Prices for advanced AI servers built around Nvidia’s latest chips have surged in China, with some configurations of the new B300 systems reportedly reaching as high as $1 million per unit, according to sources familiar with the market.
The sharp increase is being driven by U.S. export restrictions that limit the sale of cutting edge semiconductor technology to Chinese buyers. These curbs have tightened supply of high performance chips, creating a scarcity that is pushing up prices for systems that are still able to reach the market through approved channels or secondary distribution networks.
The B300 servers, designed for artificial intelligence workloads such as large language models and advanced data processing, are among the most powerful computing systems available. Demand for such hardware in China remains strong, particularly from tech companies, research institutions, and cloud providers seeking to maintain competitiveness in the global AI race.
Industry sources say the restrictions have led to a fragmented supply chain, with some buyers turning to intermediaries or alternative procurement methods to secure the hardware. This has contributed to price inflation and reduced transparency in the market.
Analysts note that the situation underscores the broader impact of geopolitical tensions on the semiconductor industry, where access to high end chips has become a strategic issue. The U.S. has imposed a series of measures aimed at limiting China’s ability to acquire advanced computing technology, citing national security concerns.
Meanwhile, Chinese firms are accelerating efforts to develop domestic alternatives, though experts say it may take time before locally produced chips can match the performance of top tier systems from Nvidia.
The reported pricing highlights the growing premium attached to advanced AI infrastructure under current export controls, as companies weigh the high costs against the strategic importance of maintaining access to cutting edge computing power.
Leave a comment