Mistral AI Launches Forge Platform to Challenge Cloud Giants with Enterprise Model Training

Quick Facts

• Mistral AI launched Forge on March 17, allowing enterprises to train custom AI models with proprietary data on their own infrastructure
• The French AI company is on track to surpass $1 billion in annual recurring revenue this year, growing from $20 million to over $400 million in 12 months
• More than 100 enterprise customers including ASML, TotalEnergies, and European governments use Mistral’s services, with 60% of revenue from Europe

Mistral AI launched Forge on Monday, a new enterprise platform that lets companies build and train custom AI models using their proprietary data. The announcement came at Nvidia’s GTC conference as the French AI startup positions itself as a European alternative to US cloud giants.

The platform addresses a key limitation of existing AI services. Most enterprise AI solutions rely on fine-tuning existing models or using retrieval augmented generation techniques. Forge enables companies to train models from scratch with their own data while keeping everything on their infrastructure.

“What Forge does is it lets enterprises and governments customize AI models for their specific needs,” said Elisa Salamanca, Mistral’s head of product. “It’s on their clusters, it’s with their data — we don’t see anything of it, and so it’s completely under their control.”

The business momentum is significant. CEO Arthur Mensch said the company will surpass $1 billion in annual recurring revenue this year. Mistral’s revenue run rate jumped from $20 million to over $400 million within 12 months, according to the Financial Times.

The company raised €1.7 billion in September 2025 at an €11.7 billion valuation, with Dutch chipmaker ASML leading the round. Mistral now serves over 100 enterprise customers including ASML, TotalEnergies, HSBC, and several European governments.

Forge includes technical features like support for dense and mixture-of-experts architectures, multimodal capabilities for text and images, and agent-first design. The platform allows teams to customize models using plain English commands with built-in evaluation pipelines.

Use cases range from hedge funds customizing models for proprietary quantitative languages to manufacturers with specialized code bases. Mistral worked with Ericsson to customize its Codestral model for legacy-to-modern code translation involving internal calling languages no off-the-shelf model had encountered.

“When you want to go a step beyond that, you actually need to create your own models,” Salamanca said. “You need to leverage your proprietary information.”

The launch caps an aggressive week for Mistral, which also released its Mistral Small 4 model and unveiled Leanstral, an open-source code agent. The company is investing €1.2 billion in AI data centers in Sweden, its first facility outside France.

Mistral’s growth reflects broader European concerns about technological dependence on US providers. The EU sources more than 80% of its digital services from foreign providers, mostly American companies. Around 60% of Mistral’s revenue comes from Europe.

Read more: Mistral AI launches Forge to help companies build proprietary AI models, challenging cloud giants