Intel and AMD are set to enter into longer-term contracts with Chinese customers for the supply of processors to data centres. According to sources at Reuters, the companies aim to streamline their orders at a time when chip availability is dwindling and prices are rising rapidly.
Most of the contracts under discussion cover around one year, although in some cases commitments of at least two years are being negotiated. The contracts are intended to guarantee specific purchase volumes, but do not fix prices in advance. This means that customers gain greater certainty of supply, but still face the risk of further price rises. Intel and AMD have not commented on this information, and Reuters’ sources have remained anonymous.
The change is yet another consequence of the rapid expansion of artificial intelligence infrastructure. Data centres require more than just graphics processing units, which are used to train and run AI models. Central processing units (CPUs) are responsible, amongst other things, for server management, data processing, mass storage, networking and coordinating the operation of accelerators.
Intel officially acknowledged in April that demand continues to outstrip supply, particularly for its Xeon server chips. The company has also signed a multi-year agreement with Google covering future generations of processors used in cloud and AI infrastructure.
AMD, on the other hand, believes that the development of so-called agent-based artificial intelligence could increase the demand for separate CPU servers operating alongside GPU clusters. The manufacturer forecasts that the value of the server processor market will exceed $120 billion by 2030. However, this is the company’s own forecast, not an independent market assessment.
According to Reuters, prices for selected server processors in China have risen by over 40 per cent since the start of the year, and lead times for some Intel products have reached up to six months. For Chinese cloud operators, this means rising costs and the possibility of delays in launching new AI services. It also shows that the artificial intelligence boom now encompasses the entire data centre infrastructure, not just the most advanced GPUs.
