# Mistral Compute [[Mistral AI]]'s AI infrastructure offering, announced in 2026. Provides customers a private, integrated stack: GPUs, orchestration, APIs, products, and services in form factors ranging from bare-metal servers to fully-managed PaaS. Positioned as a European alternative to US and China-based cloud/AI providers. ## What it provides - Full AI infrastructure stack: compute (GPUs), orchestration, APIs, and Mistral's product suite - Mistral AI's training suite for domain-specific and region-specific model training - Tens of thousands of GPUs using latest NVIDIA reference architectures, with rapid expansion planned - Deployment options from bare-metal to fully-managed platform-as-a-service ## Strategic positioning - Designed for sovereigns, enterprises, and research institutions that want to own their AI stack top to bottom - Targets Europe, Middle East, Asia, and the Southern Hemisphere as an alternative to US/Chinese cloud providers - Also serves global entities of US/Chinese companies needing regionally-operated AI (particularly in Europe) - Strong emphasis on data sovereignty and European regulatory compliance - Commitment to sustainability through decarbonized energy ## Origin Born from Mistral AI's own experience building infrastructure for model training and serving. The internal platform that trained their state-of-the-art models is now being made available externally. ## Launch partners BNP Paribas, Black Forest Labs, Kyutai, Mirakl, Orange, Schneider Electric, SLB Groupe, SNCF, Thales, and Veolia. ## References - https://mistral.ai/news/mistral-compute ## Related - [[Mistral AI]] - [[Large Language Models (LLMs)]] - [[Mistral Small 4]]