The Copilot AI, Microsoft’s replacement for the original Cortana, has long had rumors swirling that soon it will be run locally instead of on the cloud. However, nothing’s been official yet, until now. Microsoft themselves has confirmed that Copilot will soon run locally on your PC.
As reported by Tom’s Hardware, the confirmation comes from Intel’s AI Summit that took place in Taipei. Intel executives chatted with the site and confirmed that Copilot will no longer be running on the Cloud in the future. Instead, the AI will run locally on your PC. This news also comes with a 40 TOPS of performance requirement for the Neural Processing Unit (NPU) on all next-gen AI PCs.
Copilot will run locally on PCs, as confirmed by Intel
Running Copilot locally on AI PCs will allow for more privacy benefits, along with improved latency and performance. As the next generation of AI PCs comes, they will meet the requirements of what Microsoft defines as an AI PC, including an NPU, CPU, GPU, Copilot, and a dedicated Copilot button on the keyboard. Some of these PCs are already out, but the next-gen will be the one to enforce the 40 TOPS rule.
Running Copilot locally instead of on the Cloud will allow for a better overall performance of the AI companion. Running it specifically on the NPU instead of the GPU as well, will reduce the strain on the battery and allow for a longer battery life. This is overall a positive change for the AI companion, which so far has been impressive.
The next generation of NPUs will need to take a leap to match up to these requirements, though. The current Intel NPU, the Meteor Lake, only runs at 10 TOPS of performance. The Ryzen Hawk Point NPU from AMD has 16 TOPS. These both fall quite a bit short of the 40 TOPS requirement. Luckily, better TOPS performance will come, including the X Elite chips from Qualcomm which have 45 TOPS. These requirements, along with the new local-run Copilot, will allow for incredible AI PC performance.
