Nvidia Corporation, the U.S.-based leader in artificial intelligence (AI) chip manufacturing, has highlighted growing competitive pressures from Chinese tech giant Huawei, according to recent regulatory filings and public statements. The disclosure follows U.S. government sanctions aimed at restricting China’s access to advanced semiconductor technologies. Despite these measures, Huawei has accelerated the development of its proprietary AI chips, including the Ascend 910B, which industry analysts describe as China’s closest domestic alternative to Nvidia’s A100 GPU. Market data from 2024 indicates Huawei’s AI chip division has captured approximately 6-8% of China’s $7 billion domestic AI chip market, up from 2% in 2022.
The U.S. Commerce Department imposed export restrictions in October 2022, barring American firms from supplying cutting-edge AI chips and manufacturing equipment to Chinese entities. These sanctions were designed to curb China’s technological advancement in critical sectors. However, Chinese companies, including Huawei, have pivoted to domestic innovation. Huawei’s investments in semiconductor R&D surged by 25% year-over-year in 2023, reaching $22 billion, according to its annual financial report. The firm has also collaborated with state-backed foundries like Semiconductor Manufacturing International Corporation (SMIC) to produce 7-nanometer chips, a milestone previously deemed unfeasible under U.S. restrictions.
Nvidia’s latest earnings report underscores the shifting dynamics. While the company retains an 80% share of the global AI chip market, its revenue from China—once representing 20-25% of its data center sales—plummeted by 85% in Q4 2024 due to export controls. In response, Nvidia has introduced modified chips compliant with U.S. regulations, such as the A800 and H800, but these products offer reduced performance metrics. Analysts at Gartner note that these downgraded chips are 30-50% slower in training AI models compared to their predecessors, limiting their appeal to Chinese cloud providers like Alibaba and Tencent, which now prioritize local suppliers.
Huawei’s progress is partly attributed to Beijing’s $150 billion semiconductor subsidy program, launched in 2023, which prioritizes self-sufficiency in chip design and manufacturing. The initiative has enabled Huawei to expand its Shanghai-based chip design team to over 15,000 engineers, focusing on neural processing units (NPUs) and high-bandwidth memory solutions. Tests conducted by China’s Institute of Computing Technology in early 2025 revealed that Huawei’s Ascend 910B delivers 85% of the performance of Nvidia’s A100 in training large language models (LLMs), a significant leap from its earlier iterations.
Global supply chain adaptations have further aided Chinese firms. Between 2023 and 2025, Huawei increased its procurement of etching tools and lithography equipment from Japanese and South Korean suppliers by 40%, circumventing U.S.-dominated vendors like Applied Materials and Lam Research. Meanwhile, SMIC has repurposed older deep ultraviolet (DUV) lithography machines—not restricted by U.S. sanctions—to mass-produce 7nm chips, albeit at higher defect rates and costs. Canalys reports that SMIC’s 7nm yields currently hover around 50%, compared to Taiwan Semiconductor Manufacturing Company’s (TSMC) 90% for similar nodes.
The competitive landscape is reshaping procurement strategies within China. A 2024 survey by TrendForce found that 65% of Chinese tech firms now prioritize sourcing AI chips domestically, up from 35% in 2022. Baidu, for instance, replaced 30% of its Nvidia A100 GPUs with Huawei’s Ascend chips in its data centers last year. Similarly, state-owned telecom operator China Mobile announced a $1.2 billion deal with Huawei in January 2025 to upgrade its AI infrastructure, signaling confidence in local alternatives.
Nvidia’s long-term market dominance faces uncertainties. While its global data center revenue reached $47 billion in 2024—a 120% annual increase—its growth in Asia-Pacific (excluding China) slowed to 15% in the same period, down from 45% in 2023. CEO Jensen Huang acknowledged during a February 2025 earnings call that geopolitical tensions are creating “a parallel ecosystem” in China, though he emphasized Nvidia’s software ecosystems, such as CUDA, remain entrenched in AI development workflows. CUDA’s market share in China’s AI research sector stands at 70%, per IDC data, but open-source alternatives like Huawei’s MindSpore are gaining traction, doubling their adoption rate since 2023.
International regulatory bodies are monitoring the fallout. The European Commission’s 2025 Global Tech Competitiveness Report warns that China could achieve 50% self-sufficiency in AI chips by 2027 if current R&D investment trends persist. Conversely, U.S. semiconductor firms, including Nvidia and Intel, have lobbied Congress to ease certain restrictions, arguing that overly broad sanctions incentivize China to develop independent solutions. A 2024 Boston Consulting Group study estimates that U.S. chipmakers risk losing $54 billion in annual revenue by 2030 if decoupling accelerates.
Market analysts stress that Huawei’s rise does not immediately threaten Nvidia’s global hegemony. The Chinese firm’s AI chips remain absent from markets outside China due to intellectual property concerns and compatibility limitations. However, Huawei’s growing influence in emerging economies is notable. During the 2025 Belt and Road Summit, the company signed AI infrastructure deals with Indonesia, Saudi Arabia, and Egypt, leveraging affordable pricing and government-backed financing.
The U.S. Department of Commerce is reportedly reviewing its sanctions framework to address loopholes, including third-country shipments and “chiplet” technologies that bypass export controls. For now, Nvidia continues adapting its product roadmap, announcing plans to launch a China-specific GPU, the H20, in late 2025. Early specifications suggest the H20 will align with U.S. performance thresholds while optimizing for AI inference tasks, a segment less impacted by Huawei’s current offerings.
Industry stakeholders emphasize collaboration as a counterbalance. In March 2025, Nvidia partnered with Indian conglomerate Reliance Industries to build AI data centers using U.S.-approved chips, diversifying its geographic reliance. Meanwhile, Huawei’s challenges persist, including production scalability and international trust deficits. Its global market share in AI chips remains under 3%, but domestic triumphs underscore a pivotal shift in the semiconductor race—one where sanctions, rather than stifling innovation, have spurred localized resilience.
