Intel has announced that Microsoft plans to run the artificial intelligence assistant Copilot locally on artificial intelligence computers.
Intel also stated that next-generation artificial intelligence computers would require neural processing units with power up to 40 trillion operations per second, surpassing the capabilities of any current consumer processor on the market.
Intel mentioned that computer devices containing artificial intelligence must be capable of executing Copilot functions locally.
Currently, Microsoft’s artificial intelligence assistant performs most activities in the cloud, including small requests.
This leads to delays in completing large tasks, and the possibility of local computing can help reduce these delays, thus improving performance and privacy as well.
There were previous rumors indicating that Microsoft would require power up to 40 trillion operations per second across next-generation artificial intelligence computers, in addition to 16 gigabytes of RAM.
Currently, neural processing units are not extensively used in Windows except for running video effects such as background blur for webcameras in Surface Studio.
ChromeOS and macOS utilize the power of neural processing units to run video and audio processing features, as well as optical character recognition, translation, live transcription, and other features.
Apple’s M3 series processors stand out with the strongest neural processing units, offering 18 trillion operations per second through the M3, M3 Pro, and M3 Ultra models.
AMD’s Ryzen 8040 and Ryzen 7040 computer chips come second with speeds of 16 trillion and 10 trillion operations per second respectively. Meanwhile, Intel’s Meteor Lake processors offer speeds of 10 trillion operations per second.
Qualcomm has developed the first processor with sufficient power to run Microsoft’s Copilot AI assistant through Snapdragon X Elite processors, which boast a power of 45 trillion operations per second in AI computation speed.
It is expected that Intel’s Lunar Lake chips will arrive in 2025 with double the speed of current neural processing units.
Intel aims to enhance its artificial intelligence features through its processors, planning to improve 300 AI features specifically for the OpenVino platform via Meteor Lake processors.
The giant chip company also announced that it owns a set to develop AI computers based on the Asus NUC Pro device powered by the current processor Meteor Lake.
