TrendForce reports that the global lead time for server CPUs has increased to 25–30 weeks, compared to the standard 16–20 weeks. This shift is driven by the rising demand for AI agents in infrastructure.
Meta's Muse project serves as a key example of this trend. The service allows AI agents to perform multi-step tasks and manage operations via cloud-based browsers, even when the user application is closed.
Each Muse user requires a dedicated virtual machine configured with 2 vCPUs, 8GB of RAM, and 100GB of SSD storage. This architecture necessitates the deployment of millions of additional CPUs.
The report also highlights a significant supply shortage in the DRAM market. Lead times for memory have reached 20 weeks, up from the typical 8 weeks, with concerns growing over potential price hikes through early 2027.
