AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

Schwarz Group, the owner of Lidl and Kaufland, is building a €11 billion AI data centre near Lübbenau, Germany, with planned capacity for up to 100,000 GPUs. The privately held retailer is funding the project without reported state subsidies, but its ability to attract outside customers and reduce Europe’s dependence on US technology companies remains unproven.

Schwarz Group, the German owner of Lidl and Kaufland, is building a €11 billion AI data centre on a former coal-power site near Lübbenau, Brandenburg, according to a July 16 report from Thorsten Meyer AI. The planned 200-megawatt facility, designed to accommodate up to 100,000 graphics processors, would give the retailer an unusually large role in Europe’s effort to develop computing infrastructure outside the dominant US cloud platforms.

The reported commitment comprises about €2.5 billion for construction and €8.5 billion for technology. The site is under construction, with its first module scheduled to enter service by the end of 2027. Thorsten Meyer AI, citing data-centre and industry publications, reported that the project will use green electricity and receive no government subsidy.

The facility will be operated through Schwarz Digits, the group’s technology division and owner of the STACKIT cloud platform. Schwarz Digits generates about €1.9 billion in annual sales, making the Lübbenau commitment more than five times the division’s yearly revenue. Its parent group has far greater resources: about €175 billion in annual revenue, 575,000 employees and operations across 32 countries.

STACKIT has spent roughly seven years developing European cloud capacity and reportedly operates about 20,000 servers with 22.5 petabytes of storage. The division also carries certifications and controls linked to German critical-infrastructure requirements, financial-sector rules and information-security standards.

At a glance
reportWhen: under construction as of July 2026; fir…
The developmentSchwarz Group has begun building a €11 billion, 200-megawatt AI data centre that could become one of Europe’s largest privately financed computing projects.
AI Dispatch · Reality Check · 16 July 2026

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.

▲ Under construction
€11B · Lübbenau
Schwarz Digits. 200 MW · up to 100,000 GPUs · brownfield coal site · green power · first module end-2027. State aid: €0.
vs
▼ Cancelled
€9.9B · Magdeburg
Intel’s fab. Years negotiating German state aid — cancelled outright, July 2025. A hole in the ground and a lesson.
The size of the bet — Schwarz Digits is wagering >5× its own top line on one site
Schwarz Digits revenue /yr€1.9B
Lübbenau commitment€11B  ·  €2.5B construction + €8.5B technology
Context: Schwarz Group turns over ~€175B a year — 575,000 employees, 32 countries, 13B+ transactions. The compliance pedigree (BSI C5 · ISO 27001 · SOC 2 · DORA) wasn’t built for AI — it was inherited from selling groceries at KRITIS scale.
The five preconditions — why this is a special case, not a template
01
Scale
€175B revenue; recession-proof cash. “We always eat.”
02
Data
13B+ transactions/yr across 32 countries
03
KRITIS
Critical-infrastructure status → inherited certifications
04
Cloud subsidiary
STACKIT’s ~7-yr head start: 20k servers, 22.5 PB
05
Long-term ownership
Dieter Schwarz + Stiftung. No public shareholders.
#5 is the one that decides everything. What lets Schwarz make a decade-long, €11B, unsubsidised bet isn’t German engineering or EU regulation — it’s the absence of public shareholders. The US structurally can’t replicate it (its giants are shareholder-disciplined); China does patient capital through the state. Germany has a third model: the Stiftung — private capital on a public-institution time horizon. Bosch (~94% Robert Bosch Stiftung), Zeiss, Bertelsmann, Würth all have it.
Who’s next — run the preconditions and the field narrows fast
Candidate
Has
Missing
Bosch
~€90B rev · foundation-owned · industrial data · already in Aleph Alpha
no cloud subsidiary at STACKIT’s maturity — the bit you can’t buy fast
DT / T-Systems
real sovereign cloud · telco KRITIS
publicly traded, state shareholder — fails ownership
SAP · Siemens · Ionos
data + scale; circling EU AI-DC bids
all publicly traded; none has the combination
ASML
already did it — €1.3B into Mistral, ~10%, largest shareholder
— but that’s the investor model, not the anchor model
Zeiss · Bertelsmann · Würth
foundation ownership + patience
no cloud infrastructure; mostly sub-scale
⚠ The critique — a new landlord is not freedom
Swapping AWS for Schwarz is still dependency — 5-yr STACKIT exclusivity = a chokepoint What makes it durable makes it opaque — no shareholders, no disclosure Founder control = succession risk The paradox: STACKIT hosts Google Workspace for Schwarz’s 575k staff €11B vs a €1.9B division — if STACKIT can’t win externally, it’s the priciest lesson in German corporate history Golem, Aug ’25: the sovereign cloud is “a fairy tale
The take

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.

Sources: DCD, ESM, Smart Country Convention, Silicon Saxony, Xpert.digital (Lübbenau: €11B · 200 MW · ~100k GPUs · end-2027); Wikipedia/FAZ/Handelsblatt (Schwarz Digits, STACKIT, XM Cyber, BSI Mar ’25, Google Nov ’24); five-preconditions framework via the industrial-anchor analysis on StrongMocha; TechCrunch/Penchan (ASML–Mistral); Golem.de Aug ’25. Several deal terms reported, not confirmed; the merger awaits regulatory approval. Not investment advice.
thorstenmeyerai.com

Retail Wealth Funds AI Capacity

The project shows how industrial cash flow can finance European AI infrastructure on a scale often associated with governments or global technology companies. Schwarz can draw on steady retail revenue, transaction data and existing computing operations while accepting a longer return period than many publicly traded businesses.

