TL;DR
Schwarz Group is investing €11 billion in a 200-megawatt AI data center near Lübbenau, Germany, without government subsidies. The project shows how patient industrial capital could expand Europe’s computing capacity, though dependence on one private provider would not amount to full technological independence.
Schwarz Group, the owner of Lidl and Kaufland, is building an €11 billion AI data center near Lübbenau, Germany, without government subsidies, according to project details compiled by Thorsten Meyer AI. The planned 200-megawatt facility, designed for as many as 100,000 graphics processing units, tests whether Europe’s industrial companies can provide computing capacity more reliably than state-backed technology programs.
The project is being developed by Schwarz Digits, the group’s technology division, at a former coal-fired power station in Brandenburg. The reported commitment includes about €2.5 billion for construction and €8.5 billion for technology. Its first module is scheduled to enter service by the end of 2027, though the final GPU count and installation timetable have not been disclosed.
The investment exceeds five times Schwarz Digits’ reported annual sales of about €1.9 billion. The division operates STACKIT, a European cloud platform with roughly 20,000 servers and 22.5 petabytes of storage, and draws financial backing from a retail group with approximately €175 billion in yearly revenue, 575,000 employees and operations in 32 countries.
Thorsten Meyer AI reported that the Lübbenau facility is receiving no state subsidy. That separates it from Intel’s proposed semiconductor plant in Magdeburg, which had been linked to €9.9 billion in German aid before Intel cancelled the project in July 2025. The comparison is not exact: one project concerns AI computing infrastructure, while the other would have manufactured chips.
The supermarket that bought Europe’s AI: why industrial capital beats government money
The €500M cheque got the headlines. The €11 billion one is the story. On a dead coal plant in Brandenburg, the owner of Lidl is building a 200 MW, 100,000-GPU AI data centre — with no government subsidy at all.
Europe looked for its AI advantage in regulation, talent and Brussels programmes. Magdeburg is what that produces. The real advantage was sitting in the Mittelstand: enormous, foundation-owned industrials with recession-proof cash, decades of proprietary data, inherited KRITIS compliance — and nobody to answer to. Patient capital is the one thing American AI structurally cannot buy. But be precise: Europe’s sovereignty didn’t get nationalised — it got privatised. The answer to American corporate power over European AI is turning out to be German corporate power, with a toll booth attached. That may be the better trade. Just don’t call it independence — call it a change of landlord, and read the lease.
Private Capital Tests Europe’s AI Model
The Lübbenau project could give European companies access to more locally operated AI infrastructure at a time when much of the advanced cloud and accelerator market is controlled by US suppliers. A large domestic facility may help customers meet European requirements covering data location, security and operational control, though the underlying processors may still come from foreign manufacturers.
Its financing also highlights a possible advantage held by foundation-controlled industrial groups. Schwarz Group has no publicly traded shares and can make long-term investments without quarterly pressure from outside shareholders. Thorsten Meyer AI argues that this structure gives private capital an institutional time horizon more commonly associated with governments.
The model has limits. Few European companies combine large and stable cash flows, proprietary operating data, regulated-infrastructure experience and an established cloud subsidiary. Bosch, Zeiss, Bertelsmann and Würth have patient ownership structures, but none has a cloud operation matching STACKIT’s reported scale. Publicly traded groups such as SAP, Siemens and Deutsche Telekom face different capital constraints.
high performance GPU for AI data centers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Retail Scale Built the Cloud Base
Schwarz established Schwarz Digits as a separate division in September 2023, bringing together STACKIT and other digital operations. Its technology systems grew from the needs of Lidl, Kaufland and related businesses, which process more than 13 billion transactions each year across 32 countries.
That retail base gave the division experience with high-volume infrastructure and compliance frameworks including BSI C5, ISO 27001, SOC 2 and DORA. According to the source analysis, STACKIT had about a seven-year operating head start before the Lübbenau investment, reducing the need to create a cloud platform alongside the physical data center.
The development comes as European policymakers and companies seek alternatives to US cloud concentration. Other industrial investments include ASML’s reported €1.3 billion stake in Mistral AI, but that is an investor relationship rather than direct ownership of an infrastructure platform.
“A new landlord is not freedom.”
— Thorsten Meyer AI
enterprise cloud storage solutions
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Capacity, Customers and Control Stay Open
It is not yet clear how quickly the facility will reach its proposed 100,000-GPU ceiling, which accelerator models it will use or how much capacity will serve Schwarz Group itself. Public information also does not establish the expected mix of external customers, pricing or utilization.
The project’s commercial case depends partly on STACKIT winning business outside its parent group. An investment of €11 billion against €1.9 billion in divisional sales creates execution risk if demand falls short. Private ownership also provides less routine financial disclosure than public markets require, leaving questions about returns, contracts and cost overruns.
Claims of sovereignty remain qualified. A reported five-year STACKIT exclusivity arrangement could create a domestic chokepoint, while imported chips and software would preserve other dependencies. Schwarz also uses Google Workspace for its workforce, showing that its technology operations are not fully detached from US providers.
industrial AI server racks
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
First Lübbenau Module Faces 2027 Test
Attention will shift to whether Schwarz Digits completes the first Lübbenau module by late 2027 and discloses firm customers, installed accelerator capacity and power arrangements. Construction progress will provide the first measurable test of the reported schedule.
European industry and policymakers will also watch whether other companies adopt the industrial-anchor model or continue relying on subsidies and multinational cloud providers. The central test is whether private ownership can deliver durable European capacity without replacing one concentrated dependency with another corporate gatekeeper.
GPU cooling systems for data centers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is Schwarz Group building near Lübbenau?
Schwarz Group is building a 200-megawatt AI data center on a former coal-power site in Brandenburg. It is designed to support up to 100,000 GPUs.
How much is the project expected to cost?
The reported commitment is €11 billion, comprising about €2.5 billion for construction and €8.5 billion for technology.
Is the German government subsidizing the facility?
According to the supplied project analysis, the facility is receiving no government subsidy. Detailed financing terms have not been made fully public.
Does the project make Europe independent in AI?
No. It could add European-operated computing capacity, but dependence on foreign chipmakers, software providers and a single domestic operator would remain.
When will the data center begin operating?
The first module is scheduled for the end of 2027. The timetable for reaching full capacity has not been disclosed.
Source: Thorsten Meyer AI
