Digital Innovation

Abstract

Innovation in digital industries is one of the key factors to building a country’s competitiveness. With a focus on the semiconductor industry and the software industry, this chapter mainly analyzes the path of innovation of China’s digital economy. Semiconductors are the cornerstone of the global information and communications technology (ICT) industry. The semiconductor industry has a global division of labor, with upstream and downstream players being interdependent and regional cooperation being indispensable. Although Chinese companies mainly focus on downstream intermediate products and end product assembly, we suggest China undertake a range of measures to promote innovation, especially those relating to semiconductor manufacturing. Software is also an important component of the digital economy. Currently, China’s supply of system and high-end application software may not be sufficient, and China relies heavily on imports. This current situation may lead to security risks, including import suspension and data leaks, and also may result in a lack of long-term drivers for industrial upgrades. We believe the security risk of the software value chain is rooted in inadequate technological innovation, and it is important to catch up in innovation in both system and application software.

Regarding the path of innovation of China’s semiconductor industry, we think China can step up efforts in innovation in the following two directions: More Moore’s Law and More than Moore’s Law. The former indicates that China can leverage existing mature technologies and follow the path of leading global companies. China can increase capex on mature process capacity and step up R&D on semiconductor equipment, materials, and electronic design automation (EDA) tools, aiming to eliminate the supply chain bottlenecks and improve the command of core technologies. More than Moore’s Law, on the other hand, suggests that China can highlight catching up with global players more quickly in terms of advanced integrated circuit (IC) processes; deepen its presence in potentially disruptive technologies such as new semiconductor materials, new computing architectures, and chip integration; and increase investment in the wide band-gap semiconductor materials and advanced packaging markets, in which technology advancement is more visible and the gap between China and overseas economies is narrow.

There are four major driving forces behind innovation in the semiconductor industry. Industrial policies have played a crucial role in stimulating technological innovation and driving industrial development. Many countries and regions have unveiled a variety of incentive policies to bolster the semiconductor industry, mostly consisting of fiscal and taxation measures. In addition to industrial policies, capital, talent, and technology are three indispensable elements for the semiconductor innovation.

Regarding the path of innovation of China’s software industry, system software and application software take different forms. System software innovation is an aggressive process led by scientific R&D. Setting standards for system software tends to create monopolies. It promotes innovation and requires long-term input of funds and resources. Innovation of application software happens gradually, and it is rooted in the constant feedback exchanges between users and developers. Small new entrants find it difficult to compete with the low pricing of large players, making application software an ideal target for venture capital.

Therefore, we believe that domestic software companies can innovate in the following ways. Almost all new-generation system software is open source, giving domestic companies a chance to participate in the setting of new standards. In fact, Chinese software developers are already an important part of the international open source community. We suggest that domestic companies increase independent innovation to keep up with global cutting-edge system software technology. In terms of application software, we think standardization is the key to success. Some companies in China are able to define local standards for application software, but only a few can set global standards. We believe opening up more application scenarios and clients is crucial for domestic companies to strengthen the competitiveness of their products.

3.1 Digital Industries in the Global Value Chain

Digital industries are experiencing deepening global division of labor, and their global value chains are becoming increasingly complex. We believe China is moving up along the semiconductor value chain to claim “innovation premium”, but is facing risks, including suspension of oversea supply and data leaks in the software value chain. Technological innovation plays an important role in shaping the current status of China’s digital industries. Innovation enables semiconductor companies to enjoy competitive strength, while insufficient innovation results in security risks in the software industry.

3.1.1 Semiconductor: Global Division of Labor and Cooperation Along the Global Value Chain

The global division of labor along the semiconductor value chain has become common practice with the emergence of economic globalization. China has seized the opportunity to become an important participant in the global semiconductor value chain. After experiencing a rapid growth stage driven by its large market size and abundant supply of engineers, China’s semiconductor industry may transit to an innovation-driven growth stage.

3.1.1.1 Semiconductor Value Chain: The Global Division of Labor and Regional Interdependence

The semiconductor industry presents two major characteristics: Vertical and regional division of labor, and regional clusters.

