On September 29th, according to a report from market research firm JIBANG Consulting, yesterday (September 28th), they published a blog post stating that under the wave of AI intelligence, the global delivery cycle for CPU servers has extended to 25-30 weeks, whereas under normal circumstances it is 16-20 weeks.

Taking Meta's Muse as an example, this institution points out that the AI intelligent entity has become an important development direction for the current AI infrastructure, and it also drives the market demand for servers CPU.

Muse The service can execute multi-step tasks and access websites through a cloud-based browser. Even if the user closes the application, the agent can still process tasks in the background and manage specific transactions and operational procedures.

Muse runs in independent virtual machines (Virtual Machine, VM). Each virtual machine is equipped with a separate browser, and user data and credentials are saved on it. The system also incorporates Sentinel security mechanisms to continuously monitor the operating environment.

Meta Previously, it was proposed to configure a dedicated virtual machine for each Muse user. The relevant configuration included 2 virtual CPU ( vCPU), 8GB of memory, and 100GB of solid-state drive. Therefore, millions of additional CPU are needed to support this.

The report also indicates that the supply of DRAM remains very tight. In a balanced market, the supply cycle is 20 weeks, whereas under normal circumstances it is 8 weeks. Discussions about the tight supply and rising prices from the fourth quarter of 2026 to the first quarter of 2027 have been intensifying.