Microsoft wants to make it easier for developers to build AI-powered apps on Windows devices.
On Monday during its Build 2025 conference, Microsoft announced Windows AI Foundry, a rebranding and expansion of the Windows Copilot Runtime service the company launched last May. Microsoft describes Windows AI Foundry as a “unified platform for local AI development” — a way to fine-tune, optimize, and deploy the AI models underpinning Windows apps.
Microsoft increasingly sees AI, whether running locally or in the cloud, as a major profit driver, despite the high costs associated with developing tentpole models. In January, Microsoft CEO Satya Nadella said the tech giant has $13 billion in annualized AI revenue, in part thanks to its partnership with ChatGPT maker OpenAI.
Microsoft might not profit directly from third-party AI-powered apps on Windows. Still, by providing resources to make it easier to build AI software on the tech giant’s flagship operating system, Microsoft aims to foster a flourishing ecosystem at the expense of rivals. Hence, Windows AI Foundry.
According to a press release shared with TechCrunch, Windows AI Foundry can automatically detect a Windows machine’s hardware and fetch the necessary software components to run a particular AI model. Windows AI Foundry will also keep these components up to date as new devices are released, Microsoft says, and deliver tools designed to simplify the process of prepping models.
Windows AI Foundry also includes Foundry Local, a new service that “bring[s] the power of [AI] models […] to client devices,” explains Microsoft in the press release. Foundry Local, which supports macOS in addition to Windows, comes with a set of modules to run AI models and tools directly on-device, leveraging the ONXX Runtime, an open source project aimed at accelerating AI across platforms.
Foundry Local also ships with command-line interface support, letting developers use prompts like “Foundry model list, Foundry model run” to browse, test, and interact with models running on a local server.
Join 10k+ tech and VC leaders for growth and connections at Disrupt 2025
Netflix, Box, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, Vinod Khosla — just some of the 250+ heavy hitters leading 200+ sessions designed to deliver the insights that fuel startup growth and sharpen your edge. Don’t miss the 20th anniversary of TechCrunch, and a chance to learn from the top voices in tech. Grab your ticket before doors open to save up to $444.
Join 10k+ tech and VC leaders for growth and connections at Disrupt 2025
Netflix, Box, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, Vinod Khosla — just some of the 250+ heavy hitters leading 200+ sessions designed to deliver the insights that fuel startup growth and sharpen your edge. Don’t miss a chance to learn from the top voices in tech. Grab your ticket before doors open to save up to $444.
“Foundry Local will automatically detect device hardware — CPU, GPU, and NPU — and list compatible models for developers to try,” Microsoft writes. “Developers will be able to leverage the Foundry Local SDK to easily integrate Foundry Local in their app.”
