When Microsoft presented the first Copilot computers , capable of fully exploiting the upcoming AI-based Windows functionalities, only machines equipped with Qualcomm’s new Snapdragon X Elite SOCs were eligible. No machine with an AMD or Intel processor, even one with an NPU, was present. The minimum requirements imposed by Microsoft did not allow the two American giants to be operational for Copilot at the time of launch. But there was another absentee brand whose name has been associated with AI for some time: Nvidia.
Porting, since 2018 and the arrival of RTX 20xx, Nvidia has implemented AI processing and acceleration capabilities in its GPUs. The latest Geforces cards far exceed the NPU performance of Qualcomm SoCs…Yet Microsoft has not allowed its Copilot runtime to access this performance. But things are about to change.
Windows Copilot Runtime to add GPU acceleration for local PC SLMs
Microsoft and NVIDIA are collaborating to help developers bring new generative AI capabilities to their native and Web-based Windows applications. The collaboration will provide application developers with easy application programming interface (API) access to GPU-accelerated small language models (SLMs) that enable recovery augmented generation (RAG) capabilities running on a Windows Copilot Runtime-powered device.
Nvidia takes this philosophically. Better still, it’s using it as a marketing tool. For the company, NPU use is confined to light AI, while their products are intended for heavy AI processing, including on PCs.
Up to now, Microsoft seemed to be concentrating on the operation of Copilot on the NPUs of future processors and SoCs. After that, RTX support will arrive to give users access to the enhanced AI performance of dedicated GPUs. This also suggests that RTXs may not be the only cards concerned.
