Most enterprise AI projects fail not because the company lacks the technology, but because the models they use do not understand the business. Models are often trained from the Internet rather than decades of internal documents, workflows, and institutional knowledge.
French AI startup Mistral sees this gap as an opportunity. On Tuesday, the company announced Mistral Forge, a platform that allows companies to build custom models trained on their own data. Mistral announced its platform this year at Nvidia GTC, Nvidia’s annual technology conference focused on enterprise AI and agent models.
This is an important move for Mistral, a company that has built its business around enterprise customers while its competitors OpenAI and Anthropic are surging in terms of consumer adoption. CEO Arthur Mensch says Mistral’s laser focus on the business is paying off. The company is on track to surpass $1 billion in annual recurring revenue this year.
Mistral says a big part of doubling the size of the company is giving the company more control over its data and AI systems.
“Forge’s role is to enable enterprises and governments to tailor AI models to their specific needs,” Elisa Salamanca, head of product at Mistral, told TechCrunch.
Several companies in the enterprise AI space already claim to offer similar capabilities, but most focus on fine-tuning existing models or layering proprietary data on top through techniques such as search augmented generation (RAG). This approach does not fundamentally retrain the model. Instead, it uses corporate data to reconcile or query it at runtime.
In contrast, Mistral says it helps companies train their models from scratch. In theory, this could address some of the limitations of more general approaches. For example, you can better handle non-English or domain-specific data and have greater control over model behavior. Enterprises can also use reinforcement learning to train agent systems and reduce reliance on third-party model providers, avoiding risks such as model changes or deprecation.
Tech Crunch Event
San Francisco, California
|
October 13-15, 2026
Forge customers can build custom models using Mistral’s extensive open AI model library, which includes small models such as the recently released Mistral Small 4. Forge can help drive more value from existing models, according to Timothée Lacroix, co-founder and chief technologist at Mistral.
“The trade-off in building a smaller model is that it can’t be as good as a larger model for every topic, so customization allows us to choose what we emphasize and what we omit,” Lacroix said.
Mistral advises which model and infrastructure to use, but both decisions are up to the customer, Lacroix said. And for teams who need more than guidance, Forge comes with Mistral’s team of forward-deployed engineers who embed directly with customers to surface the right data and adapt to their needs. This model is borrowed from companies such as IBM and Palantir.
“As a product, Forge already comes with all the tools and infrastructure to create synthetic data pipelines,” Salamanca said. “But understanding how to build the right assessments and making sure they have the right amount of data is something companies typically don’t have the appropriate expertise for, and FDE brings that to the table.”
Mistral has already made Forge available to partners including Ericsson, the European Space Agency, Italian consulting firm Reply, and Singapore’s DSO and HTX. Early adopters also include Dutch chipmaker ASML, which led Mistral’s Series C round last September at a valuation of 11.7 billion euros (about $13.8 billion at the time).
This partnership symbolizes what Mistral anticipates will be key use cases for Forge. This includes governments, which need to tailor models to their language and culture, according to Marjorie Janiewicz, Mistral’s chief revenue officer. Financial companies with high compliance requirements; Manufacturers needing customization; For example, technology companies that need to adapt their models to their code base.
