About 600,000 Nvidia H200 chips stranded in Hong Kong; China has not yet approved large-scale imports

This previously undisclosed scale of inventory reflects the complex trade-offs China faces between developing its artificial intelligence industry and reducing its dependence on American technology. On the one hand, Chinese tech companies' demand for high-performance AI computing power continues to grow; on the other hand, regulators remain cautious about expanding dependence on advanced American technology.

According to people familiar with the matter, although large-scale imports have not yet been approved, China has recently allowed some H200 chips to enter the domestic market. Over the past few months, several major Chinese internet companies have each obtained about 10,000 H200 chips. In addition, some chips have entered the Chinese market through middlemen and indirect channels, though the specific quantities have not been disclosed.

As of now, the relevant Chinese authorities have still not approved large-scale imports of the H200. Neither China's Ministry of Industry and Information Technology nor the National Development and Reform Commission has responded to the situation. Nvidia, as well as companies such as Alibaba, ByteDance, DeepSeek and Tencent, have also not publicly commented.

Although the H200 is no longer Nvidia's latest-generation AI processor, its performance remains highly competitive, especially when used with Nvidia's software ecosystem, and it still holds significant value for AI model training. At present, domestic Chinese alternatives have not yet fully reached the same level in some key capabilities.

Since 2022, the United States has continued to tighten export restrictions on advanced semiconductors and their manufacturing equipment in order to reduce China's channels for obtaining high-end AI chips. With legal supply limited, Chinese buyers have begun to rely more on intermediaries and indirect procurement chains, which has also caused gray-market prices to keep rising.

People familiar with the matter said that as of the end of October, the quoted price in the Chinese market for a server equipped with eight H200 chips had exceeded 6 million yuan, roughly three times the price a year earlier. Meanwhile, servers equipped with Nvidia's newer-generation Blackwell architecture chips, also affected by U.S. export controls, have seen per-unit prices exceed 20 million yuan.

It is worth noting that the H200 chip is currently at the intersection of the U.S.-China technology and trade rivalry. The U.S. government had banned exports of the product to China, but after approval by the Trump administration last year, Nvidia regained a license to sell H200s to vetted Chinese customers. The adjustment is believed to be related to progress in U.S.-China economic and trade negotiations at the time.

However, the U.S. export license has not automatically translated into large-scale access to the Chinese market. The current situation has created a peculiar phenomenon: the chips can leave the U.S. supply system, yet they cannot smoothly enter the mainland Chinese market on a large scale.

Analysts believe the current impasse over import approvals highlights the practical challenges facing China's tech industry. As AI competition continues to escalate, Chinese companies need access to advanced computing power to support their own model development, while regulators hope to secure more room for the domestic chip industry to grow.

With demand continuing to grow and approval progress limited, large quantities of H200 chips remain piled up in Hong Kong awaiting a final decision. When this enormous trove of computing resources can enter the Chinese market is becoming a focus of attention for the global AI industry chain and the chip trade.

Related articles:

U.S. confirms Nvidia H200 chips have begun exports to China, involving several leading Chinese AI companies

Revenue from H200 products shipped by Nvidia to Chinese customers accounts for less than 1% of total data center revenue