Bessent sees sparse H200 orders as Nvidia faces barriers on both sides of China trade
Treasury Secretary Scott Bessent expects no change to the US policy permitting limited H200 sales to China despite sparse orders. Nvidia’s earlier filing shows Chinese restrictions and tariff costs also obstruct sales, while the company warns that exclusion helps competitors.
- Bessent expects the H200 sales policy to continue despite sparse Chinese orders.
- Nvidia’s earlier filing disclosed Chinese sales restrictions, a $400 million H200 charge and tariff costs it could not pass on.
- Weak licensed sales do not establish how far Chinese chips have advanced or which barrier is decisive.
US Treasury Secretary Scott Bessent, a former hedge fund chief who founded Key Square Capital Management, said on November 20 that Chinese orders for Nvidia’s H200 artificial intelligence chips were sparse and that he expected no change to Washington’s policy allowing their sale under restrictions.
Speaking to CNBC after accompanying President Donald Trump to Asia, Bessent linked weak orders to China’s efforts to develop its own chips. But Nvidia’s earlier disclosures show that Chinese purchasing restrictions and US licensing costs also obstructed sales. Permission from Washington has not assured access to China’s data-center market.

Nvidia’s HGX H200 platform, pictured in its November 2023 announcement. Image: Nvidia. Source: Nvidia H200 announcement.
What Washington is allowing
Bessent said Trump had assigned him to manage the US-China economic relationship. His Treasury appointment record describes responsibilities spanning economic policy, government finances and national security; his investment career included leading the macro-focused Key Square hedge fund.
In the interview transcript, Bessent described the H200 as the upper threshold for chips that could be sold to China. He said he would not anticipate a policy change and attributed sparse orders to Chinese efforts to develop their own chip technology, which he assessed as probably many years behind.
He supplied no order count, value or measurement period, and no technical benchmark for that assessment. His remarks establish the administration’s position, without measuring Chinese substitutes’ capabilities or explaining how much of the order weakness they account for.
The H200 is a graphics processing unit, or GPU, designed for generative AI and high-performance computing in data centers. Nvidia, the US computing company that expanded from PC graphics into AI infrastructure, . Its announcement described larger, faster memory and H200 server boards compatible with H100 hardware and software.
