The artificial intelligence race is creating a new bottleneck, and it is no longer concerning Nvidia’s GPUs. Intel and AMD are allegedly negotiating long-term purchase commitments with Chinese server customers, as demand for data center processors surpasses supply. While GPUs have been the highlight of discussions around AI infrastructure shortages, the newest negotiations suggest that server CPUs are becoming the next point.
Prices for some processors in China have surged more than 40% this year, with timely increases exceeding 10% for certain products. The development also emphasizes an unusual reality in today’s damaged U.S.-China technology relationship. Despite export constraints and tensions, Chinese cloud providers continue to rely on Intel and AMD CPUs, while the chip makers are looking for multi-year purchasing commitments to secure production planning and monetize tight supply.
Why Are Intel and AMD Seeking Longer CPU Commitments?
The AI race is not only about driving demand. Every AI data center also needs thousands of CPUs to manage servers, networking, storage, orchestration, and inferencing workflows. According to sources, Intel and AMD are asking Chinese customers to buy volumes for roughly one year, with some discussions extending to two years. Unlike conventional procurement, these agreements secure supply volumes rather than fixed costs, allowing chipmakers to manage the amount while demand continues rising.
The shift reflects what has already happened in the AI memory market, where shortages have pushed customers towards massive purchasing requirements. Intel CEO Lip-Bu Tan has acknowledged the demand for Xeon server processors, but this has surpassed supply, while AMD has increased its long-term server CPU market forecast, quoting growing demand from agentic AI workloads.
For much of the generative AI boom, GPUs were looked upon as the industry’s primary restriction. That picture is now changing. As organizations rapidly expand AI clusters, demand is rising simultaneously for CPUs, memory, network equipment, storage systems, and supporting infrastructure. Server processors, which were considered easy to acquire, are now experiencing longer lead times and sustained price increases. China shows this shift clearly. Despite U.S. constraints on advanced AI GPUs, the country continues building data centers, cloud infrastructure, and national computing facilities, increasing competition for Intel and AMD processors.
The result is a comprehensive instance shortage, where CPUs are becoming as crucial as accelerators. Negotiations therefore represent more than routine supply contracts; they show that the AI model is spreading across the entire computer stack, forcing customer to secure long-term access to parts that were previously considered abundant. Even amidst fragmented U.S.-China technology tensions, demand for AI infrastructure continues to join global semiconductor supply chains together, making server CPUs the latest resource in the AI race.
Why Server CPUs Are the New AI Bottleneck
Initially, the conversations revolved around GPUs. They became synonymous with artificial intelligence because they perform the necessary computations required to train and run large language models. However, AI infrastructure cannot work on GPUs alone. Every AI server rack also requires CPU to manage workloads, manage storage, handle networking, allocate resources, and complement inference tasks across data centers. As more organizations build AI clusters, demand for these processors increases alongside the GPUs. This explains why CPU inventories are beginning to tighten.
Unlike prior years, when CPUs were relatively easy to acquire, cloud providers are now competing for the same proprietary-grade processors needed to support expanding AI infrastructures. Reuters earlier reported that Intel had informed Chinese customer of lead time stretching up to six months for some Xeon server processors, while AMD also predicted strong long-term growth in the server CPU market because of agentic AI workloads. This indicates that processors once viewed as basic components are now becoming strategic resources.
The deal between Intel, AMD, and Chinese customer shows that the infrastructure race will enter a new phase. The biggest bigger restriction is no longer limited to GPUs, as CPUs are becoming equally necessary. Even amid fragmented U.S.-China technology tensions, both sides remain monetarily dependent on each other, with Chinese firms seeking dependent processor supply and American chip makers looking for long-term demand. As AI adoption continues to expand, securing every layer of the compute stack will determine how organizations build the next generation of AI infrastructure.
