Microsoft announced Thursday the formation of a brand-new company aimed at helping large businesses figure out how to make artificial intelligence actually work for them — and make money doing it.
The new entity, called Microsoft Frontier Company, is launching with $2.5 billion in backing from the tech giant. Among its first clients are major global companies including Unilever and Novo Nordisk.
The move comes as big corporations are stepping away from relying on a single AI provider and instead piecing together a variety of technologies — including open-source models — customized to fit their specific needs. That approach can be expensive and slow down the time it takes to see a financial return.
Microsoft Frontier Company will step in to help clients sort through and combine AI tools — whether they come from Microsoft or other companies — and connect them with that client’s own proprietary data. Importantly, whatever is built through that process stays with the client rather than being returned to Microsoft.
Microsoft is not alone in this space. Palantir Technologies is already doing similar work using Nvidia’s open-source models with large clients, and cloud competitor Amazon Web Services recently launched its own embedded-engineer program worth $1 billion.
Patrick Moorhead, CEO of the analyst firm Moor Insights & Strategy, noted that many large businesses are growing concerned that relying on AI models from companies like Anthropic and OpenAI could eventually give those AI labs enough knowledge to compete against them — particularly in areas like legal services and software coding.
Microsoft holds a partial ownership stake in OpenAI, the maker of ChatGPT, and earlier this year added models from Anthropic to its Copilot AI assistant, responding to strong demand from business customers for that technology.
Judson Althoff, CEO of Microsoft Commercial Business, said the idea for the new company grew in part from lessons Microsoft learned when AI models from China’s DeepSeek and Google’s Gemini began closing the gap with OpenAI’s offerings.
“Three years ago, when we built Copilot, we made a mistake by binding it to OpenAI models only,” Althoff told Reuters. “You wanted models to amplify your intelligence and be able to have that sort of swappability for state-of-the-art and fine-tuning.”
Althoff added that what matters most to customers is the combination of their own data with AI models — not any single model in particular — and that businesses need the ability to quickly switch between different AI options as the technology evolves.
