Advancing Windows ML Acceleration with AMD at Microsoft Build 2026
AMD, Tuesday, June 2nd, 2026
AMD announces NPU and GPU acceleration improvements for Windows ML at Microsoft Build 2026.
AMD has unveiled significant enhancements to its Windows ML acceleration capabilities, including improved NPU and GPU Execution Providers that boost AI inference performance across diffusion models and large language models (LLMs).
The company has doubled model coverage, introduced WebNN support for browser-based inference, and upgraded to ROCm 7.1 for better kernel performance and memory efficiency. Key advancements include a new GPU Execution Provider plugin interface for faster iteration, optimizations for popular diffusion models like Stable Diffusion 3.5 and FLUX.1, and expanded hardware support for AMD Ryzen AI 400 Series processors.
Additionally, AMD has launched a preview driver for the DirectX Compute Graph (DxCGC) integration with Windows ML, enabling more efficient mapping of ML graphs to AMD Radeon GPUs and supporting Microsoft PhiSilica workloads.