French AI startup Mistral AI has launched Mistral Compute, a new AI infrastructure platform aimed at providing frontier AI access to enterprises, governments, and research institutions globally.
Mistral Compute is described as a 'private, integrated stack' offering compute, orchestration, APIs, and tailored services.
The platform positions Mistral as a European alternative to US and Chinese hyperscalers, emphasizing decentralised access to high-performance AI systems.
Mistral Compute includes tens of thousands of GPUs based on NVIDIA architectures and supports diverse workloads like defence and drug discovery.
Partners backing this initiative include BNP Paribas, Orange, Thales, and Veolia, aligning with European sustainability and data sovereignty norms.
The infrastructure operates on decarbonised energy and complies with regional regulatory requirements.
Mistral aims to address GPU scarcity by offering its scaling platform to external users for custom AI environments.
They also introduced Magistral, a new reasoning-focused language model in two variants - Magistral Small and Magistral Medium.
Magistral enhances transparent, multilingual, domain-specific problem-solving with improved reasoning and traceable logic.
The launch of Mistral Compute signifies the company's expansion from open science to sovereign AI infrastructure.