Microsoft bets $2.5bn on a unit to help firms adopt AI
TL;DR:
- Microsoft is creating Microsoft Frontier Company, a new operating entity seeded with $2.5bn to help customers choose and integrate AI tools.
- Early clients include Unilever and Novo Nordisk, and firms will keep the results of the work rather than hand them back to Microsoft.
- It joins AWS and Palantir in a fast-growing market for hands-on enterprise AI services.
Microsoft is standing up a new company to help large customers select the AI technologies that actually earn a return, backing it with $2.5bn, Reuters reports. Microsoft Frontier Company will integrate tools — its own and rivals’ — with each customer’s internal data, a tacit admission that buying AI and getting value from it are two very different things.
Multi-model, by necessity
The launch reflects how big corporations now buy AI. Rather than renting from a single provider, they blend closed and open-source models and tune them to their needs, which is costly and slow to pay off. Judson Althoff, chief executive of Microsoft’s commercial business, was candid about learning this the hard way: “Three years ago, when we built Copilot, we made a mistake by binding it to OpenAI models only,” he said, adding that customers wanted the “swappability” to move between state-of-the-art systems as China’s DeepSeek and Google’s Gemini caught up.
The competitive read is sharper still. Analyst Patrick Moorhead noted that large firms increasingly suspect the frontier labs could use what they learn to compete with them in fields such as coding and law — an incentive to work through an integrator rather than hand data straight to a model provider. Microsoft is not alone in spotting the gap: Amazon Web Services recently committed $1bn to embed its own AI engineers with customers, and Palantir is doing similar work with open models. For UK businesses — collectively on track to waste some £67bn a year on failed AI projects — the arrival of well-funded “help you actually deploy it” units is arguably more consequential than the next model release.
Looking forward
The margin these units chase is real, but so is the risk of dependency: outsourcing integration can leave firms locked into a vendor’s view of which models and data pipelines matter. British decision-makers weighing Frontier Company or its rivals should press on who owns the resulting systems and skills. Microsoft’s promise that customers keep the output is a start; whether that extends to the underlying capability is the question worth asking before signing.