China has prohibited the import of NVIDIA's RTX 5090D V2 while Jensen Huang was present in Beijing.

      The prohibition on the China-exclusive Blackwell card occurred during the Trump delegation's state visit, which saw Nvidia's CEO joining late. Chinese AI purchasers had been utilizing the 5090D V2 as an alternative amid the gap in procuring the H200. On May 15, China ceased issuing import permits for Nvidia’s RTX 5090D V2 gaming card, coinciding with Jensen Huang's presence in Beijing as part of Donald Trump's delegation, according to the Financial Times.

      This ban affects the Blackwell-architecture card that NVIDIA launched last August specifically to adhere to U.S. export regulations. The 5090D V2 was promoted in China as suitable for gamers and 3D animators, but in reality, it was being used by Chinese AI buyers who were cut off from the more advanced H100, H200, and Blackwell-class data-center versions as an alternative, since it maintained the Blackwell architecture and could be utilized effectively for training and inference tasks outside the direct export-control framework.

      Huang's attendance in Beijing during the ban was a late addition to the Trump delegation; he joined the trip on May 13 after receiving a phone call from the president, who was in Alaska at the time. The NVIDIA CEO was present alongside Tim Cook, Elon Musk, and other American tech leaders during the formal state visit program.

      The broader directive regarding domestic chip procurement in China, within which this ban falls, has been escalating throughout spring. Beijing has instructed its major AI firms to halt the acquisition of NVIDIA processors, including the H20 and the RTX Pro 6000D, on the grounds that Huawei’s Ascend line and Cambricon’s Siyuan accelerators now offer comparable performance for relevant tasks.

      Alibaba’s launch of the T-Head Zhenwu M890 aligns with the same procurement directive expressed in the chip-design arena, with company executives asserting ‘scaled mass production’ capabilities for their domestic alternative. The narrative from the U.S. perspective is more pointed. Trump noted earlier this month that China is obstructing H200 purchases despite the U.S. having sanctioned the necessary export licenses. This framing suggests Beijing's procurement policy actively hampers Nvidia’s revenue in China, contrary to the U.S. export controls, representing an atypical diplomatic stance for a U.S. administration to adopt publicly.

      The Trump-Xi summit in Beijing focused on AI regulations while leaving procurement and licensing issues on the agenda. The 5090D V2 ban marks the first clear enforcement action taken by Beijing since the closing of the summit.

      Regarding commercial implications, NVIDIA projected $91 billion in revenue for Q2, exceeding the $86.84 billion consensus. In their remarks, they characterized the China revenue stream as ‘small but material,’ while the overall demand from data centers is described as still in the early stages. Based on the publicly available economic data, the 5090D V2 constitutes a low single-digit percentage of Nvidia’s quarterly revenue. However, its signaling value is more significant. The RTX 5090, the global variant, has now reached $5,300 on the Korean grey market due to procurement pressure-driven demand.

      NVIDIA has not publicly commented on the 5090D V2 ban beyond confirming Huang's role in the delegation. Moreover, China's customs administration has yet to provide a formal written notice detailing the permit decision, and the underlying rationale—whether it is due to national-security reviews, anti-competitive framing, or simple procurement policy enforcement—remains undisclosed. The forthcoming indicator will be the Q2 earnings report concerning China-specific revenue, set to be included in Nvidia’s August reporting cycle, where the impact of the V2 ban on the China revenue stream will be formally unveiled.