U.S.-China High Tech Friction and Japan’s Response: Implications from comparison of innovation processes in Japan, the United States, and China

1. U.S.-China Friction in the High-Tech Sector and Japan’s Response

Competition for technological supremacy reminiscent of the U.S.-Soviet conflict during the Cold War era is being fought between the United States and China in the cutting-edge technology fields of the 21st century. While the United States is strengthening export restrictions against Chinese companies and is calling for stronger cooperation from allied countries, China is developing domestic technologies at a rapid pace. The China bashing by the United States evokes the memories of the Japan bashing in the 1980s, which came against the backdrop of accusations of Japan getting “a free ride on American technology” and the perception of Japan as being “different.” However, the U.S.-China friction is distinctive in that it has extended to cutting-edge science and technology fields, as shown by the lists of critical technologies (e.g., AI/machine learning, and quantum computing) that are subject to the export restrictions adopted by the two countries. One underlying factor is that advanced sciences may be converted to military applications and are therefore relevant to national security. More fundamentally, there is an ongoing “co-evolution” of science and innovation, which refers to the ever-decreasing distance between basic science achievements, which have long been considered a non-competitive field, and innovation for their commercialization and monetization. As a result, in addition to the intervention in free trade seen in the U.S.-Japan friction of the 1980s, the U.S.-China friction is also affecting basic research activities at universities and other public institutions.

It is not yet clear what specific policies the second Trump administration will adopt toward China for the high-tech sector, or how Japan and other U.S. allies will be treated, but its stance will likely become tougher. Amid the deepening confrontation between the United States and China, in Japan, the Economic Security Promotion Act was enacted in 2022, and a research and development program that promotes critical technologies (the K Program) was launched in order to counter economic coercion by other countries. Under the K Program, 20 broad-ranging “specified critical technology” fields, which include biotechnology, computing, materials, space, and ocean transportation, have been designated. Although the policy is meant to make research achievements publicly available in principle, a public-private consultation council has been established to implement the program, so that the freedom allowed for researchers may be restricted to some degree depending on the government’s intentions.

The selection of specified critical technologies is mainly determined based on the potential military diversion of technologies. However, it remains unclear at what point in time government intervention could be required and targeting what technologies from the national security perspective. While providing transparency is preferable in terms of predictability in research and development and avoiding research conflicts, it also presents significant national security challenges. As the boundary between civilian and military technologies becomes increasingly ambiguous, undermining research efficiency in cutting-edge fields where significant innovation is expected in the future could affect the fundamentals of economic security, the purpose of which is ensuring strategic autonomy.

Furthermore, as technology export controls are being strengthened in order to prevent technology leakage, trade control screening related to the activities of universities is becoming increasingly strict because the acceptance of foreign researchers and students by universities falls within the category of technology export. Scientific research activity at universities, which previously involved brisk personnel exchanges under the principle of open science, has become subject to constraints.

2. Comparison of Innovation Processes in Japan, United States, and China

In order to consider how national security-related constraints in cutting-edge science and technology fields may affect innovation using the results of research, and therefore the promotion of domestic high-tech industries in the country, and how to determine which specific technologies should be subject to constraints, it is important to deepen the understanding of innovation processes. In other words, it is necessary to quantify all processes, including how research achievements at universities are put into practical use, the relationship between science papers and patents in that process, and the role of companies in commercializing science achievements in the form of products and services. We have researched science-based innovation processes for many years by developing a database linking science papers (scientific results), patents (technologies), and corporate information (Motohashi, Koshiba and Ikeuchi, 2023; Motohashi, Ikeuchi and Yamaguchi, 2024). Here, I will introduce the results of one of the studies, which examined the differences between Japan, the United States, and China in terms of innovation processes (Motohashi and Zhu, 2024).

In this study, we developed a machine learning model that shows the relationship between technology-related texts and product-related texts using company-level patent text data (technology information) and product-related keyword information extracted from corporate website content (product information). We used the model to conduct an international comparison of the processes of conversion from technology to product (innovation processes). Data available from corporate websites include not only product information but various other forms of information, so we applied a machine learning model intended to extract product information from among all present information (Dual Attention Model, Motohashi and Zhu, 2023) to the companies under analysis (around 15,000 companies listed on stock exchanges in each of Japan, the United States, and China).

We made estimations under nine technology-to-product machine learning models, classified by company type (three types: emerging companies, domestic incumbent companies, and multinational companies) and by country (three countries: Japan, the United States, and China) and looked at differences in innovation processes between models classified by country and by company type. The characteristics of technologies and products in firms can be converted from text information into vector information by using document embedding technology. Therefore, machine learning models for innovation processes can be expressed as conversion models in vector spaces. For example, differences between Japanese and U.S. innovation systems can be measured as differences between the product vector of a Japanese company and the virtual vector of Japanese companies’ products obtained by applying a U.S. conversion model to the vector of Japanese companies’ technologies (the vector-to-vector cosine similarity).

