Nvidia Develops New AI Chip for China Amid Revenue-Sharing Agreement

Nvidia is developing a new artificial intelligence (AI) chip, tentatively named the B30A, specifically for the Chinese market. This chip is based on Nvidia's latest Blackwell architecture and is designed to offer approximately half the computing power of the company's flagship B300 chip. The B30A will feature a single-die design, high-bandwidth memory, and NVLink technology for efficient data transmission. Nvidia plans to provide samples to Chinese clients as early as next month.

This development follows a recent agreement between Nvidia, Advanced Micro Devices (AMD), and the U.S. government, allowing the companies to sell specific AI chips to China. Under this arrangement, Nvidia and AMD will pay the U.S. government 15% of the revenue generated from these sales. The agreement permits the export of chips like Nvidia's H20 and AMD's MI308, which are designed to comply with U.S. export restrictions.

The revenue-sharing deal has sparked debate among industry analysts and policymakers. Some view it as a pragmatic approach to balance economic interests with national security concerns, while others argue it sets a concerning precedent by monetizing export controls. Legal experts have also raised questions about the constitutionality of the arrangement, suggesting it may function as an export tax, which is prohibited by the U.S. Constitution.

Nvidia's spokesperson stated, "We evaluate a variety of products for our roadmap, so that we can be prepared to compete to the extent that governments allow. Everything we offer is with the full approval of the applicable authorities and designed solely for beneficial commercial use."

As Nvidia continues to navigate the complexities of international trade and regulatory compliance, the development of the B30A chip represents a strategic effort to maintain its presence in the Chinese market while adhering to U.S. export regulations.