What is the current level of China's advanced chip research and development?

Background introduction

Semiconductors are a basic technology that is indispensable to all aspects of modern society. Their applications go far beyond typical information and communication technologies such as smartphones, laptops, and cloud infrastructures, such as hospitals and automobile manufacturing. Both commercial and power companies rely on increasingly complex chips.

The global semiconductor value chain that produces these chips has received extensive attention from policymakers and the media. It is the core of technology competition between the United States and China. Many countries and regions, including the European Union, are striving to strengthen their semiconductor industries in order to reduce exposure to foreign countries. Reliance on technology suppliers.

However, an often overlooked aspect of the semiconductor industry is the amount of research and development (R&D) required to advance cutting-edge technology. The chip industry is one of the industries with the highest R&D profit margins of all industries-semiconductor companies can easily spend more than 18% of their average revenue on R&D. In addition, most of the research and development is done by a few countries, which are the center of this research work.

Why focus on R&D? The recent policy initiatives of the United States, Europe, and South Korea are very focused on the semiconductor manufacturing industry-how to best subsidize fabs and increase domestic wafer production capacity. In the context of the soaring investment costs of modern fabs, such initiatives are Understandable.

However, the semiconductor industry also faces many technical challenges in terms of energy efficiency, sustainability, and new materials. Therefore, the research and development of all process links are crucial. From the perspective of geopolitics and geoeconomics, the production location of chips may be important, but who will develop, define and shape our future chips is equally important.

In the first attempt to assess the semiconductor R&D capabilities of different countries and regions, SNV’s Data Science Department collaborated with the “Technology and Geopolitical Program” to jointly analyze the contributions of three major semiconductor academic conferences:

The International Electronic Devices Conference (IEDM) started in 1955 and covers the entire semiconductor value chain of international research from semiconductor and electronic device technology to technological innovation in design, manufacturing, physics, and modeling.

The International Solid-State Circuits Conference (ISSCC) began in 1954 and brought together top experts in the field of solid-state circuits and system-on-chips, covering scientific achievements in different stages of the semiconductor design process.

The VLSI Technology and Circuit Symposium started in 1981. It connects two international multi-stakeholder conferences on semiconductor technology and circuits, aiming to create synergies on topics of common interest ranging from process technology to system-on-chip. The focus of the VLSI seminar is to study the manufacturing and design of very large-scale integrated circuits (VLSI).

Here are some key insights derived from the quantitative analysis, and we invite you to explore these data yourself in the interactive chart at the end of the report. To learn more about our method, its limitations, and its challenges, please check the FAQ at the bottom. To learn more about SNV's data science department, please feel free to contact Pegah Maham. If you are interested in SNV's work in semiconductors and geopolitics, please contact Jan-Peter Kleinhans.

analysis

Analysis conclusion 1: In the past 25 years, the United States and Japan have developed future chips

It can be seen that, for example, an organization or university in Taiwan (cooperation) has written 1201 research papers in all three conferences (IEDM, ISSCC, VLSI) in the past 25 years, that is, these 1201 papers In each paper, at least one author is from an organization or university in Taiwan, China.

This means: First of all, it is difficult to overestimate the importance of the US and Japanese organizations and universities to semiconductor R&D. In the past 25 years, the combined contributions of these two countries have surpassed that of the rest of the world combined (10.338 vs. 8.187). Although the research forces of various countries have been constantly shifting over time, especially in Japan, their overall importance and contribution to semiconductor research and development are enormous.

Second, only a few regions have developed future chips. The United States, Japan, Europe, South Korea, Taiwan, and China are not only the most important regions in the semiconductor value chain but also the most important regions for semiconductor R&D so far. In the past 25 years, papers from five major countries and regions, the United States, Japan, South Korea, Taiwan, and Belgium, accounted for 75% of all papers.

Analysis conclusion 2: The United States and the European Union have always had high research capabilities

As you can see: The above chart shows the percentage of paper contributions from the United States and the 28 EU countries (including the United Kingdom) each year. For example, in 2005, the European Union (co-) wrote 21% of conference papers, and the United States (co-) wrote 40% of conference papers.

