French AI darling Mistral is expanding into AI infrastructure services as it unveils a new offering for a compute platform in partnership with Nvidia.
The announcement, which was made at French tech conference Viva Technology in Paris on Wednesday, means Mistral is set to tackle a new part of the AI value chain. The two-year-old startup originally launched to build generative AI models similar to those powering OpenAI’s ChatGPT.
Since launching, Mistral has raised over $1bn and was last valued at €6bn, with backers including DST Global and General Catalyst — and is considered one of Europe’s most promising companies in the sector.
Through the platform, dubbed Mistral Compute, it intends to provide hardware and software resources for users to develop AI systems. They will be able to use Mistral’s own models as well as other providers’ open-source models.
Mistral Compute will be supported by 18k Nvidia GPUs specialised for AI, called Grace Blackwell Superchips, and full deployment is expected in 2026.
“We’re still doing models,” said Mistral CEO Arthur Mensch, speaking at VivaTech on Wednesday. “On top of that we’re going to be operating all of our software platform on digital assets that we’re deploying with Nvidia.”
“We’re expanding from an AI company doing software to a cloud company.”
Mensch added this will enable developers to build AI workloads without “relying on certain US providers.”
Mistral Compute
In addition to Nvidia’s GPUs, Mistral Compute will provide users with software and services to code, develop models and build AI chat. Targeting enterprise customers and the public sector, it will be available as a cloud service as well as for private deployment on-premises.
Speaking at VivaTech, Nvidia founder Jensen Huang said: “We’re going to be partnering to build a very sizable AI cloud here.”
Mistral did not share the cost of launching the new service, but it is likely to be a significant investment, with Nvidia’s AI superchips estimated to cost up to tens of thousands of dollars each.
The French startup is reported to be considering raising $1bn in funding this year.
Mistral says the service has already been adopted by launch partners including German generative AI startup Black Forest Labs, French bank BNP Paribas and defence corporation Thales.
Nvidia’s European push
The news came as Nvidia announced a wealth of projects related to building AI infrastructure across Europe, including working with large corporations like Germany’s Siemens and France’s Schneider Electric, and building data centres across the region.
Huang said during the event that Nvidia will increase the amount of AI computing capacity in Europe “by a factor of ten” over the next two years.
Earlier this week, speaking at London Tech Week, Huang also said the UK needs to build its own AI infrastructure.
“You need to have some ability to develop your own intelligence,” Huang reiterated at VivaTech on Wednesday.
Update: This article has been edited on 11 June to add content from Huang and Mensch’s discussions at VivaTech.
Read the orginal article: https://sifted.eu/articles/mistral-compute-launch/
