Sovereignty
Sovereign, strictly defined
The word is everywhere. For graal, it has a single definition, in four points you can verify on your infrastructure. It is the same on every page, in the documentation and in meetings.
The definition
Four points, all verifiable
01
The platform and the data run on the infrastructure you choose.
02
No third-party service is required for it to work: no SaaS, no telemetry, no external fonts or captcha.
03
The vendor, which also integrates and supports the platform, is a French company (GRAAL.SYSTEMS SAS).
04
Pipeline code is exportable and formats are open (Parquet, Iceberg, SQL, MLflow API).
Why now
Sovereignty has become a buying criterion
Within a few months, major public projects have changed providers, and the European sovereign cloud market is changing scale.
23/04/2026
Health data leaves Azure
Hosted on Microsoft Azure since 2019, France’s Health Data Hub (Plateforme des données de santé) selects Scaleway after a tender launched in February 2026.
16/06/2026
The DGSI replaces Palantir with a French vendor
The French Prime Minister announces that ArgonOS, from ChapsVision, will replace Palantir on the DGSI’s OTDH project; the rollout is described as gradual over 2026 and 2027.
20/06/2025
Saagie joins the Scaleway cloud
Scaleway, part of the iliad group, acquires the assets of Saagie, a French DataOps platform founded in 2013, to build a data & AI platform in its cloud.
09/02/2026
Sovereign cloud triples in two years
Gartner estimates European spending on sovereign cloud IaaS at $6.9 billion in 2025, $12.6 billion in 2026 and $23.1 billion in 2027.
The regulatory calendar (NIS2, DORA, Data Act, AI Act, SREN law) is detailed on the Compliance page.
FAQ
What we are asked most often
Is graal SecNumCloud-qualified or HDS-certified?
Those labels qualify hosting offers, not software. graal deploys with a SecNumCloud-qualified or HDS-certified hosting provider, just as it does in your own datacenter.
Does our data leave our infrastructure?
No. The platform and the data stay with you. If you connect an external language model, it only receives what the tools it calls return: you choose the model according to the sensitivity of the data.
Are the technical building blocks French?
graal builds on open source projects of diverse origins: Kubernetes, Apache Spark, Trino, Jupyter, MLflow. No foreign cloud service is required for it to work.
What if we want to change platforms?
Pipeline code is exportable and formats are open: you leave with your work and your data.
Check the four points on your own infrastructure
The 8-week pilot installs graal on your infrastructure: every point of the definition can be observed there.