Nvidia has told data-centre partners to cut their exposure to Chinese customers, the Financial Times reported on Tuesday, in a move aimed at the cloud-rental route Chinese AI developers use to reach advanced chips.
According to the report, cloud clients must now direct at least 51 per cent of their services to non-Chinese companies. Nvidia had set no such limit before. The FT cited five people with knowledge of the matter, who said the instructions were passed on verbally through Nvidia’s sales force and regional teams rather than in writing.
The rule targets overseas cloud arrangements, where Chinese firms lease computing capacity to train AI models without importing the hardware themselves. Washington has been trying to close that gap as part of wider limits on Chinese access to advanced U.S. chips.
The report lands as enforcement pressure on the chip supply chain builds. Several senior employees at Super Micro, which assembles Nvidia-based servers, have been charged with allegedly smuggling about US$2.5 billion of Nvidia AI servers to China earlier this year.
The story carried no comment from Nvidia, nor does it name the chips or server models covered. Because the guidance is reportedly verbal, how strictly it will be applied is not yet clear.
Source: Financial Times via Investing.com; Bloomberg.
