TradingKey - The supply-demand tightness brought by AI infrastructure construction may be spreading from GPUs to the server CPU market.
According to Reuters citing sources familiar with the matter, Intel ( INTC) and AMD ( AMD) are recently negotiating long-term supply agreements with Chinese server manufacturers and large customers. Unlike traditional orders, these agreements primarily agree on future purchase volumes rather than fixed transaction prices. Most contracts cover a supply period of about one year, but some customers have begun discussing procurement arrangements for two years or even longer.
This change reflects a shift in the supply-demand dynamics of the server CPU market. Over the past few years, the AI industry's greatest shortage has been in GPUs, but now, with the accelerated deployment of AI servers, storage systems, networking equipment, and inference clusters, demand for server CPUs—as a vital component of overall data center infrastructure—is climbing in tandem.
Industry insiders pointed out that an increasing number of data center projects require not only large quantities of AI accelerators but also high-performance CPUs to handle tasks such as server control, storage management, network scheduling, and inference. As a result, the wave of AI investment is starting to drive the entire data center industry chain into a state of tight balance.
Sources revealed that since the beginning of this year, cumulative price increases for certain server CPU products in the Chinese market have exceeded 40%, with single-month increases for some models exceeding 10%. Meanwhile, delivery times have also lengthened significantly. Previous reports indicated that Intel had notified some Chinese customers that lead times for some of its server CPU products could reach up to six months.
Intel Chief Executive Officer Lip-Bu Tan stated in April this year that demand for the company's Xeon server processors continues to outstrip supply, and disclosed that multiple long-term supply contracts had been signed in the first quarter, including a multi-year cooperation agreement with Google. The market expects that server CPU supply and demand will also be a key focus when Intel announces its quarterly results this week.
AMD is also optimistic about this market prospect. The company had previously raised its forecast for the global server CPU market size by 2030 to over $120 billion, believing that next-generation AI applications such as Agentic AI will continue to drive up data center demand for high-performance processors.
AMD plans to announce its latest quarterly results in early August, and the market will also focus on its server business orders and future capacity arrangements.
As one of the world's largest server markets, China continues to advance the construction of data centers, AI computing clusters, and various computing power infrastructures. Although advanced AI GPUs are still subject to U.S. export restrictions, growing demand for computing power continues to keep server CPU procurement at high levels.
With suppliers' bargaining power increasing and long-term procurement agreements becoming more common, the server CPU market is exhibiting a development trajectory similar to that of the previous HBM and DRAM markets. The supply-demand tightness brought by AI has further expanded from acceleration chips to the entire data center hardware ecosystem.
Find out more
Comments