There are two reasons behind the vertical division of labor. First, companies benefit from economies of scale by expanding production. With the advancement of manufacturing processes and the increase in silicon wafer size, the number of transistors on a chip has increased dramatically, and the yield has improved substantially. Therefore, capacity expansion can lower unit production cost, thereby enhancing competitiveness. Second, companies in the semiconductor industry must commit high levels of sunk cost. Expansion of wafer manufacturing capacity for advanced chips requires far more capital expenditure and R&D spending than before. Aside from some tech giants, most companies are unable to continue to move forward due to lack of sufficient funds.

The semiconductor industry is also undergoing regional division of labor, and each country or region plays a different role, some pivotal, along the value chain (Fig. 3.1). In design tools, three electronic design automation (EDA) tool companies (two in the US and one in Europe) have a combined global market share of 85%. In chip manufacturing, about 75% of production capacity is concentrated in East Asia (Japan, South Korea, the Chinese mainland, and the Taiwan region of China). In manufacturing equipment and materials, the US has a market share of more than 50% in at least five subdivided wafer fab equipment (WFE) markets. Japanese companies have high shares in the semiconductor materials market. European company ASML dominates the extreme ultra violet (EUV) lithography market. In end products markets, the US has a share of more than 90% of the high-end logic chips market.

Fig. 3.1

Source SIA, BCG, CICC Research

Semiconductor value chain and global division of labor.

Overall, while various sections of the semiconductor value chain are relatively scattered, there is a high degree of regional concentration within some of the sections. Therefore, in the event of force majeure such as earthquakes, flooding, and fire, the supply chain can be highly fragile.

3.1.1.2 The Innovation Premium Curve of the Semiconductor Value Chain: An Analysis on Premium from Competitive Strength

We construct an “innovation premium” chart to illustrate and analyze¹ the competitive landscape of the global semiconductor value chain based on the size of the global semiconductor market and industrial value chain, as well as the current positions and innovation capabilities of different countries and regions.

From a static perspective, the semiconductor value chain demonstrates a typical smile curve (Fig. 3.2). The value added in design, equipment, foundry, and packaging & testing is gradually decreasing, while the value added of integrated device manufacturer (IDM) is rising. Given different levels of strength in innovation, product premiums of different segments vary. The design process has high value added given its rapid technological advances and as it is an asset-light business model. The asset-heavy business models of the foundry and packaging & testing segments affect their margins. IDM enjoys high premiums given the high R&D spending and fixed-asset investments.

Fig. 3.2

Source Wind, Bloomberg, IC Insights, Trendforce, Digitimes, China Semiconductor Industry Association, CICC Research

The “innovation premium” curve of the global semiconductor value chain in 2020. Note Size of bubble refers to total revenue of respective country and region in the corresponding markets (based on CICC estimates).

US companies are concentrated at the two ends of the smile curve, and Chinese companies are moving up the value chain to achieve “innovation premium”. The heavy R&D investments in the design, equipment, and IDM segments help US companies maintain their innovation premium and leading positions. Companies from the Chinese mainland and the Taiwan region of China focus on the foundry, packaging, and testing businesses. In 2020, the Chinese mainland’s R&D spending on foundry, packaging and testing was much higher than that of other countries and regions. In addition, high R&D spending on the design segment has brought about considerable returns for companies from the Chinese mainland.

From a dynamic perspective, innovation is an effective way for companies to maintain their competitive strength. By analyzing the changes in the “innovation premium” curve for 2000–2020, we have found that global IDMs have maintained stable gross margins over the past two decades (Fig. 3.3). Fluctuating market demand and intensifying competition have posed a limited impact, mainly thanks to R&D expense ratio rising steadily by 0.25 pct annually. US IDMs maintained a gross margin of over 50% with an R&D expense ratio of 12–15%, while Korean IDMs maintained a gross margin of over 30% with an R&D expense ratio of 3–8%.

Fig. 3.3

Source Wind, Bloomberg, IC Insights, Trendforce, Digitimes, China Semiconductor Industry Association, CICC Research

The “Innovation premium” curve of the IDM industry of major countries and regions in 2000–2020. Note Size of bubble refers to total revenue of respective country and region in the corresponding markets (based on CICC estimates).

