Microsoft and the French AI developer Mistral are expanding their strategic partnership. In addition to a multi-billion investment in European AI infrastructure, the companies will make Mistral’s AI models more widely available through Microsoft Foundry, Copilot Studio, and Azure. The partnership is primarily aimed at organizations that want to deploy AI while adhering to strict requirements regarding data location, security, and regulations.
A key component of the agreement is the expansion of AI computing capacity in Europe. To this end, Mistral is investing in a new GPU infrastructure that utilizes thousands of NVIDIA Vera Rubin GPUs. Microsoft will gain access to a portion of this capacity to meet the growing demand for AI and cloud services. According to both companies, the agreement is worth several billion dollars. Azure customers will be able to utilize AI capacity housed in Mistral’s French data centers.
The partnership aligns with Microsoft’s broader strategy to expand its European AI and cloud infrastructure. As part of this, the company is combining its own data centers with leased capacity and collaborations with external infrastructure partners.
On the software side, the Mistral Medium 3.5 and OCR 4 models will be integrated into Microsoft Foundry, the development platform for AI models and AI agents. Medium 3.5 will also be available in Copilot Studio, which enables organizations to develop their own AI assistants.
Microsoft is explicitly positioning this collaboration for organizations that want more choice in the AI models they use. Developers can continue using the same development environment while choosing from different models for various applications. OCR 4 focuses on document processing and automated workflows, while Medium 3.5 is a general-purpose language model.
European AI sovereignty
The announcement fits into the broader discussion about digital sovereignty in Europe. European governments and companies are trying to reduce their dependence on U.S. AI infrastructure and models without sacrificing access to the latest AI technology. According to SiliconANGLE, this discussion has intensified in recent weeks following the United States’ restrictions on access to certain advanced AI models for some European allies, citing security concerns.
Microsoft President Brad Smith told Reuters that the combination of Mistral’s infrastructure and Azure Local is intended to ensure that European organizations retain access to advanced AI while allowing them to manage their data and workloads locally if they choose.
A single platform for cloud and on-premises infrastructure
A key part of the announcement is the ability to run the same AI applications across different environments. Organizations can use Mistral models in Azure, on Azure Local, or in completely offline installations.
Organizations in regulated sectors, in particular, are requesting such deployment options. Sectors such as financial services, healthcare, manufacturing, and critical infrastructure often have requirements related to data sovereignty, privacy, export controls, or operational continuity. By using the same development environment and management processes, organizations do not have to redevelop their AI applications for each environment.
The partnership aligns with Microsoft’s previous commitments to make more AI capacity available in Europe. At the same time, it strengthens Mistral’s position as a European provider of advanced AI models. The French company has long positioned itself as a European alternative to U.S. AI providers.
According to SiliconANGLE, Mistral CEO Arthur Mensch confirmed that the agreement does not involve an increase in Microsoft’s equity stake in the company. However, the partnership is expected to contribute to Mistral’s goal of having approximately 1 gigawatt of AI computing capacity operational by 2030.
In addition to the technical collaboration, Microsoft and Mistral also aim to jointly support customers in adopting AI. This will be achieved through proof-of-concepts, workshops, and support for implementation projects, among other things. As a result, the agreement provides not only a technological but also a practical framework for organizations seeking to deploy AI within their own infrastructure.