As AI computing demand continues to grow, data center infrastructure requirements are shifting. After being overshadowed by the rise of GPUs for over a decade, CPUs are once again taking center stage. Since late last year, CPU supply has suddenly become very tight. Both Intel and AMD have decided to prioritize server products in their production capacity, gradually raising CPU prices and extending delivery cycles. To improve overall profitability, Intel also plans to streamline its product lineup by discontinuing some lower-margin products.
According to Wccftech, recent reports indicate that CPU supply is becoming increasingly tight, driven by the demand for personal AI agents like Meta Muse. This is reflected in CPU lead times, which have extended from the usual 16 to 20 weeks to 25 to 30 weeks.
These personal AI agents require cloud-based computing services. Meta has promised that each Muse user will have a dedicated virtual machine configured with 2 vCPUs, 8GB of RAM, and a 100GB SSD. Initial estimates suggest this could generate millions of additional CPU demands to support daily active users. Although personal AI agents are still in their early stages and there are many variables in projected CPU demand, the trend is clear, representing an extremely optimistic outlook for both Intel and AMD.
The entire semiconductor supply chain is now under strain, not just CPUs. TrendForce reports that DRAM lead times have reached 20 weeks, far above the normal 8 weeks, while SSDs and HDDs have reached 16 and 50 weeks respectively, compared to their normal 8 and 16 weeks. The only component remaining at normal levels is GPUs, which are still maintaining a lead time of 30 to 40 weeks.