3.1.2 Software: China is Facing Security Risks in the Global Value Chain

3.1.2.1 Insufficient Supply of Local Software

China’s digitalization demand exceeds overall software supply. We believe local software supply is unlikely to suffice during the new round of digitization in China. IDC expects IT spending by Chinese enterprises to reach US$700bn in 2021, ranking No. 2 in the world, while Bloomberg and Gartner forecast China’s software industry output value at US$439bn in 2021, revealing a significant supply gap. In contrast, the supply and demand for software is largely balanced in developed countries such as the US (Fig. 3.4).

Fig. 3.4

Source Bloomberg, IDC, Gartner, CICC Research

Software industry output value and IT spending in major regions, by company (2021e).

China’s software industry faces trade deficit. China’s software export value has maintained steady growth since 2013, except for declines due to trade frictions or COVID-19. Export value reached US$191.5bn over January to May in 2021, rising 15.4% YoY, and is around 3% higher than the same period in 2019.² Data from the National Bureau of Statistics (NBS) shows that China’s import value of computers, software, and auxiliary equipment bottomed out at end-2015 before increasing sharply from 2017 onwards. Overall, we think the trade deficit still exists, but is gradually narrowing.

Supply shortage essentially lies in system software. We divide the software and services sector into three parts: System software (operating system, database, middleware, virtualization technology, and cybersecurity); application software (industrial software, management software, and industry application software); and IT services (IT consulting and implementation). By scale, the US is the largest market in all three major subsectors—especially in system software.

There are relatively large gaps in localization of different software subsectors in China. Chinese companies have developed competitive strength in the fields of enterprise resource planning (ERP), construction cost estimation, and healthcare IT, with the import substitution rate measured by the share of software demand met by domestic companies surpassing 50% (Fig. 3.5). However, the system software and industrial software segments are monopolized by overseas industry giants due to the R&D barriers to entry and the mainstream of industry standards set by foreign companies, making breakthrough in system and industrial software a key focus of R&D in China.

Fig. 3.5

Source Bloomberg, CICC Research

Size of software and services subsectors in major regions. Note Based on data from select companies; aggregate revenue of the 2,712 companies we selected from Bloomberg’s software and technological services sector; data as of end-July 2021.

In sum, there is visible supply gap in China’s system and application software industry. Despite a high import substitution rate for some software subsectors, a large proportion of domestic demand for system and application software (e.g., industrial software and ERP) is dependent on imports (Fig. 3.6). We believe the technological innovation-driven enhancement of China’s software technology is important for ensuring the security of the domestic software value chain.

Fig. 3.6

Source IDC, Qianzhan Industrial Research Institute, CCID, corporate filings, CICC Research

Reliance on imports, by software subsector. Note Data as of end-December 2020; size of bubbles indicates the market size; reliance on imports = 1-import substitution rate.

3.1.2.2 Potential Security Risks in China’s Software Value Chain

China’s software supply chain faces the risk of suspended software authorization and services. The bulk of industrial design software (an important production tool) and system software (the foundation for the operation of application software) in China are imported from the US.

There are also concerns over data security. “Backdoor” in the context of software offers access to privileged information bypassing normal authentication. Software developers can modify or test bugs during the development process through a so-called “backdoor,” which is a software vulnerability that may become an entry point for hackers However, if hackers manage to gain access to this backdoor, or developers fail to close it before the software hits the market, it would create risk of data leaks. This is an issue that has been gaining increasing attention across the globe.

Software value chain risk is more worrying in the long term. The US restrictions on tech exports to China are now focused on hardware, mainly because the software shortage can be addressed with alternative solutions such as open source software. However, the software value chain risk will affect China’s digitalization and industrial upgrade over the long term, in our view. The upgrading of software technology is a constant process globally, and contributing to this process is important for China’s ability to narrow its gap with other countries in IT.

3.1.2.3 Software Value Chain Risk Rooted in China’s Lagging Technological Innovation

Gross margin is a key measure of a product’s competitiveness and a company’s innovation capability. Software companies in China tend to have lower gross margins compared to their European and US counterparts, particularly in the system software sector (Fig. 3.7). High barriers to entry for system software also contribute to higher gross margins. In contrast, labor-intensive IT services tend to have lower gross margins. European and US software companies generally have higher gross margins than Asian companies.

Fig. 3.7

Source Bloomberg, CICC Research

Software innovation capability in terms of GM and R&D expense ratio. