NVIDIA is expected to introduce a new and more affordable artificial intelligence chipset intended for the Chinese market, and mass production is scheduled to begin as early as June. This endeavor comes as the company makes its way around strict U.S. export regulations that are aimed at repressing China’s technological advancements.

The soon-to-come GPU is a part of NVIDIA’s latest Blackwell-architecture AI processors. It is anticipated to be priced between $6,500 and $8,000. This range is much lower than the considerable $10,000-$12,000 cost of the recently restricted H20 model. Audiences familiar with the launch indicate that the narrower price is due to the chip’s weaker specifications and more straightforward manufacturing requirements. 

Reportedly, it will be based on NVIDIA’s RTX Pro 6000D which is a server-class processor, utilizing conventional GDDR7 memory instead of more advanced high-bandwidth memory (HBM), and will forgo Taiwan Semiconductor Manufacturing Co.’s (TSMC) advanced Chip-on-Wafer-on-Substrate (CoWoS) packaging technology, Jefferies estimated that the new regulations ranged the memory bandwidth at 1.7-1.8 terabytes per second, a limit the new GPU is expected to meet, compared to the H2O’s 4 TB/s.

This is reportedly the third time NVIDIA has had to construct a GPU specifically for the Chinese market, influenced by U.S. restrictions. The initial impact of U.S. export curbs was felt before 2022, which led to the development of earlier tailored chips for the region. China is still a significant market for NVIDIA, as it accounted for 13% of its sales in the last fiscal year. However, NVIDIA’s market share in China has steeply declined from 95% before 2022 to 50% currently, fronting growing competition from Huawei’s Ascend 910B chip. These consecutive adjustments make evident the continuous challenges that NVIDIA faced in maintaining its current market position while adapting to constantly changing international trade policies.

CEO Jensen Huang’s Warning & the Future

NVIDIA CEO Jensen Huang has conveyed the company’s challenges, keeping in mind that their older Hopper architecture (used in the H20) can no longer be refined to comply with current U.S. export rules. He also warned that the continued boundaries could push more Chinese customers toward solutions. The ban on the H20 chip forced NVIDIA to write off $5.5 billion in inventory and forbear an estimated $15 billion in sales. I believe this warning has the potential to be a strategic concern for NVIDIA because if this continues much longer, the position of domestic competitors could become more prominent.

NVIDIA is reportedly also developing another Blackwell-architecture chip for China, with production potentially starting in September. While NVIDIA acknowledges that it is evaluating “limited” options, the company states that it is “effectively foreclosed from China’s $50 billion data centre market” without U.S. government approval for new product designs. The long-term implications of these export controls continue to tailor global technology supply chains and regional market dynamics, which pushes companies like NVIDIA to constantly adapt their strategies.