China's racing to build its AI ecosystem as U.S. tech curbs bite. Here's how its supply chain stacks up

China is investing heavily in domestic semiconductor alternatives, particularly Huawei, to overcome U.S. restrictions on advanced AI chips; however, significant challenges remain across the entire value chain, from design to manufacturing and memory components. While Huawei's Ascend 910B and upcoming 910C GPUs show progress, they still lag behind Nvidia's offerings, and China's reliance on foreign technology, especially in chip fabrication equipment (ASML) and high-bandwidth memory (SK Hynix, Samsung, Micron), hinders its ability to create a fully self-sufficient AI chip ecosystem despite efforts from companies like SMIC and CXMT.

Analysis

U.S. restrictions on advanced semiconductor sales to China are compelling Beijing to accelerate domestic AI chip development, with Huawei emerging as a central figure in this endeavor. Despite substantial state investment, significant technological and supply chain hurdles persist across the entire semiconductor value chain. In AI chip design, while Nvidia (NVDA) retains its global leadership with its GPUs, U.S. export controls have curtailed sales of its advanced H20 processor to China. Huawei's HiSilicon, with its Ascend 910B and the anticipated Ascend 910C, is demonstrating notable progress, reportedly narrowing the performance gap with Nvidia's export-restricted chips to less than a full generation, or approximately one year for the 910C according to SemiAnalysis. However, this advancement is confined to design, as China faces formidable challenges in fabrication. TSMC (TSM), the world's leading foundry capable of 3-nanometer production, complies with U.S. restrictions, barring orders from blacklisted entities like Huawei. Consequently, Chinese designers rely on domestic foundries such as SMIC, which officially produces 7-nanometer chips and is suspected of 5-nanometer capabilities for specific applications like Huawei's Mate 60 Pro, but lags significantly behind TSMC in advanced GPU mass production and cost-efficiency. This gap is exacerbated by restrictions on critical manufacturing equipment, particularly ASML's (ASML) advanced EUV lithography machines, which are essential for producing cutting-edge chips at scale. SMIC's workaround using less advanced DUV systems for 7nm production reportedly suffers from poor yields and is nearing its technological limits. Furthermore, the AI ecosystem's reliance on high-bandwidth memory (HBM) presents another obstacle, with South Korea's SK Hynix leading the market and, alongside Samsung and Micron (MU), adhering to U.S. export controls. Chinese efforts, such as those by CXMT, are estimated to be three to four years behind global HBM leaders and face equipment procurement challenges. Huawei's Ascend 910C, despite domestic design, reportedly still depends on stockpiled foreign HBM, underscoring China's continued reliance on external technology across multiple critical segments.

Key Decisions for Investors

• Investors should closely monitor the development trajectory of Huawei's Ascend GPUs, particularly the 910C, as its performance relative to Nvidia's export-controlled chips will be a key indicator of China's ability to indigenize advanced AI hardware.
• The persistent technological lead and market dominance of TSMC in advanced fabrication and ASML in EUV lithography, highlighted by China's significant struggles to replicate these capabilities, suggest sustained competitive advantages for these firms, though geopolitical risks affecting their operations warrant ongoing assessment.
• Consider the high execution risk and extended timelines associated with China's efforts to achieve self-sufficiency in the high-end semiconductor sector; while domestic firms like SMIC and CXMT may benefit from state support, their ability to compete globally in advanced AI chips will likely remain constrained by technology access for the medium term.
• Factor in the ongoing U.S.-China tech rivalry and export controls as a persistent source of volatility and supply chain risk for the global semiconductor industry, potentially impacting valuations and growth prospects for companies with significant exposure to either market.