SNCF’s Definition of Strategic Autonomy

▶ Watch (0:05)

SNCF operates the second-largest rail network in the world. Security sits at the center of everything: reliability, safety, cybersecurity, and what the company calls strategic autonomy. That term means reducing dependence on external vendors while keeping internal teams in control of their own infrastructure. SNCF chose “strategic autonomy” over “sovereignty” to signal a preference for deliberate action rather than reactive lock-in avoidance. The goal is giving every business unit genuine choice over where and how their applications run.

The 30/40 Split: Data Centers and Cloud-Native Adoption

▶ Watch (0:42)

SNCF started a cloud migration program in 2018 and introduced systematic application classification to sort which apps move to the cloud. Today 30% of applications stay in SNCF data centers. Of the rest, 40% use cloud-native technology. Eight years into the program, demand for cloud-native infrastructure keeps growing. The company is now working with what Comtet called the late majority of internal teams, meaning the harder-to-convert groups, not the early adopters who signed up first.

Why the First On-Premise Clusters Failed

▶ Watch (1:36)

Six years ago SNCF built on-premise Kubernetes clusters as a practical alternative to public cloud managed services. The clusters proved operationally stable, but they never matched the feature velocity of public cloud offerings. Automation was the specific failure point. The gap between what the clusters could do and what teams expected from a managed service was too wide. That gap forced the decision to start over, rebuilding the platform layer by layer rather than patching the original approach.

Rebuilding Layer by Layer with Open Infrastructure

▶ Watch (2:07)

SNCF rebuilt everything from scratch, assembling internal foundations layer upon layer. Each layer was designed to expand the capabilities of the platform rather than replace the one below it. The stack relies on open infrastructure and what Rotilio described as innovative technology, chosen specifically to avoid tying the platform to a single vendor’s roadmap. The result is a self-sufficient alternative to public cloud managed services that all SNCF business units can use to maintain strategic autonomy over their own workloads.

Notable Quotes

we decided to rebuild everything from scratch purposefully assembling layer upon layer of internal foundations as to expand the capabilities of our platform and finally reach our goal Yann Rotilio · ▶ 2:07

this world alternative made thanks to open infrastructure and innovative technology all business to Ensure they have strategic autonomy Yann Rotilio · ▶ 2:22

Key Takeaways

  • SNCF keeps 30% of applications in its own data centers after six years of migration work.
  • The first on-premise Kubernetes clusters were stable but failed on automation, forcing a full rebuild.
  • Layered open infrastructure now gives every SNCF business unit genuine strategic autonomy from public cloud vendors.

About the Speakers

Thomas Comtet has worked at SNCF for 20 years, starting on embedded systems in TGV high-speed trains before moving into cloud infrastructure. He has worked with containers and Kubernetes since 2018 and heads SNCF’s container and cloud-native platforms.

Yann Rotilio has been in IT since 2007, moving from Java development down to infrastructure automation through DevOps practice. He joined Kubernetes in 2016 and serves as SNCF’s Kubernetes specialist, focused on building better-operated software across the organization.