AINEWS
Mistral Bets on 'Build-Your-Own AI' as It Takes On OpenAI, Anthropic in the Enterprise
French AI lab launches Forge platform, enabling companies to train custom models from scratch on proprietary data.
French AI lab launches Forge platform, enabling companies to train custom models from scratch on proprietary data.
Mistral has unveiled Mistral Forge, a platform that lets enterprises build custom AI models trained from scratch on their own data—a pointed differentiation from rivals OpenAI and Anthropic, which focus on API-based services and fine-tuning.
The company announced Forge at Nvidia's GTC conference, positioning it as an alternative for enterprises that want full control over their AI rather than relying on third-party models.
"Most enterprise AI projects fail not because companies lack the technology, but because the models they're using don't understand their business," said CEO Arthur Mensch. "The models are often trained on the internet, rather than decades of internal documents, workflows, and institutional knowledge."
What makes Forge different: Most enterprise AI solutions offer fine-tuning or retrieval-augmented generation (RAG)—adapting existing models or layering company data on top at runtime. Mistral says Forge enables companies to train models from scratch using their proprietary data, which could better handle non-English languages, domain-specific terminology, and give more control over model behavior.
"Forge lets enterprises and governments customize AI models for their specific needs," said Elisa Salamanca, Mistral's head of product.
The business angle: Mistral is on track to surpass $1 billion in annual recurring revenue this year, according to the Financial Times—remarkably, achieved while focusing almost entirely on enterprise rather than consumers, the opposite of OpenAI and Anthropic's trajectories.
The platform includes access to Mistral's open-weight models (including Mistral Small 4), infrastructure guidance, and what the company calls "forward-deployed engineers"—consultants who embed directly with customers to help build custom models. This model mirrors IBM and Palantir's consulting-style approach.
Early adopters: Ericsson, European Space Agency, Italian consulting firm Reply, Singapore's DSO and HTX, and ASML (the Dutch chipmaker that led Mistral's €11.7 billion Series C).
Primary source: TechCrunch