Data sovereignty

Ready for AI, without your data leaving Europe

A sovereign data platform with Qlik, Snowflake and STACKIT, with language models in Germany. You hold the keys; cimt builds the platform and runs it for you.

Discuss your situation Download the overview (PDF)

An open vault door in a concrete room, with an orange key on the ledge

What is data sovereignty?

Data sovereignty means your organisation decides where data is stored, who can access it and which law it falls under. For a data platform that comes down to three things you can demonstrate: data and AI processing stay in Europe, you hold the keys, and you can leave without your data being locked into one vendor.

The question comes up more often now. AI applications work with order data, customer data and contracts. The EU Data Act, applicable since 12 September 2025, gives you the right to switch cloud providers. And customers, regulators and your own DPO ask more and more precisely where data is processed. If you cannot show that, AI plans stall. See also how we get your data ready for AI.

Use cases

Four situations where the platform pays off straight away

Data sovereignty is rarely a project in itself. Usually something is already going on: a system under heavy load, SAP data that is locked in, a database that needs replacing or an AI plan. That is the moment to set up the platform as sovereign from the start.

Wholesale and distribution

IBM i (AS/400)

Take the load off the core system

Situation: A wholesaler runs orders, inventory and invoicing on IBM i (AS/400). Every report puts load on that system, and every new question needs scarce expertise.
Approach: Qlik Replicate reads changes from the DB2 database (change data capture) and loads them into the platform in near real time. The tables sit in Apache Iceberg on your STACKIT storage. Reports and AI run on that copy from then on.
Result: The core system does what it does best. The data is available for analysis and AI, and stays in Europe.

Unlock SAP data

Situation: A manufacturer runs SAP ECC and is preparing the move to S/4HANA; mainstream maintenance for ECC ends at the end of 2027. Reports are locked inside SAP and every migration step affects them.
Approach: Qlik Replicate loads SAP data continuously into the platform; the tables sit in Iceberg on STACKIT. In Snowflake we build a bronze, silver and gold data model that is independent of the SAP version.
Result: Reports and AI run on a stable model outside SAP. The move to S/4HANA changes the source, not the reports.

Migrate Oracle databases

Situation: A logistics provider has an Oracle data warehouse that is due for replacement: licence and maintenance costs keep rising and the hardware is written off.
Approach: Qlik Replicate keeps Oracle and the new platform in sync during the migration. We move the data into Iceberg on STACKIT and rebuild the models in Snowflake, subject by subject.
Result: No big bang: a report moves over only when the figures match. Then Oracle is switched off.

AI on your own data

Situation: A technical wholesaler in the Netherlands and Belgium wants an AI assistant that answers questions about orders and delivery times. The DPO first wants to know where the data and the language model live.
Approach: cimt puts the order data in Iceberg on STACKIT, loads SAP data with Qlik Replicate and connects Qlik Sense to a STACKIT language model through the Qlik MCP server.
Result: The DPO sees in one overview where the data lives and who holds the key. The assistant goes live without data leaving Europe.

Qlik, Snowflake and STACKIT as one sovereign platform

Proven software for integration, analytics and storage, on a European foundation. cimt connects the layers and takes over operations.

Integration, analytics and AI

Qlik

Client-managed on your own servers, for example on STACKIT. Qlik Replicate and Talend Data Fabric load data from SAP and other sources; Qlik Sense Enterprise makes it usable. From the November 2026 release the Qlik MCP server is available, so AI assistants work directly with your Qlik apps.

More about Qlik

Storage and processing

Snowflake

The central data platform in bronze, silver and gold: from raw source data to reliable data for analytics and AI. With Snowflake Horizon you define per role who sees which columns and rows, and you can see afterwards who queried what. We put the data in Apache Iceberg on your STACKIT storage.

More about Snowflake

The connecting layer

cimt

We design the architecture, set up governance based on DAMA DMBoK and build the platform on STACKIT. After that our Managed Services take the work off your hands: monitoring, operations and upgrades, as much as you need. You stay in control.

The European foundation

STACKIT

The cloud of the Schwarz Group (Lidl, Kaufland), with data centres in Germany. This is where your data and Qlik servers live, and where your language models run through STACKIT AI Model Serving, without your data being stored or used for training. Snowflake and STACKIT announced in September 2026 that they will integrate their platforms step by step.