Companies were classified into three types—emerging companies, domestic incumbent companies, and multinational companies—because multinational companies are presumed to be relatively immune to the effects of international differences in innovation systems because they operate in various international markets in addition to domestic markets. From the comparisons between Japan and the United States, between Japan and China, and between the United States and China by company type, we found that the international difference in innovation processes was the smallest for multinational companies for all country pair comparisons. We also find that emerging company innovation processes also differ from established (mature) firms as emerging companies are more likely to embrace new technologies and markets. The separate classification of emerging companies was made to clarify the characteristics of emerging companies based on product life cycle theory.

Innovation processes are also affected by the characteristics of the industries to which the companies belong. In order to control international differences in the industrial structure, here, comparisons were made by industry (21 industries), by country, and by company type. Figure 1 shows international differences regarding domestic incumbent companies, while Figure 2 shows differences between emerging companies and domestic incumbent companies in each country. Table 1 shows the industry-by-industry average and standard deviation regarding the results shown in those two figures. It should be noted that Figure 1 (Table 1, left) may be interpreted as indicating international differences in innovation processes and Figure 2 (Table 1, right) may be interpreted as indicating the innovation propensity of each country.

Regarding the international comparison of incumbent companies, the difference between Japan and the United States (43.61 degrees difference) is relatively small compared with the differences between Japan and China (49.52 degrees) and between the United States and China (49.91 degrees). The differences within all pairs of countries in the standard deviation are relatively small. This means that regardless of industry, the difference between Japan and the United States in innovation processes is relatively small compared with the differences between both Japan and the United States and China. As for the difference between emerging and incumbent companies, the indicator measures the degree of difference in innovation as represented by efforts by emerging companies to address new technologies and markets. The United States scored the highest (28.20 degrees) in terms of this indicator, followed by Japan (26.14 degrees), and China (24.98 degrees). Regarding the standard deviation, the deviation across industries is larger in Japan and China than in the United States. This indicates that in the United States emerging companies’ efforts to address new technologies and markets are advanced regardless of industry, while in Japan and China, such efforts are uneven across industries.

3. U.S. and Chinese Controls of Flows of High Technology and Japan’s Response

The United States is trying to control the leakage of high technology through export regulations (ECRA: Export Control Reform Act) and investment regulations (FIRRMA: Foreign Investment Risk Review Modernization Act), mainly targeting China. Meanwhile, China has also enacted an export control law and is strengthening industrial policies to enhance the competitiveness of domestic high-tech industries. In the United States, there are moves to promote decoupling from China focusing on research funding and personnel exchanges in the field of academic research. How will those moves affect the U.S. economy?

Science-based innovation processes can be modeled as in Figure 3. Research results from public research institutions, mainly universities, are typically published as public knowledge in the form of papers. However, recently, there has been a growing tendency to commercialize scientific knowledge through university-derived ventures and industry-academia partnerships (science innovation). Science-based startups like those ventures and partnerships are often marketed by being integrated into the products and services of large companies (business innovation). At that stage, there is a widespread ecosystem model in which central companies (keystones) collaborate with many niche companies. The co-evolution of science and technology and innovation is proceeding, as mentioned at the beginning of this article, because science innovation is becoming more and more vigorous and is playing an increasingly important role in business innovation. Additionally, the growing role of universities and other public institutions in innovation processes has brought economic security issues into academic research.

The results of our international comparison regarding innovation processes suggest that science innovation is active in the United States, led by emerging companies, and that Japan comes in second in terms of the vigorousness of science innovation, while China lags behind in terms of innovation of high originality among emerging companies. Meanwhile, the results show that the United States is dominant in terms of research capability as seen in scientific papers and other data (the capability to expand science frontiers in Figure 3), and its decoupling from China can be regarded as an attempt to prevent China from catching up. However, China, too, has been rapidly enhancing its research capabilities, so in AI and some computer science fields, it is keeping up with or has even overtaken the United States.

On the other hand, Japan lags far behind both the United States and China in quality and quantity of scientific papers (Motohashi, 2022). Therefore, it is necessary for Japan to incorporate cutting-edge international research results into its leading-edge research, in order to aim for world-first research. Adopting a closed approach to information, even if it were possible to do, is likely to further isolate Japan from global research trends and to lower its science and technology expertise relative to other countries, making it difficult to attract leading researchers from around the world. Therefore, intervening in academic research in the name of national security should be avoided in Japan because doing so is likely to stall the entire process described in Figure 3.

On the other hand, it is possible to identify technologies that are relevant to national security, such as those with potential military diversion, at the point when scientific innovation has been realized. First, as a countermeasure at that stage of scientific innovation for innovations with potential national security implications, possible policy measures would be investment restrictions or other measures to prevent leakage. At the business innovation stage, policy support should be provided to make Japanese businesses keystones in the formation of these ecosystems. In any case, measures targeting basic research and trade controls should not be taken in isolation but should be considered in an integrated manner.