Intel and AMD Face Server CPU Delivery Delays in China Due to 3nm Capacity Bottleneck

According to foreign media reports, the surge in server CPU demand in China has been triggered by a sharp increase in DRAM and NAND flash memory prices at the end of last year. As a result, domestic cloud service providers and large-scale data centers are accelerating their equipment upgrades. Intel's Xeon 6-series processors are now facing a delivery cycle of up to six months, while AMD's fifth-generation EPYC (Milan-X) processors are seeing a delivery cycle of approximately 10 weeks.

The significant delay in Intel's server CPU deliveries compared to AMD is primarily due to a combination of factors, including the slower production progress of Intel’s 3nm process, limited wafer fabrication capacity, and internal capacity allocation strategies.

Specifically, AMD’s 3nm (N3) process, which is manufactured by TSMC, began mass production in 2022, and its yield control and capacity release have since entered a stable and mature phase, providing a reliable supply for the fifth-generation EPYC processors. In contrast, Intel's self-developed Intel 3 process is not expected to begin mass production until the second half of 2024, with both production scale and manufacturing maturity significantly lagging behind TSMC's 3nm process.

On the production capacity front, AMD has the flexibility to tap into TSMC’s global foundry network to adjust resource allocation, while Intel’s server CPU production capacity primarily relies on its own wafer fabs. Under the current surge in demand, Intel’s in-house capacity bottleneck has become increasingly prominent. Additionally, to prioritize the high-margin data center market, Intel has redirected a larger portion of its Intel 3 process capacity towards Xeon 6-series processors. While AMD also faces intense competition in securing additional 3nm capacity from TSMC, it benefits from greater flexibility in resource allocation within its existing capacity pool, giving it a competitive edge in delivery times compared to Intel.

Industry sources have also noted that, in addition to delays in the delivery of Intel Xeon 6-series processors and AMD’s fifth-generation EPYC processors, large-scale data center operators and cloud service providers are facing sustained high prices for server ECC memory and NAND flash storage. This combination of pressures could prompt some companies to reassess their server upgrade plans and even delay procurement timelines.