Koyeb is Joining Mistral AI to Build the Future of AI Infrastructure

Koyeb, a high-performance serverless infrastructure for intensive workloads, led by Yann Léger, Edouard Bonlieu, Bastien Chatelard, and the team, has entered into a definitive agreement to join Mistral AI to advance cutting-edge AI infrastructure.

Koyeb will bring its platform, technology, and team to accelerate Mistral Compute offering. Compute is designed to provide leading teams across the globe with the same state-of-the-art infrastructure Mistral AI uses to build, run, and scale frontier models and AI software.

Founded in 2021, Koyeb set out to build next-gen cloud infrastructure, a global, high-performance serverless platform deployable in seconds. As generative AI and agentic workloads scaled rapidly, it expanded to serverless GPUs, specialized accelerators, and autoscaling environments that support APIs, agents, and high-performance inference without infrastructure management.

Today, the platform runs tens of thousands of apps on bare-metal servers across 10 global locations, optimized for performance and cost efficiency. Koyeb is now joining Mistral AI to help build the future of AI infrastructure.

Koyeb chose to join Mistral AI after aligning on a shared goal: accelerating world-class AI infrastructure in Europe.

Mistral is launching Mistral Compute with massive scale, starting at 40 MW of data-center capacity and ~18,000 Nvidia Blackwell GPUs, alongside a planned $1.4B data-center expansion in Sweden. Beyond building frontier models, the company is now vertically integrating the infrastructure layer needed to run them at a global scale.

Discussions with co-founder & CTO Timothée Lacroix revealed a common mission: make high-performance AI infrastructure widely accessible, not just to hyperscalers but to all developers and enterprises.

By joining Mistral AI, Koyeb can accelerate that vision, combining its serverless platform with large-scale compute capacity to help deliver state-of-the-art AI infrastructure to everyone.

What this means for customers and for Koyeb

The platform will continue running as usual, but over the coming months, it will become a core component of Mistral Compute within Mistral AI. Development will concentrate on inference, sandboxes, and serverless capabilities for MCP servers.

As part of the transition:

• Focus shifts to Pro plans and above; new sign-ups will only be available on paid tiers.

• Existing organizations keep their current plans unchanged.

• The free tier is being phased out to prioritize performance and reliability.

Koyeb will remain a serverless platform for deploying production-grade, AI-intensive applications on high-performance infrastructure.

Completion of the acquisition remains subject to closing conditions.