THE CRUNCH

France's technology ecosystem has moved from AI ambition to working deployments of agentic computing, according to an NVIDIA blog post published on 13 October 2026. The roundup, which is a vendor account rather than independent reporting, spans space, financial services, the public sector, retail and scientific research, with open models and NVIDIA infrastructure as the common thread.

The most striking example is in orbit: Altair-1, the first satellite in a planned $1 billion, 50-satellite constellation, launched this month and reached orbit successfully. It flies NVIDIA Jetson AGX Orin edge modules running Mistral's models for real-time Earth observation and autonomous operations, and was integrated by Loft Orbital and built in Abu Dhabi by Orbitworks, a joint venture between Loft Orbital and Marlan Space.

On the ground, sovereign AI is the other pillar. LINAGORA and Docaposte have secured the CANUT framework agreement, making Twake.ai, built on NVIDIA Nemotron, the sole referenced sovereign digital workplace for more than 2,500 French public organisations. In banking, SBS is launching a platform with NVIDIA to let banks run governed generative and agentic services inside their own environments. Mistral also introduced its Mistral Large 4 open model under private preview last week.

Elsewhere, L'Oréal is using NVIDIA ALCHEMI in research workflows, with the company claiming computational predictions combined with lab work can make discovery up to 100x faster than traditional methods. Mirakl has onboarded more than 70 million products with models on NVIDIA GPUs, cutting seller catalog onboarding to under two hours, while ThinkDeep and Akila built a digital-twin agent managing nearly 5,800 kilometres of roads and buildings in Côte-d'Or, cutting operating costs by 10% so far.

Hardware plans are also in motion: Bull is extending its European AI server production to NVIDIA Vera Rubin Ultra NVL72 systems planned for 2028, with rack integration at its newly inaugurated Angers factory. In quantum computing, Alice & Bob and Quandela have both integrated NVIDIA's NVQLink platform to connect GPUs with their respective superconducting and photonic qubits.

WHAT HAPPENS NEXT

Bull's Vera Rubin Ultra NVL72 server production is planned for 2028, and Fondation Inria's working group on open source modelling, launched with NVIDIA's support, is due to publish findings in spring 2027.