Nvidia is developing a new artificial-intelligence processor for the Chinese market, according to people briefed on the plan. The chip, tentatively called the B30A, is based on the company’s latest Blackwell architecture and is designed to deliver roughly half the raw computing power of the dual-die B300 accelerator while outperforming the Hopper-based H20 model that U.S. regulators currently allow the company to sell in China. The single-die B30A will incorporate high-bandwidth memory and Nvidia’s NVLink interconnect to speed data transfers between processors. Nvidia hopes to send engineering samples to Chinese customers for testing as early as September, the sources said, though detailed specifications have not been finalised. Any large-scale shipments would still need clearance from the U.S. Commerce Department amid tighter rules on advanced chip exports. China accounted for about 13% of Nvidia’s revenue last fiscal year, making the market strategically important as the company seeks to fend off domestic rivals such as Huawei. Separately, Nvidia is preparing a lower-specification Blackwell-based RTX6000D card aimed at AI inference tasks that is also designed to stay within Washington’s performance thresholds.