This means: First of all, the United States is a major semiconductor research and development country on average, more than 40% of conference papers are written or co-authored by American organizations or institutions. In addition, the United States has been able to maintain such a high level of R&D contribution in the past 25 years, partly because of the large-scale chip design industry in the United States.

In the past 30 years, American semiconductor companies have controlled about 50% of the global chip market (sales). Since the semiconductor industry is one of the industries with the highest R&D profitability—the company invests about 18% of its revenue in R&D—higher revenue directly translates into more R&D power.

Second, the European Union’s contribution to papers has almost doubled in the past 25 years (1995: 13%, 2020: 25%). In the past 30 years, European companies have accounted for a small market share in global chip sales, only Less than 10%.

Comparing the research power of a region with its market share can reflect the ability of innovations to transform into inventions. Compared with research capacity, the market share is significantly reduced, which indicates that the European ecosystem has potential obstacles and inefficiencies in obtaining value from R&D.

Analysis conclusion 3: In-depth analysis of Europe-member states with leading RTOs to contribute the most

It can be seen that the above chart shows the relative share of the EU member states' annual paper investment. For example, in 2020, Belgian organizations and universities (cooperatively) wrote 8.7% of conference papers, while France contributed 6.1% of conference papers.

The cumulative contribution of each member state in this chart is higher than the contribution of the EU in the previous chart, because France, Belgium, and Germany may co-author the same paper, which makes each country have 1 contribution, but the paper contributes to the EU Say there is only 1 contribution.

This means: First of all, the vast majority of the EU's R&D power comes from a few member states—Belgium, France, Germany, the Netherlands, Italy, and the United Kingdom—in the past 25 years, these countries have accounted for more than 80% of EU paper contributions. In 2020, Belgium and France alone accounted for more than half of the total number of EU papers.

Secondly, the member states (Belgium, France, and Germany) with the most paper submissions also have important semiconductor industry research and technology organizations (RTO): imec (Belgium), CEA Leti (France), and Fraunhofer (Germany). Over the past 25 years, Belgium’s contribution to papers has increased significantly because imec is at the forefront of process upgrades and has worked closely with companies such as TSMC, Samsung, and Intel to produce smaller transistors.

Analysis conclusion 4: The R&D capabilities of China, South Korea, and Taiwan have improved significantly

It can be seen: the proportion of paper submissions in China, South Korea, and Taiwan each year. For example, in all conference papers submitted in 2014, mainland China contributed 5%, South Korea 8.8%, and Taiwan China accounted for 13%.

this means:

First, China, South Korea, and Taiwan are playing an increasingly important role in semiconductor R&D-these three countries and regions accounted for more than one-third of conference papers in 2020. They are no longer just the manufacturing centers of the semiconductor industry chain but are deeply embedded in the development of future chips.

Second, in particular, Mainland China has significantly improved its R&D capabilities in the past decade, which is related to the rise of globally competitive Chinese semiconductor companies such as Huawei, SMIC, and Goodix. Although China is highly dependent on foreign technology providers throughout the semiconductor industry chain, the contribution of its original research papers is already higher than that of Belgium.

Third, since South Korea and Taiwan are the most important countries and regions for cutting-edge wafer manufacturing, their companies (TSMC in Taiwan, Samsung in South Korea, and SK-Hynix in South Korea) and research institutions (ITRI in Taiwan, China) makes sense to also engage in a lot of research and development.

Analysis conclusion 5: Development in the opposite direction-Japan’s share has fallen, while China’s share has increased

As you can see: The above chart shows the relative share of contributions from Chinese and Japanese organizations and universities in all three conferences each year. In 2015, Japan accounted for 18% of all conference papers, while China accounted for 3.8%.

This means: First of all, Japan's scientific research strength has fallen sharply in the past 25 years, from nearly 40% in 1995 to less than 10% in 2020. The same is true for Japan’s overall share in the semiconductor value chain: in 1990, six of the top ten semiconductor companies (sales) were Japanese companies, and in 2020, no Japanese company was among the top ten.