Design choices

Four choices we fix for you

Sovereignty sits in four choices in the architecture. We record them so you can demonstrate them later.

Location

Data on STACKIT in Germany, Qlik on your own servers, Snowflake in an EU region. Fixed in contract and configuration.

Your own control

You manage storage and keys on STACKIT. Revoke them and Snowflake can no longer read your Iceberg tables.

AI inside the border

Language models run on STACKIT in Germany. AI assistants work through the Qlik MCP server with your Qlik apps and a model on STACKIT.

Exit path

Open Apache Iceberg format and a documented data model. Your data stays readable without Snowflake, in line with the Data Act.

To be clear

American software, European control

Snowflake and Qlik are American companies. We address that in the architecture: your data and language models sit on STACKIT, Qlik runs client-managed on STACKIT or in your own data centre, and Snowflake reads your tables inside the EU, only for as long as you allow it. cimt handles setup and operations.

How we work

From discovery to operations

You decide after each step whether to continue. A sensible start is discovery with one source and one use case.

Step 1

Discovery

Together we map where your data and AI processing sit today, which requirements apply (GDPR, Data Act, sector rules, customer contracts) and which systems come first.
Step 2

Architecture

We fix the four design choices: location, key management, AI processing and exit path. You get an architecture your DPO and auditor can follow.
Step 3

Build

We build the platform on STACKIT, load the first sources and set up governance based on DAMA DMBoK: ownership, access rights, data quality and lineage.
Step 4

Operate

After go-live our Managed Services take the work off your hands: monitoring, operations and upgrades, as much as you need. You stay in control.

Qlik Elite Partner since 2009
Snowflake partner since 2018
Knowledge partner of DAMA NL
300+ data specialists across Europe

Free download

Sovereign data platform: Qlik, Snowflake and STACKIT (PDF)

Two pages on a data platform where data and AI processing stay in Europe. How Qlik, Snowflake and STACKIT work together, the four design choices we fix for you and four use cases from logistics, manufacturing and wholesale.

· The platform at a glance: Qlik, Snowflake, STACKIT and cimt
· Four design choices: location, control, AI inside the border, exit path
· Use cases: IBM i (AS/400), SAP, Oracle and AI on your own data

2 pages, PDF

Your contacts

Talk to the people who design the platform

A focused half-hour conversation about where your data and AI processing sit today. Afterwards you know which steps it takes to keep that demonstrably in Europe.

Book a conversation

Taco van het Reve

Managing Director
Henk Teurlings

Solution Architect

Discuss your situation

Find out how sovereign your data is today

Together we map where your data and AI processing sit today and what it takes to keep that demonstrably inside the EU.

Book a conversation Download the overview (PDF)

Frequently asked questions

Data sovereignty: what you want to know

Can a data platform built on American software be sovereign?

Yes, if the architecture takes care of it. Your data sits in Apache Iceberg on STACKIT storage in Germany, you manage the keys and Qlik runs client-managed on your own servers. Snowflake reads your tables in an EU region and only for as long as you allow it. Revoke the keys and Snowflake can no longer read your Iceberg files on STACKIT.

What is STACKIT?

STACKIT is the cloud of the Schwarz Group, the company behind Lidl and Kaufland. Its data centres are in Germany and the organisation falls under European law. In this platform your data, your Qlik servers and your language models run on STACKIT.

Can Qlik run in our own data centre?

Yes. We work with Qlik’s client-managed products: Qlik Replicate, Talend Data Fabric and Qlik Sense Enterprise. They run on STACKIT or in your own data centre, whichever fits your policy best.

How do we use AI without our data leaving Europe?

Language models run through STACKIT AI Model Serving in Germany, without your data being stored or used for training. With the Qlik MCP server, available from the November 2026 release, AI assistants work with your Qlik apps and a model on STACKIT.

What if we want to move away from Snowflake later?

Your data sits in the open Apache Iceberg format on your own STACKIT storage, with a documented data model. Other engines can read those tables without Snowflake. That fits the EU Data Act, which is meant to make switching between cloud providers easier.

Do we have to replace our current data platform?

Not in one go. During discovery we decide which data and which AI applications need to become sovereign first. A sensible start is one source and one use case; the rest follows step by step, without interrupting running reports.

What does a sovereign data platform cost?

That depends on the number of sources, the data volumes and how much of the operations you hand over to us. After discovery you receive a reasoned estimate for build, licences and operations, so you can decide on facts.