Case study ·
TGV hops on the Data Mesh train
Original title, in French: “TGV prend le train du Data Mesh”
- Publication
- LeMagIT (France)
- Author
- Christophe Auffray
- Published on
- 5 October 2026 · read in the page structured data
Summary
SNCF's TGV division has adopted a Data Mesh architecture to break down data silos across Ouigo, Eurostar and Lyria, as rail competition opens up. According to chief data officer Julien Iris, this transformation aims to lay the foundations for AI, along three lines: "AI for everyone" (SNCF-GPT, Copilot), "AI for digital production" and "AI for the business lines" (train set diagnostics, responses to tenders). The company stresses the need to document processes before any AI optimisation. FinOps practices with smart routing between Claude models make it possible to cut token costs by up to 70%. The subsidiary is also exploring an "agent marketplace" to counter Shadow AI, inspired by the example of LVMH.
Summary based on the full text of the article.
Read the article on LeMagIT →Figures cited
-
70 %
Reduction in LLM token costs through smart routing
SNCF's TGV division
« Il permet de réduire jusqu’à 70 % les coûts, par rapport à une utilisation non arbitrée, chiffre le CDO de l’entreprise. »
Classification
- Country
- France
- Company size
- Large companies
- Industries
- Transport, logistics, supply chain
- Companies
- DataGalaxy (supplier), LVMH (cited), AWS (supplier), SNCF (cited), SNCF TGV (deploys AI), Databricks (supplier), Microsoft (supplier), Anthropic (supplier)
- Keywords
- data mesh, data governance, generative AI, finops, shadow ai, rail transport
Entry written by Marsanne Studio from the cited article, with the help of an AI model (claude-sonnet-5). The rights to the article and its illustrations belong to LeMagIT and its author. Entry translated from French by Claude Sonnet 5.5.