Second, on the contrary, China's development is particularly impressive. Just 10 years ago, they had no interest at all. It was not until the last five years that China successfully approached Japan and eventually surpassed Japan in 2020. As a latecomer and the fastest-growing region in the past 25 years, China's share has more than doubled from 4% in 2015 to 10% in 2020.

Analysis conclusion 6: The EU’s most important research partner is the United States, and China surpasses Japan to become its second-largest research partner

As you can see: The above chart shows the number of cooperation between the EU and foreign countries. A hypothetical joint paper jointly written by imec (Belgium), Intel (USA), Sony (Japan), and JCET (China) will be regarded as a collaboration of these countries. Here, we have chosen the EU's research cooperation with China, Japan, and the United States. The chart below shows the total number of paper contributions from the European Union. In 2020, the EU will publish 11 papers in cooperation with Chinese institutions or universities (7% of all 164 EU papers), and 4 papers in cooperation with Japan (2% of all EU papers), about one-fifth of the papers (34 articles in total) are in cooperation with the United States.

This means: First of all, for Europe, by far the most important R&D partner is the United States. In the past five years, at least 15% of the EU's papers were research and development cooperation with the United States.

Second, it is interesting that China ranks second since 2016, Europe has co-authored more papers with China than with Japan.

Analysis conclusion 7: China cooperates with well-known R&D partners

As you can see: Here, we give examples of the paper cooperation between China and the European Union, Taiwan, and the United States. In 2005, China submitted 3 conference papers, one of which was a collaboration with Taiwan, and the other was a collaboration with the United States.

This means: First of all, as a fast follower of the semiconductor industry, China has focused on research cooperation from the very beginning. China’s most important R&D partners are the United States and Europe: nearly half of China’s conference papers are co-authored with Europe or the United States.

Second, the smaller chart at the bottom illustrates the speed and magnitude of China's R&D participation rate growth since 2010.

in conclusion

Who is developing our future chips? The answer to this question has begun to change in the past 25 years, and even more so in the past 10 years. First, although the United States is and will continue to be a major semiconductor research country, Asian countries, and regions—especially China, South Korea, and Taiwan—are playing an increasingly important role in semiconductor research and development.

In a value chain that relies heavily on research and development, this should not come as a surprise: Currently, only Taiwan and South Korea have the most advanced manufacturing capabilities of 7nm and below in the world.

In order to improve cutting-edge technology and develop future manufacturing processes, companies such as TSMC (Taiwan, China) and Samsung (South Korea) have invested heavily in their research and development work-usually in cooperation with European RTOs or American chip design companies. Therefore, China, South Korea, and Taiwan have not only become chip manufacturing centers, but also important research partners.

Second, scientific research collaboration is playing an increasingly important role. Although the current policy debates on semiconductors in Europe, the United States, and China are mainly about "technical sovereignty", "economic security" and "self-reliance," international cooperation in the semiconductor research community is also greatly enhanced: in 1995, only 11% of the papers were published in the United States. It is co-authored with foreign research partners, and in 2020, about 36% of the papers in the United States are based on international cooperation.

Today, to keep up with Moore's Law, the number of researchers is 18 times that of the 1970s-strengthening international research cooperation is the only way to overcome the technical challenges facing the semiconductor industry in the future. Therefore, it is best for policymakers to encourage international cooperation in the field of semiconductor research.

Third, the decrease in market share seems to be related to the decline in R&D capabilities: Japan’s global chip sales fell from 49% in 1990 to 6% in 2020, and its research contribution dropped from 40% in 1995 to 2020. Of less than 10%.

Whether these developments are simply going on in parallel or one leads to the other, we cannot get the answer through data analysis. But for Europe, there may be some lessons (and warnings). Europe’s R&D capabilities are much higher than its market share, partly because of the very successful RTOs such as imec, CEA-Leti, and Fraunhofer.