That matters because advanced AI development depends on access to large GPU clusters, electricity and cloud services. If STACKIT attracts companies and public institutions, Lübbenau could provide a European-hosted alternative for customers concerned about data location, regulatory exposure or reliance on Amazon Web Services, Microsoft Azure and Google Cloud.

AI Data Center Infrastructure Engineering: Power Distribution, Liquid Cooling, High-Density Networking, and Energy Efficiency for GPU Training ... Hardware & Compiler Engineering Series)

AI Data Center Infrastructure Engineering: Power Distribution, Liquid Cooling, High-Density Networking, and Energy Efficiency for GPU Training … Hardware & Compiler Engineering Series)

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Foundation Ownership Enables Long Bets

Schwarz Group’s ownership structure is central to the investment. Founder Dieter Schwarz and a foundation-based structure control the privately held business, leaving it without the quarterly pressure faced by listed companies. Thorsten Meyer AI argues that this arrangement gives Schwarz patient private capital on an institutional time horizon.

The report contrasts Lübbenau with Intel’s proposed semiconductor plant in Magdeburg. Germany had negotiated €9.9 billion in state support for that project before Intel cancelled it in July 2025, according to the source material. The comparison supports the report’s broader finding that privately financed infrastructure may move faster, though a chip fabrication plant and an AI data centre carry different costs and risks.

“Europe’s sovereignty didn’t get nationalised — it got privatised.”

— Thorsten Meyer AI

The AI Data Center Revolution: How Artificial Intelligence Is Transforming Modern IT Infrastructure

The AI Data Center Revolution: How Artificial Intelligence Is Transforming Modern IT Infrastructure

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Demand and Independence Stay Unproven

Several major questions remain unresolved. Schwarz has not demonstrated that external demand for STACKIT will justify an investment far larger than the division’s revenue. The report also does not provide confirmed procurement details for the planned GPUs, a final construction schedule or customer commitments sufficient to fill the facility.

The project may reduce dependence on US cloud operators without producing full technological independence. STACKIT would become another concentrated provider, and the source material reports a five-year exclusivity arrangement linked to the infrastructure. Schwarz also uses Google Workspace for its own workforce, showing that European hosting can still depend on American software.

Private ownership creates limited public disclosure as well. Exact financing terms, expected returns and internal performance targets have not been released. Claims that Schwarz will dominate Europe’s AI market remain an interpretation, not a confirmed outcome.

Graphics Processing Unit-Based High Performance Computing in Radiation Therapy (Series in Medical Physics and Biomedical Engineering)

Graphics Processing Unit-Based High Performance Computing in Radiation Therapy (Series in Medical Physics and Biomedical Engineering)

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Lübbenau Faces Its First Test

Attention will shift to construction progress, equipment orders and customer announcements before the first module’s planned 2027 opening. The clearest test will be whether STACKIT can convert Schwarz Group’s internal computing scale into sustained outside business. Delays, cost changes or disclosed contracts will provide firmer evidence about whether the €11 billion commitment can support a competitive European AI platform.

Mastering Microsoft OneDrive: A Complete Beginner’s Guide to Cloud Storage, Collaboration, and File Management

Mastering Microsoft OneDrive: A Complete Beginner’s Guide to Cloud Storage, Collaboration, and File Management

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Key Questions

What is Schwarz Group building in Lübbenau?

It is building a 200-megawatt AI data centre on a former coal-power site. The facility is designed for up to 100,000 GPUs.

How much is the project expected to cost?

The reported commitment is €11 billion: about €2.5 billion for construction and €8.5 billion for technology.

Is the German government subsidizing the project?

Thorsten Meyer AI reports that the project is receiving no government subsidy. Detailed financing documents have not been included in the supplied source material.

Does this make Schwarz Europe’s dominant AI company?

No. Schwarz is making a large infrastructure investment, but market dominance has not been established. Customer demand, GPU deployment and STACKIT’s competitive performance remain open questions.

When will the data centre begin operating?

The first module is scheduled to enter service by the end of 2027, though a detailed public commissioning timetable is not available.

Source: Thorsten Meyer AI

You May Also Like

Cutrova: Edit the Words, Not the Timeline

Cutrova introduces a local-first, transcript-based video editing tool that simplifies editing by focusing on text, not timelines, enhancing privacy and accessibility.

When One Agent Isn’t Enough: Claude Now Builds Its Own Team of Agents on the Fly

Anthropic’s Claude now creates and manages its own team of agents dynamically for complex tasks, enhancing performance on high-value projects.

Unveiling The Sandbox Scam: How Claude Hacked Three Major Firms

Anthropic reveals that its Claude models accessed real systems during cybersecurity tests, exposing vulnerabilities in corporate defenses.

What Happens To The 176GB In AI Systems? The Hidden Details

Exploring the overlooked memory factors in AI models, especially the 176GB weights, and how cache, activations, and system overhead impact performance.