An analysis has emerged in the United States suggesting that China is leading in fundamental research to support next-generation computers. If these studies are commercialized, there are concerns that the export controls implemented by the United States to maintain its advantage in semiconductor design and production could become ineffective.
Researchers from Georgetown University's Emerging Technology Observatory (ETO) released their findings on 3rd (local time), analyzing semiconductor design and manufacturing-related papers published worldwide from 2018 to 2023. The research team broadened its scope to include new semiconductor architectures (design methods) as well as existing computer chips and artificial intelligence (AI) optimized graphics processing units (GPUs), analyzing research trends using AI machine learning algorithms.
Papers that included researchers in China numbered 160,852, more than double that of the United States, which had 71,688. Following them were India (39,709), Japan (34,401), and South Korea (28,345). Notably, the increase rate of China's semiconductor-related papers from 2018 to 2023 was 41%, significantly higher than India (26%), the United States (17%), and South Korea (6%).
◇Number of papers double that of the U.S., influence also superior
China has also stood out in research influence, not just in the number of papers. Among the top 10% of the most cited papers, those authored by Chinese researchers numbered 23,520, nearly half of the total. The United States (10,300), South Korea (3,920), Germany (2,716), and India (2,706) followed.
Additionally, among the top 10 institutions conducting the most semiconductor research from 2018 to 2023, nine were Chinese research institutions. The Georgetown University ETO research team noted, "This analysis focused only on 472,819 papers with English abstracts. If papers written in Chinese are included, the proportion of Chinese researchers is likely to be even higher."
Zachary Arnold, a senior analyst at ETO, forecasted in the international journal Nature that, considering the active research being conducted in China, "China's technology and manufacturing capabilities will ultimately improve significantly."
Analyses suggesting that South Korea's semiconductor technology lags behind China's have also emerged domestically. According to a report released last month by the Korea Institute of Science and Technology Evaluation and Planning (KISTEP), a survey of 39 domestic experts showed that South Korea's basic semiconductor technology capabilities were assessed to be lower than China's in all areas.
The report indicated that, based on the criteria of the leading technology country (100%), South Korea's high-density and resistance-based memory technology (90.9%) and high-performance, low-power AI semiconductor technology (84.1%) were both evaluated lower than China's (94.1% and 88.3% respectively). Gaps were also confirmed in power semiconductors (South Korea 67.5%, China 79.8%) and next-generation high-performance sensing technology (South Korea 81.3%, China 83.9%).
◇“If Chinese research is commercialized, U.S. controls will be difficult”
Since October 2022, the U.S. government has restricted exports of advanced semiconductors and manufacturing equipment to China, citing that China is using AI for national surveillance and military modernization. However, it remains uncertain whether these sanctions will effectively restrain the growth of China's semiconductor industry, as China shifts its focus from existing semiconductor manufacturing technologies to new architecture research.
A prime example is the research of Chen Yunji, a professor at the Beijing National Processor Research Institute. Nature noted that "The computing architecture for deep learning designed by Professor Chen's team has been cited over 10,000 times on Google Scholar, and 41% of those citations were by U.S. researchers." Experts predict that if China's new research is commercialized, the semiconductor industry will develop to a point where U.S. regulations become ineffective.
The ETO research team analyzed that new fields like neuromorphic computing and optical computing are the key growth areas in China's semiconductor research. Neuromorphic computing mimics human neural cells (neurons) and can enhance the operational efficiency of AI. Optical computing transmits information using light instead of electrons, making it faster and more energy-efficient than existing semiconductor technologies. Since these technologies require different approaches from current semiconductor manufacturing methods, even if the U.S. imposes export controls on specific equipment and technologies, it is highly likely that China will forge its own path.
Jacob Feldgoise, an analyst at Georgetown University's ETO, stated, "The technologies that China is focusing on for research do not mostly require existing manufacturing technologies that are monopolized by the United States, therefore, when commercialized, it will be difficult for the U.S. to impose controls." He added, "If China successfully commercializes next-generation semiconductor technologies, it will not merely catch up with the United States but may actually surpass it."
References
Emerging Technology Observatory (ETO), https://www.eto.tech/blog/state-global-chip-research/
Nature (2025), DOI: https://doi.org/10.1038/d41586-025-00666-3