Regarding the state of China’s semiconductor industry, propaganda clouds the truth on all sides. In the West, commentary often reflects wishful thinking, unless the commentator tries to make points opposing a government’s stance or policies. Details from China are sparse, deliberately vague, and sometimes inflated to project strength beyond what reality supports. Understanding the actual state of China’s semiconductor capabilities is critical for investors, policymakers, and technologists alike. This report, distilled from hundreds of pages of rigorous research and primary evidence, aims to provide clarity. We present our findings in a series of concise tables, offering a clear view of where China stands, where it’s headed, and the remaining gaps. Clients or prospective clients seeking deeper insights into our methodology or the raw work behind today’s note are welcome to contact us for further details.

While the rest of this note may at times appear dense with technical terms and acronyms, that’s an unfortunate but unavoidable feature of the subject matter. That said, readers unfamiliar with semiconductor jargon should feel no hesitation in skimming over the specifics. The core insights—about where China stands versus global technology frontiers—are clearly laid out and accessible. The patterns, trends, and implications emerge clearly even without mastering every technical detail.

‍A List of Unfounded Perceptions Best Ignored

The reflex in the West to any Chinese semiconductor breakthrough is often a chorus of skepticism: “It’s stolen tech,” “It’s sanction evasion,” or “It’s just government subsidies.” Loose claims, like ASML CEO Christophe Fouquet’s December 2024 assertion that China’s industry is “10 to 15 years behind,” gain traction despite scant evidence, perpetuating a cycle of dismissive narratives. Such statements reflect a broader bias, where Chinese innovation is assumed to be derivative until undeniable proof forces reluctant acceptance.

For instance, DeepSeek’s AI efficiency gains in 2024 were initially met with accusations of intellectual property theft, only acknowledged as legitimate when their performance became indisputable. Similarly, SMIC’s 7nm chips (2022) and Huawei’s Kirin 9000S (2023) faced early denials, with Western analysts citing low yields or copied designs, only to be validated by teardowns showing unique DUV-based processes.

Conversely, it becomes “new to me is new to the world.” Around the end of the month, multiple, high profile outlets wrote similar articles on Huawei’s facilities and Chinese innovation. These reports, while grounded in observation, risk being co-opted by political agendas, with some Western critics of US sanctions using China’s progress to argue against restrictive policies. Both extremes—reflexive dismissal and uncritical praise—distort the truth, underscoring the need for evidence-based analysis over narrative-driven agendas.

‍

The list of meaningful innovations within Semiconductor or AI fields where China is ahead is extremely short. China has found new efficiency measures or workarounds, but it is difficult to come up with a list in these fields where China is ahead.

Semi Manufacturing: China Stays in the Race

When one starts with the fact that semiconductor manufacturing is possible in only a small number of countries, with few able to master more than a handful of domains, one begins to appreciate the all-round and competent nature of China’s semiconductor industry. Through relentless investment and innovation, companies like SMIC, CXMT, and YMTC are pushing boundaries in logic chips, memory, and beyond. The following tables highlight China’s progress in foundry nodes, memory technologies, and associated fields like packaging and EDA tools.

SMIC’s Node Progress vs. TSMC/Samsung (Logic Chips)

Note: SMIC’s lag has shrunk from 4 years (14nm) to ~2 years (7nm), but 5nm/3nm delays reflect EUV sanctions. Other Chinese foundries like Hua Hong and Nexchip are also ramping mature nodes (28nm and 14nm), contributing to growing domestic capacity. Additionally, China’s push in domestic lithography and materials suggests groundwork for more advanced nodes across multiple players by the late 2020s.

China’s Memory Progress vs. Global Leaders

Note: YMTC’s NAND is nearly at parity, but CXMT lags in DRAM yields and HBM development due to sanctions.

China’s Progress in Associated Semiconductor Fields vs. Global Leaders

System Strengths and New Paths: China’s Innovation Model

While China still lags behind in many individual segments of AI and semiconductor innovation—often by 2 to 5 years—it is the breadth of its presence and its ability to derive synergistic advantages across the stack that set it apart. Unlike most competitor systems, which excel in several niches, China maintains a full-spectrum capability, from compute cores and memory to packaging, optics, and industrial deployment. This holistic participation enables system- and product-level innovation that can outpace subsegment limitations.

Moreover, constrained access to bleeding-edge tech has forced Chinese companies to explore unconventional paths, driven by necessity but often revealing new efficiencies.

This pattern shines in Huawei’s latest Ascend 920 announcement (March 2025), where chip-level disadvantages—lower efficiency and speed compared to NVIDIA’s GB200—are overcome by leveraging Huawei’s networking strengths. The Ascend 920’s CloudMatrix 384 system, with 384 chips, delivers 300 PFLOPs BF16, doubling NVIDIA’s 150 PFLOPs, proving system-level competitiveness despite individual chip gaps.

Cloud-Scale Efficiency by Architecture, Not Just Chip

Huawei has essentially outsystemed its disadvantage: where others would focus on transistor performance, it built on telecom-grade networking to unify 384 Ascend chips into a powerful AI machine.

Conclusion: The Innovation Race Heat Is On

This note’s mission is to assess the innovation gap through evidence, and the data speaks clearly: China is closer than many would like to admit or to see.

It is not just about semiconductors: China is within the zipcode of the global cutting-edge leaders across innovation segments and is working relentlessly. While there is an industry denigrating the announcements, the real-life industry leaders, policymakers, and politicians of the rivaling nations have the need to maintain, if not intensify, their efforts not to fall behind. We do not think the path of a sustained lead is through sanctions – they may provide a cushion near-term, but even in fields like the EUV machines or the most cutting-edge fabs, Chinese innovations continue to build.

As we enter a period of economic slowdown and industries in the developed world feel the need to scale back on expansion and innovation plans, competitive pressure will remain firmly in the opposite direction. Every week or month, with another innovation announcement from China, we will have another journal in some country doing a cover piece on the Chinese innovations, even if littered with offhand comments about subsidies, stealing, and them still not being cutting edge. And, every such media attention will have the community’s leaders urging industries to keep working on their innovations to maintain their leads. This is why the arguments on innovation capex slowdown in the current downturn are not as straightforward as economic textbooks suggest.