Sovereignty

Built for sovereignty

Two ways to keep control of your data: our managed service, run in the EU by a European company, or LakehouseBox on infrastructure you control, air-gapped if you need it.

Either way your tables are open Apache Iceberg, and you know exactly where they run and who handles them.

Get started Explore self-hosting

Managed service: your data path

Every step of the data path is in the European Union.

SpainLakehousebox, S.L.
Company & operationsEU operator

GermanyData storage & compute
Frankfurt; a replicated copy in Stockholm, SwedenEU infrastructure

FranceHosted-agent models (preview), emailEU provider

Or self-host it, even air-gapped

Data path locationEuropean Union (EU)

See details below

Two ways to run LakehouseBox

Option 1 · Managed service

Run by us, in the EU

Operated by Lakehousebox, S.L. from Spain
Data and backups in Frankfurt, Germany, replicated to Stockholm, Sweden
Nothing to install; free to start, 5 GB, no card

Option 2 · Self-hosted

Run by you, where you choose

Your cloud, your data centre, your jurisdiction
Air-gapped if you need it: no connection to us or to the internet
The same containers as the managed service

Option 1 · Managed service

The managed service, step by step

Who operates it, where your data is stored and processed, and what we still depend on.

The operator, the data path and the dependencies

European operator

Lakehousebox, S.L. is a Spanish company, operating under EU law and the GDPR.

EU data path

On the managed service, your tables are stored and processed within the European Union.

Transparent dependency ledger

We list every third-party service, and separate the data path from the services around it.

What is in the path

The components that handle your data, from storage to results.

Component	Provider	Location	Purpose	Handles your data?
LakehouseBoxOperator, support	Lakehousebox, S.L.	Spain	Operates the service	Yes
Data storage & computeCatalog, object store, server-side queries	Our EU infrastructure	Germany, Sweden	Stores your tables and files (in Frankfurt, Germany, with a replicated copy in Stockholm, Sweden) and runs the catalog, queries, scheduled SQL and hosted agents in Germany	Yes
BackupsOff-host copies	Our EU infrastructure	Germany, Sweden	Encrypted copies of the catalog's state every 5 minutes, in Frankfurt, Germany, replicated to Stockholm, Sweden	Yes
Model servicesHosted agents (preview)	Scaleway	France	Model inference for hosted agents, unless you bring your own model key	Yes
Transactional emailAccount mail	Scaleway	France	Sign-in codes, invitations, notices	Your address
Content deliveryCatalogs you publish or cache	Our CDN, an EU provider	EU, edge worldwide	Caches and serves the files of catalogs you make public, and of private catalogs you turn edge caching on for	Only what you choose

Processing also happens wherever your own agent or query engine runs. The catalog's state is copied off-host every 5 minutes; the data files are replicated to Stockholm as they are written; one serving site today, no standby yet (availability & recovery).

What we still depend on

Services outside the EU that never hold your data, but that we use to build and run the service. We list them so you can judge for yourself.

GitHubSource code hosting (no customer data)Outside the EU
Docker HubOne gateway image (no customer data)Outside the EU
Let's EncryptTLS certificates (sees hostnames only)Outside the EU
DuckDB extensionsInstalled on your machine by your engineYour side

These services are outside the data path and do not process or store your data. See the full dependency ledger.

Option 2 · Self-hosted

Run it on infrastructure you control.

The same containers the managed service runs, on one Linux host with Docker: in your cloud, in your own data centre, or cut off from the internet entirely. You choose the provider, the location and the law it runs under.

Self-hosting details

Your cloud
Your data centre
Air-gapped

No call home. Nothing in the containers calls a cloud API or our service; there is no licence server.
Without internet. The images load onto the host without a registry, the query engine carries its own extensions, and mail and an external sign-in are optional. Certificates come from your own authority instead of Let's Encrypt; we set that up with you.
Your copies. Backups go to the S3-compatible store you choose, inside your own network.

Beta: one node per install; licence and support terms are being decided. Talk to us.

Compliance and provenance

Built on open data, sources you can verify, and European rules.

GDPR: our DPA
On the managed service, a data processing agreement (draft) for every customer; every sub-processor is listed in the Privacy Policy.
Open data foundation
Public datasets from open sources such as Overture Maps, the European Environment Agency and the World Bank.
Transparent lineage
Every public dataset names its source, licence and attribution, so results can be traced and checked.
Ready for SEAL-4
Built to the EU Cloud Sovereignty Framework. Open standards throughout.

European ownership
and jurisdiction.

The CLOUD Act can reach data held by providers subject to US jurisdiction, even when stored in Europe.

Our approach is European operation and infrastructure, without a US cloud provider holding your hosted tables.

Hosting location alone does not determine legal exposure. Exposure follows the operator, its providers and the law that binds them.

Read our legal and security details

European peopleEuropean infrastructureEuropean control

Ready for SEAL-4.

Built to the European Commission's Cloud Sovereignty Framework, the assessment public buyers use for cloud services.

View the framework

Cloud Sovereignty Framework

SEAL-4

Full Digital Sovereignty

Evidence and assessment are required across the framework's eight objectives. SEAL levels are assigned by a contracting authority in a tender; nobody has assessed us.

What we're improving.

We're continuing to reduce external dependencies.

Source code hosting	Under review	Evaluating European hosting options.
TLS certificates	Under review	Evaluating alternatives for certificate issuance.

These services do not store your analytical tables. Certificate issuance remains necessary for HTTPS availability.

The full list of what we depend on outside the EU, checked 2026-09-22

None of these parties holds customer data or sits in the data path. Each entry says what an outage or a withdrawal could affect. We re-check this list and change the date when we do.

Dependency	Who	What it can affect	What it cannot affect
Source code hosting	GitHub (Microsoft, US)	Where we collaborate on the code and where CI runs. Every checkout is a full copy of the repository.	The running service, your data.
Container images	Docker Hub (US) for the reverse proxy and base images our image is built from; the storage and catalog part is our own build, from a European registry (Scaleway, Paris); the finished image is pulled from GitHub's container registry (Microsoft, US; updated 2026-10-30)	Pulling images for a new deployment. Images are pinned by digest; a registry outage delays a deploy, not the service already running.	The running service, your data.
TLS certificates	Let's Encrypt (ISRG, US)	Certificate renewal for our hostnames. A withdrawal would mean changing the certificate authority.	Your data; the certificate authority never sees traffic.
Query-engine extensions on your side	DuckDB's extension repository, on your machine when you run `INSTALL iceberg`	Installing the engine where your agent runs. Our browser panel serves DuckDB and its extensions from lakehousebox.com instead.	Your stored tables; any Iceberg reader can read them.
Sign in with Google (optional; listed 2026-10-16)	Google (US), chosen by the person who presses “Continue with Google”	Google learns that this person signs in to LakehouseBox and sends us their name and verified address. If Google is unavailable, that button does not work; the email sign-in does.	Your data: no catalog data, key or query goes to Google, and everything works without it.

Mirrors for the source code and the container images inside the EU are planned; there is no date. When they exist, this list will say so with the date of the check.

Take your data with you.

Apache Iceberg tables and Parquet files. Use supported engines and retain control of your data.

Apache Iceberg
tables
Parquet
files

Choose where your data lives.

Start free Discuss self-hosting

5 GB free, no card · Security overview · Data processing agreement · Terms (Spanish law)