Intel has said that Microsoft’s Copilot AI service will soon be capable of running locally on PCs. They stated that next-generation AI PCs would necessitate built-in neural processing units (NPUs) with over 40 trillion operations per second (TOPS) of power — surpassing the capabilities of any current consumer processor.
Read: Samsung Galaxy S24 Review: Pocketable powerhouse
According to Intel, these AI PCs will be equipped to handle “more elements of Copilot” locally, reducing the reliance on cloud processing for small requests and minimizing lag. This enhancement in local compute capability not only promises to improve performance but also addresses privacy concerns associated with cloud-based processing.
Previously, there were rumours that Microsoft would require 40 TOPS on next-gen AI PCs, along with 16GB of RAM. While Windows currently underutilizes NPUs, other operating systems like ChromeOS and macOS leverage NPUs for various video and audio processing tasks, OCR, translation, and live transcription.
Among existing processors, Apple’s M3 series offers the highest NPU speed with 18 TOPS. Following closely are AMD’s Ryzen 8040 and 7040 laptop chips with 16 and 10 TOPS respectively, while Intel’s Meteor Lake laptop hits 10 TOPS. Qualcomm’s Snapdragon X Elite is anticipated to be the first processor with sufficient power for Copilot, offering 45 TOPS of AI compute speed.
Looking ahead, Intel’s Lunar Lake chips, expected in 2025, will triple its current NPU speeds. Additionally, Intel recently introduced 300 new AI features optimized for its OpenVino platform and unveiled an AI PC development kit based on the ASUS NUC Pro utilizing Meteor Lake silicon.
Intel emphasized its commitment to advancing AI capabilities across different segments, including desktop and next-gen AI PCs, as outlined in its roadmap.
