Laptop NPU race outpaces software support
Neural processing units are becoming standard in midrange and higher laptops as manufacturers pursue Microsoft’s Copilot+ PC certification. These dedicated chips handle certain AI tasks at lower power than CPUs or GPUs, supporting battery-sensitive uses such as background blurring, noise removal, real-time translation and local search. Cloud services such as ChatGPT and Gemini do not require an NPU on the user’s device.
Microsoft’s certification requires at least 40 TOPS of NPU performance, 16 GB of memory and 256 GB of storage, excluding GPU computing performance. Qualcomm’s Snapdragon X Elite offers 45 TOPS, Intel’s Lunar Lake reaches up to 48 TOPS and AMD’s Ryzen AI 300 delivers up to 50 TOPS. Qualcomm’s Snapdragon X2 Elite raises performance to as much as 85 TOPS.
Software support remains limited, with prominent uses concentrated in Windows Studio Effects, Live Captions, Recall and selected conferencing and content-creation features. In a local test of Mistral 7B using Intel OpenVINO, NPU utilization reached 98 percent while CPU utilization was 3 percent. Running the same model through Ollama pushed CPU utilization to 52 percent. Dedicated tools and technical expertise are often needed to use NPUs effectively, leaving practical software availability a key competitive factor.