Microsoft is stepping up its push to simplify AI-powered application development for Windows devices. During its recent Build 2025 conference, the company unveiled an updated and rebranded offering called Windows AI Foundry, a major evolution from the Windows Copilot Runtime service initially launched the previous May. According to Microsoft, the newly announced Windows AI Foundry serves as a comprehensive, streamlined platform designed specifically for local AI application development, providing developers with essential tools and infrastructures to fine-tune, optimize, and deploy AI models seamlessly within Windows applications.
Artificial intelligence continues to emerge as a key revenue-driving area for Microsoft. Despite the substantial costs associated with building flagship AI models, CEO Satya Nadella revealed in January that Microsoft’s annualized AI-related revenues have already hit roughly $13 billion, buoyed notably by its partnership with OpenAI, creator of ChatGPT.
Although Microsoft doesn’t directly profit from third-party AI-powered apps deployed on Windows, the company is investing strategically in developer resources and productivity enhancements to cultivate a vibrant ecosystem on its flagship operating system. The launch of Windows AI Foundry fits neatly within this broader objective, aiming to attract developers by making AI integration simpler and more accessible, potentially creating an environment less hospitable to competing platforms.
A key feature of Windows AI Foundry is its ability to automatically recognize the hardware capabilities of Windows machines and install the necessary software packages to execute specific AI models smoothly. The platform also ensures these components are regularly updated, accommodating new devices and developments. Additionally, Microsoft highlights specialized toolsets within Windows AI Foundry that streamline the often complex preparations required for model deployment.
Another significant addition to the Windows AI Foundry is Foundry Local, a service dedicated to bringing sophisticated AI model capabilities directly onto client devices. Notably, Foundry Local supports macOS along with Windows, signaling Microsoft’s intention to appeal broadly to the developer community, irrespective of the underlying operating system used. Built around the ONNX Runtime, an open-source engine designed to speed up AI processing across multiple platforms, Foundry Local comes complete with various modules enabling developers to execute AI models directly on their hardware.
For increased ease of use, Foundry Local provides a command-line interface, letting developers list available models, run tests, and interact directly with locally hosted AI solutions through simple text commands such as “Foundry model list” and “Foundry model run.” Furthermore, Foundry Local’s software development kit (SDK) helps developers seamlessly integrate these locally deployed architectures into their own applications.
According to Microsoft, Foundry Local automatically identifies a device’s hardware—including CPUs, GPUs, and neural processing units (NPUs)—and displays compatible models, allowing developers to quickly find effective solutions tailored precisely to their hardware configurations. The ultimate aim is to streamline development significantly, empowering developers to create robust AI-driven software more efficiently and reliably.
