Most business AI projects fail not because companies don’t have the technology, but because the models they use don’t understand their business. Models are often trained online, instead of decades of internal documents, workflows, and institutional knowledge.
This void is where Mistral, the French artificial intelligence startup, sees opportunity. On Tuesday, the company announced Mistral Forge, a platform that allows businesses to build custom models trained on their own data. Mistral announced the platform at Nvidia GTC, Nvidia’s annual technology conference, which this year is heavily focused on artificial intelligence and agent models for enterprises.
It’s a bold move for Mistral, a company that has built its business on enterprise customers, while rivals OpenAI and Anthropic have soared in consumer adoption. CEO Arthur Mensch says Mistral’s laser focus on the business is working: The company is well on its way to exceed $1 billion in annual recurring revenue this year.
A big part of doubling down on business is giving companies more control over their data and AI systems, Mistral says.
“What Forge does is it allows businesses and governments to tailor AI models for their specific needs,” Elisa Salamanca, chief product officer at Mistral, told TechCrunch.
Several companies in the business AI space already claim to offer similar capabilities, but most focus on improving existing models or layering proprietary data on top through techniques such as retrieval augmented production (RAG). These approaches do not fundamentally retrain the models. Instead, they customize or query at runtime using company data.
Mistral, by contrast, says it enables companies to train models from scratch. In theory, this could address some of the limitations of more common approaches—for example, better handling of non-English or highly domain-specific data and greater control over model behavior. It could also allow companies to train agent systems using reinforcement learning and reduce reliance on third-party model providers, avoiding risks such as model changes or deprecation.
Techcrunch event
San Francisco, California
|
13-15 October 2026
Forge customers can build their custom models using Mistral’s extensive open-weight AI model library, which includes small models like the recently introduced Mistral Small 4. According to Mistral co-founder and chief technologist Timothée Lacroix, Forge can help unlock more value from its existing models.
“The trade-offs we make when building smaller models is that they simply can’t be as good at every subject as their larger counterparts, and so their customizability allows us to choose what we emphasize and what we drop,” Lacroix said.
Mistral advises on which models and infrastructure to use, but both decisions remain with the customer, Lacroix said. And for teams that need more than guidance, Forge comes with Mistral’s team of forward-looking engineers that integrate directly with customers to surface the right data and adapt to their needs — a model borrowed from companies like IBM and Palantir.
“As a product, Forge already comes with all the tools and infrastructure so you can build synthetic data pipelines,” Salamanca said. “But understanding how to build the right one evaluates And making sure you have the right amount of data is something that businesses typically don’t have the expertise for, and that’s what FDEs bring to the table.”
Mistral has already made Forge available to partners including Ericsson, the European Space Agency, Italian consultancy Reply and Singapore’s DSO and HTX. Early adopters also include ASML, the Dutch chipmaker that led the way Mistral’s Series C last September at a valuation of 11.7 billion euros (about $13.8 billion at the time).
These partnerships are emblematic of what Mistral expects to be the primary use cases for Forge. According to Mistral’s director of revenue, Marjorie Janiewicz, these include governments that need to adapt models for their language and culture. financial players with high compliance requirements; manufacturers with customization needs; and technology companies that need to tune models in their code base.
