Sanctions on China Could Allow Huawei to Establish New Global Standards. Nvidia Says the U.S. Is to Blame

Losing access to the Chinese market would be a “tremendous loss” for Nvidia, according to CEO Jensen Huang. The company’s future in the Asian country seems increasingly uncertain. U.S. sanctions are becoming stricter, limiting the range of products Nvidia can sell to its Chinese customers despite its efforts to navigate these restrictions.

A lot of money is at stake. In the last fiscal year, ending on Jan. 26, 2025, China represented around 13% of Nvidia’s total revenue, amounting to around $17 billion. In fact, the Asian nation is Nvidia’s third-largest market, following only the U.S. and Taiwan. There’s also the risk that Huawei could establish global standards that disadvantage Nvidia’s technologies.

CUDA’s Dominance Is at Risk

“We are at an inflection point; the United States needs to decide if it is going to continue to lead the global development and deployment of AI or if we are going to retreat and retrench. America cannot lead by slowing down. If we step back, others will step in. And the global AI ecosystem will fragment–technologically, economically, and ideologically,” Huang told U.S. lawmakers.

The inability to sell its AI GPUs to Chinese customers poses a significant problem for Nvidia. However, the potential loss of CUDA’s current dominance in the AI industry is an even greater concern. CUDA is crucial to Nvidia’s business, given that most AI projects currently in development use it.

CUDA is the compiler and development toolset programmers use to create software for Nvidia GPUs. Replacing it with an alternative in ongoing projects could be an issue. Huawei wants to capture a substantial portion of this market in China with its own alternative, CANN. For now, CUDA remains the dominant player.

More from Xataka On

Nvidia’s AI Chips Are All the Rage. However, Gamers Are Still Complaining About Issues With Its Graphics Cards Months Later

CANN has faced criticism for being reportedly difficult to use and exhibiting unstable performance. Some have also criticized insufficient documentation and reliability. However, underestimating Huawei’s ability to address these shortcomings would be a serious mistake. The company has acknowledged these problems and confirmed that it’s actively working to resolve them. Additionally, Huawei isn’t the only company challenging Nvidia’s leadership position.

CANN isn’t China’s only asset. One example is Moore Threads, a Chinese company dedicated to producing AI hardware. Companies aligned with U.S. interests and its allies can’t sell software or advanced equipment to Moore Threads. Though it was only created in 2020, the company has a significant advantage. Founder Zhang Jianzhong was the former general manager of Nvidia’s subsidiary in China, which gives him valuable expertise.

Moore Threads has developed several GPUs for AI applications that, theoretically, rival advanced solutions from Nvidia, AMD, and Huawei. However, the company also offers a software package to challenge CUDA’s dominance. Its MUSA software is compatible with the MTT range of graphics cards. It includes a compiler, runtime libraries, specialized libraries, and debugging tools. One of its most appealing features for China is its capability to reuse code written in CUDA, allowing it to run on Moore Threads AI cards.

Image | Mariia Shalabaieva

Related | Nvidia Eyes Another Disruptive Technology After the Success of Its AI Chips: Quantum Computers