Nvidia’s New Chip for China Sparks Geopolitical Tech Tug-of-War

Nvidia is developing a new AI chip for the Chinese market that is based on its latest Blackwell architecture, offering enhanced capabilities compared to the H20 chip currently permitted for sale in China, according to sources. The new chip, provisionally named the B30A, will feature a single-die design, providing approximately half the computing power of the dual-die configuration in the flagship B300 accelerator card. Despite the potential of this chip, regulatory approval from the U.S. government remains uncertain due to concerns over national security and the technological edge it may provide to China. President Donald Trump has suggested that he may allow the sale of a modified version of the Blackwell chip to China, but any regulatory approval is far from guaranteed. The new chip is expected to be delivered to Chinese clients for testing as early as next month. Nvidia has emphasized that all its products comply with the applicable authorities and are designed for beneficial commercial use.

The U.S. government's export restrictions have had a significant impact on China’s access to advanced AI chips, creating a major point of contention in U.S.-Sino trade relations. In July 2025, Nvidia was permitted to resume sales of the H20, a chip specifically designed for the Chinese market following the 2023 export restrictions. However, this authorization was abruptly halted in April 2025. Trump's recent comments suggest a potential relaxation of these restrictions, proposing a deal that would see Nvidia and AMD share 15% of their revenue from sales of advanced chips in China with the U.S. government. This proposal has been met with skepticism from U.S. legislators, who fear that even scaled-down versions of advanced AI chips could erode the U.S. lead in artificial intelligence.

Complicating Nvidia’s efforts to retain its market share in China, there have been recent concerns raised by Chinese state media regarding the security risks associated with U.S. chips. Authorities have advised Chinese tech firms to exercise caution when purchasing the H20 chip, citing potential security issues. Nvidia has consistently denied these allegations, asserting that its chips do not contain backdoors or vulnerabilities. The company is also developing another China-specific chip based on the Blackwell architecture, tailored for AI inference tasks. This chip, currently dubbed the RTX6000D, is expected to be priced lower than the H20, reflecting its reduced specifications and simpler manufacturing requirements. It is designed to comply with U.S. government thresholds, with a memory bandwidth of 1,398 gigabytes per second, just below the 1.4 terabyte threshold set by recent export restrictions. Nvidia is planning to deliver small batches of the RTX6000D to Chinese clients in September 2025.

Chinese AI data centers are encountering difficulties in transitioning from Nvidia’s AI GPUs to Huawei’s products due to software compatibility issues, according to the South China Morning Post. The Chinese government has mandated that publicly funded AI data centers use at least 50% domestic chips to reduce dependence on foreign technology. This mandate, initially introduced in Shanghai and now extended nationwide, is intended to promote the use of domestically produced chips. Huawei’s latest AI chips, designed to replace Nvidia’s hardware, face limitations due to the 7-nanometer manufacturing process, as access to advanced EUV equipment is restricted by U.S. sanctions. Despite this, Huawei’s chips are not yet able to fully replace Nvidia’s in all applications, particularly in the training of new AI models. Nvidia’s CUDA platform remains essential for running these models, making the transition to Huawei’s CANN platform challenging for data centers forced to adopt a significant proportion of domestic chips.

The evolving landscape of U.S. export controls and the geopolitical implications for AI development highlight the complex challenges facing both U.S. and Chinese tech companies. Nvidia’s strategic efforts to develop compliant chips for the Chinese market reflect the company’s recognition of the importance of maintaining its presence in one of the world’s largest AI markets. However, the effectiveness of these strategies will depend on regulatory developments and the broader geopolitical environment. As the U.S. government continues to refine its export control policies, the balance between national security concerns and the need to foster global technological leadership remains a critical issue.

Source: [1] Exclusive: Nvidia working on new AI chip for China that outperforms H20 - sources say (https://www.reuters.com/world/china/nvidia-working-new-ai-chip-china-that-outperforms-h20-sources-say-2025-08-19/) [2] TIMELINE: GPU Export Controls, NVIDIA GPU Bans, & AI GPU Black Market (https://gamersnexus.net/gpus-news/timeline-gpu-export-controls-nvidia-gpu-bans-ai-gpu-black-market) [3] NVIDIA's GPU Software Makes Chinese Efforts To Switch To Domestic AI Chip Use Full of Headaches, Says Report (https://wccftech.com/nvidias-gpu-software-makes-chinese-efforts-to-switch-to-domestic-ai-chip-use-full-of-headaches-says-report/)