Muse Scale Challenge: Meta Might Need 1.58M CPU Cores, 100PB RAM for 100M Users
Analysts estimate that delivering Meta's Muse personal AI to 100 million users with dedicated cloud VMs (2 vCPUs, 8 GB RAM, 100 GB SSD per user) could require about 1.58 million AMD EPYC CPUs, roughly 100 petabytes of RAM and 10,000 petabytes of SSD; CPU sharing could reduce core counts but RAM/storage demands are daunting, and expansion to platforms like WhatsApp/Instagram could further drive compute needs.