As the United States deliberates on imposing stricter trade restrictions aimed at preventing advanced chip technology from being exported to China, Nvidia, a prominent chipmaker based in the U.S., is reportedly developing a modified version of its latest artificial intelligence chips to adhere to these regulations.
According to sources familiar with the situation, Nvidia is working on a new product named the “B20,” which is a variant of its Blackwell AI chips intended for the Chinese market. This initiative will involve collaboration with a local distribution partner, Inspur, to facilitate the launch and sale of the B20 in China. The product is anticipated to begin shipping in the second quarter of 2025, although Nvidia has opted not to comment on this information.
The company already has three chips designed to comply with existing U.S. export controls, including the H20. To boost its competitive position against domestic rival Huawei, Nvidia has previously lowered prices on the H20 due to sluggish sales. However, recent reports indicate that sales of the H20 are now increasing. Nvidia is projected to ship over one million H20 chips to China this year, amounting to approximately $12 billion, despite the ongoing U.S. trade restrictions. In comparison, this figure is almost double what Huawei is expected to sell of its Ascend 910B chip.
Analysts from Jefferies have raised concerns that the H20 chips may face additional risks under potential new U.S. trade regulations. While the U.S. is set to conduct its annual review of semiconductor export controls in October, analysts predict that it is highly likely the sale of H20 chips to China could be banned. Such a ban could be enforced in various ways, including a “product-specific ban,” reducing the computing power threshold, or imposing restrictions on memory capacity.
Moreover, there is speculation that the U.S. might extend its export controls on chips to other nations in the region, like Malaysia, Indonesia, and Thailand, or to Chinese firms operating abroad, though implementing such measures would be more complex.