schwarz Group are working on plans for a large artificial intelligence data centre in Germany. The facility is described as an AI gigafactory, a site built for very high compute loads. Talks include applying for European Union support under new digital infrastructure programs. Negotiations are progressing, but no binding deal has been signed, with reporting pointing to advanced discussions to apply for EU funding.
The move fits a wider policy push in Europe. Brussels wants new AI factories and gigafactories across member states to expand training capacity, and to keep data within the bloc. The EU’s InvestAI facility aims to mobilize about €20 billion (C$29 billion) for gigafactories, alongside a plan to lift overall data centre capacity over the next seven years. Berlin has said it will support competitive bids that anchor jobs and power-efficient designs.
Consortium targets EU-backed compute
Deutsche Telekom has already mapped out a delivery path with a separate build in Germany. With Nvidia, it is developing an Industrial AI Cloud intended to serve manufacturers, with initial service by early 2026. The partners plan roughly 10,000 GPUs and around 0.5 exaflops of capacity, targeting secure, in‑country workloads. Telekom has framed the project as a sprint to scale.
“Europe’s technological future needs a sprint, not a stroll,” said Timotheus Höttges.
Nvidia’s leadership has pitched the same build as a model for industry. The idea is that compute becomes core to product lines, not a back office tool. That logic mirrors the gigafactory concept now under study with Schwarz. “One for making things, and one for creating the intelligence that powers them,” said Jensen Huang. The Industrial AI Cloud is described as a roughly €1 billion (C$1.45 billion) partnership.
Existing builds show delivery path
Schwarz has been expanding sovereign cloud capacity in Germany through Stackit, its in‑house cloud platform. In November, Schwarz Digits broke ground on a large data centre campus in Lübbenau, Brandenburg, to serve retail and external clients. Early phases target staged power and heat recovery into the local network. The brownfield site benefits from established grid and fibre, reducing time to connect.
Site, power, and grid are decisive for a gigafactory, and so are chips. Europe’s access to advanced accelerators and power purchase agreements will shape procurement. Developers will need municipal permits, grid upgrades, and long‑term contracts for renewable supply. Construction phasing can align capacity with demand, cutting carrying costs while EU funding decisions progress.
The next steps appear to centre on a formal proposal to EU bodies and German authorities. If approved, the project would likely be financed through a mix of private capital and EU‑backed instruments, plus potential state and municipal supports.
Long‑lead equipment and transformer orders would follow quickly to hold schedule. Delivery would then hinge on power interconnections, cooling choices, and recruitment for operations.
For now, the partners are signalling intent and testing feasibility. Germany’s industrial base seeks local compute for design, simulation, and training. If backed, the gigafactory would add to Telekom’s pipeline and Schwarz’s growing footprint, and anchor a larger EU push to build sovereign AI capacity at scale.