Advancing AI developer solutions: Microsoft and Qualcomm Technologies collaborate on Windows 11, Copilot+ PCs and Windows AI Foundry
In the dynamic field of artificial intelligence, there is a persistent drive to develop effective inference solutions for client platforms. With new workloads such as Agentic AI and large language models (LLMs) demanding increased compute and memory capabilities, Qualcomm Technologies and Microsoft have joined forces to propel AI workloads with solutions that enable new workloads on a vast range of devices powered by Snapdragon X Series processors.
This collaboration leverages Qualcomm Technologies' advanced hardware, including the Snapdragon X Elite Processor, and Microsoft's robust AI tools, such as Windows AI Foundry. Together, they provide developers with the resources needed to create and optimize generative AI (GenAI) applications, enabling efficient performance and scalability across diverse hardware platforms.
By integrating the Qualcomm Hexagon NPU with Windows 11 technologies, developers can achieve exceptional performance and energy efficiency for AI tasks, making advanced AI accessible and practical for a broader audience. This collaboration not only enhances AI capabilities on Windows 11 platform but also drives innovation, empowering developers to push the boundaries of what's possible with GenAI and Agentic AI.
"The integration of Windows ML and the Hexagon NPU is another example of how Microsoft and Qualcomm Technologies’ continued collaboration is propelling AI workloads and enabling developers to accelerate Windows ecosystem innovation," says Logan Iyer, distinguished engineer and Windows Platform+ developer at Microsoft.
Upendra Kulkarni, vice president of product management at Qualcomm Technologies, Inc., adds, “The new Windows ML's cutting-edge runtime not only optimizes on-device model inference but also simplifies deployment, making it easier for developers to harness the full potential of advanced AI processors in Snapdragon X Series platforms. Windows ML’s unified framework and support for diverse hardware, including our NPUs, GPUs and CPUs, allow developers to create AI applications that deliver exceptional performance and efficiency across a wide range of devices. We look forward to continuing our collaboration with Microsoft to drive innovation and velocity of development to bring the best AI experiences on Windows Copilot+ platforms.”
Microsoft and Qualcomm Technologies: A strong collaboration
Microsoft and Qualcomm Technologies’ collaboration bring new advanced AI features into Copilot+ PCs, leveraging the Snapdragon X Elite. Microsoft Research has played a pivotal role by optimizing new lightweight LLMs, such as Phi Silica, specifically for on-device execution with the Hexagon NPU. These models are designed to run efficiently on Hexagon NPUs, enabling multimodal AI experiences like vision-language tasks directly on Copilot+ PCs without relying on the cloud.
Additionally, Microsoft has made DeepSeek R1 7B and 14B distilled models available via Azure AI Foundry, further expanding the AI ecosystem on the edge. This collaboration marks a significant step in democratizing AI by making powerful, efficient models accessible on everyday devices
This collaboration focuses on utilizing the Hexagon NPU to deliver exceptional performance and energy efficiency for AI tasks on Windows 11 PCs.
Scaling optimization through Olive toolchain
The Windows ML foundation of the Windows AI Foundry provides a unified platform for AI development across various hardware architectures and brings silicon performance using QNN Execution provider. This stack includes Windows ML and toolchains like Olive, easily accessible in AI Toolkit for VS Code, which streamlines model optimization and deployment.
Qualcomm Technologies has contributed to Microsoft’s Olive, an open-source model optimization tool that enhances AI performance by optimizing models for efficient inference on client systems. This tool is particularly beneficial for running LLMs and GenAI workloads on Qualcomm Technologies’ platforms.
Real-World Applications
Through Qualcomm Technologies and Microsoft’s collaboration we have partnered with top developers to adopt Windows ML and have demonstrated impressive performance for their AI features. Independent Solution Vendors (ISVs) such as Powder, Topaz Labs, Camo and McAfee, are able to effortlessly port their AI features over to the Hexagon NPU with the developer preview of Windows ML.
"The Hexagon NPU delivers standout performance for Powder’s AI features. There are many runtime options available to target different SoCs but thanks to the Windows ML API from Microsoft we don't have to worry about all the different runtime configurations on each hardware platform. Thanks to all the work Microsoft and Qualcomm Technologies put in to get Windows ML in the hands of developers — it's a game-changer for on-device AI," says Maryan Morel, co-founder and CTO of Powder.
Carl Woodward, senior principal engineer at McAfee, adds, “We’re excited about the efficiencies Windows ML on the Heaxgon NPU can bring to the development and management of the new scam detection capabilities in McAfee+. Windows ML will allow us to focus on high-impact areas like model accuracy and performance, while providing confidence that AI components work well across the entire ecosystem, including new hardware revisions.”
