Intel and AMD Reportedly Lock In Year Long CPU Deals With Chinese AI Data Centers

Intel and AMD are reportedly signing long term server CPU supply agreements with Chinese customers as artificial intelligence infrastructure demand pushes processor prices higher and places additional pressure on availability. According to Reuters, most of the agreements cover approximately 1 year of processor supply, although commitments lasting 2 years or longer have also been discussed with selected customers.

The agreements primarily secure purchase volumes rather than fixed prices, allowing Intel and AMD to continue adjusting prices as market conditions change. This provides Chinese cloud providers and data center operators with greater supply visibility, but it does not completely protect them from additional cost increases. Intel and AMD did not comment publicly on the reported negotiations.

Demand for AI infrastructure is increasingly extending beyond GPUs into CPUs, memory, storage, and networking equipment. While accelerators perform the largest training and inference calculations, server CPUs remain essential for coordinating workloads, feeding data to accelerators, managing storage systems, controlling networking, and operating the wider software environment supporting AI services.

The resulting supply pressure has caused server CPU prices to rise rapidly in China. Reuters reports that prices for certain processors have increased by more than 40% since the beginning of 2026, while monthly increases have exceeded 10% for some products. Intel previously warned Chinese customers that delivery times for selected server CPUs could reach as long as 6 months, increasing pressure on companies expanding AI computing clusters.

China remains one of the largest server markets worldwide, supported by extensive investment in cloud services, national computing infrastructure, and artificial intelligence data centers. Restrictions affecting access to advanced NVIDIA accelerators have not reduced the need for conventional processors, since every AI cluster still requires large quantities of CPUs to operate servers, storage, networking, and inference services.

Intel has repeatedly highlighted the return of the CPU as a critical component of AI infrastructure. During its Q1 2026 financial results, the company said the transition from training large foundational models toward deploying specialized AI agents at scale was significantly increasing demand for CPUs, manufacturing capacity, and advanced packaging. Intel later reported that CPU orders had exceeded its available production capacity, contributing to stronger data center revenue and additional long term customer agreements.

"This shift is significantly increasing the need for Intel’s CPUs and wafer and advanced packaging offerings."
— Quote by: Lip Bu Tan

AMD has also raised its outlook for the server CPU market to more than US$120 billion by 2030, citing the expansion of agentic AI workloads. Its upcoming Zen 6 EPYC Venice processors will scale to 256 cores and are being positioned as the CPU foundation for cloud, enterprise, high performance computing, and rack scale AI platforms.

The companies are also collaborating on the future of the x86 architecture through the x86 Ecosystem Advisory Group, which was established in 2024. Its first technical milestones included FRED for modernized interrupt handling, AVX10 for consistent vector computing, ChkTag for memory safety, and ACE for standardized matrix multiplication capabilities across laptops, workstations, and data center processors.

The reported agreements show that the AI supply crisis is no longer limited to accelerators and high bandwidth memory. CPUs are becoming another strategically allocated component as agentic systems require more orchestration, storage access, networking, security, database processing, and general computing around each GPU cluster.

For Intel and AMD, long term volume commitments provide stronger demand visibility and greater confidence when expanding production. For Chinese customers, however, the agreements represent supply protection rather than price protection. Buyers may secure the processors required to complete future data centers while still paying more if market prices continue rising.

The wider risk is that enterprise and AI customers receive production priority while smaller server builders and conventional PC markets face longer lead times or higher prices. As AI infrastructure consumes an increasing share of global semiconductor capacity, access to CPUs may become almost as important as access to GPUs.

Could server CPU shortages become as disruptive as the GPU and memory supply crisis as AI data centers continue expanding?