Advanced Micro Devices (AMD) CEO Lisa Su took a notably understated approach during her visit to China last week, in contrast to the high-profile public appearances of Nvidia CEO Jensen Huang. According to Reuters, Su’s visit included a developer event in Shanghai on May 19 and a meeting with Chinese Vice Premier He Lifeng, who encouraged AMD to deepen its cooperation with China. Beijing did not grant Huang a comparable senior-level government meeting during his recent visit, a notable diplomatic signal.
The differing styles of the two executives illuminate the complex geopolitical landscape surrounding the world’s second-largest artificial intelligence hardware market. While Huang has been vocal about the risks of U.S. export controls handing China’s AI market to domestic rivals like Huawei, Su’s quieter diplomacy appears tailored to navigate the politicization of U.S.-China commercial ties without attracting unwanted attention from Washington. Both Su and Huang are from Taiwan and are distant relatives, a personal connection that has not gone unnoticed in media coverage of the two companies’ China strategies.
(Related: Nvidia Warns US Chip Curbs Are Handing China’s AI Advantage to Huawei in Global Markets)
AMD’s Push for a Foothold in China’s AI Chip Market
AMD currently holds a 4% share of China’s AI chip market, according to IDC data cited by Reuters. In contrast, Nvidia’s market share in China has effectively fallen to zero due to stringent U.S. export controls, a dramatic reversal from the dominant position it held just a year ago. Huang has previously estimated that China’s AI chip market would be worth $50 billion this year, making it one of the most valuable battlegrounds in the global semiconductor industry.
During her visit, Su emphasized AMD’s commitment to its Chinese customers, noting that China accounts for approximately 20% of the company’s revenue. “Frankly, you look at the size of the market and the size of our portfolio, and we’ll continue to partner very closely with our Chinese customers,” she stated. AMD’s current China-eligible product lineup includes the MI300X accelerator, which has found traction with several domestic cloud providers seeking alternatives to Nvidia hardware subject to restrictions.
The Software Ecosystem Challenge
At the Shanghai developer event, Su promoted ROCm, AMD’s open-source software stack, as an alternative to Nvidia’s dominant CUDA ecosystem. As Chinese AI firms seek alternatives to restricted Nvidia hardware, AMD sees a significant opportunity to expand its footprint. However, closing the software gap remains AMD’s most formidable challenge in the market.
A source familiar with AMD’s operations in China told Reuters that, while the company sold a significant number of AI chips to Alibaba last year, the Chinese tech giant had to dedicate substantial engineering resources to debugging and adapting the hardware because AMD’s software ecosystem was less mature than Nvidia’s CUDA. This friction has slowed adoption and tempered enthusiasm among some potential customers who have grown accustomed to CUDA’s deep integration with popular AI frameworks.
The challenge is not purely technical. Nvidia has spent over a decade cultivating a developer community around CUDA, creating a switching cost that AMD must overcome through sustained investment in tooling, documentation, and developer support. Su’s Shanghai event was a direct attempt to accelerate that effort, positioning ROCm as a credible, open alternative that Chinese developers can adopt without dependence on a U.S. proprietary platform.
(Related: China’s Domestic Chipmakers Seize 41% of the Local AI Market as Nvidia’s Grip Loosens)
Navigating the Geopolitical Tightrope
AMD’s low-key approach reflects a broader strategic calculation. Unlike Nvidia, which has been outspoken about the economic damage caused by export controls, AMD has largely avoided public criticism of U.S. policy. This restraint may be deliberate: by maintaining a lower profile, AMD reduces the risk of triggering additional regulatory scrutiny from Washington while still pursuing commercial opportunities in China.
The contrast between Su and Huang also reflects the different positions of their respective companies. Nvidia, with its dominant market share and brand recognition, can afford to make bold public statements. AMD, still building its China presence, benefits more from quiet, relationship-driven diplomacy. The meeting with Vice Premier He Lifeng, a senior official who oversees economic policy, signals that Beijing views AMD as a viable long-term partner worth cultivating, even as it accelerates investment in domestic alternatives like Huawei’s Ascend series.
As both AMD and Nvidia navigate the complexities of the Chinese market, their divergent strategies underscore the high stakes of securing a foothold in one of the world’s most critical AI battlegrounds. For AMD, the opportunity is real but the window may be narrow, domestic Chinese chipmakers are advancing rapidly, and the government’s push for self-sufficiency could erode the market for all foreign suppliers within a few years.