The U.S. government has moved to block Nvidia from selling its scaled-down B30A AI chips to China, extending export restrictions to cover even lower-performance models that were previously considered compliant. The decision, first reported by The Information and confirmed by sources including Channel News Asia and Lowyat.net, marks a significant escalation in Washington’s efforts to limit China’s access to advanced AI hardware.
The B30A chip was designed as a scaled-down version of Nvidia’s flagship B300, intended to meet earlier U.S. export rules. However, officials now argue that even these chips retain enough computing power to train large language models when deployed in clusters — a capability deemed too sensitive for Chinese use.
According to The Information, the White House has instructed federal agencies not to approve export licenses for the B30A, effectively shutting Nvidia out of China’s AI datacenter market. Nvidia responded by stating it has “zero share in China’s highly competitive market for datacenter compute” and does not include China in its financial guidance.
This move follows a series of tightening controls on AI and semiconductor exports to China, reflecting growing geopolitical tensions and concerns over national security. Analysts suggest the decision could further strain U.S.-China tech relations and prompt Chinese firms to accelerate development of domestic alternatives.
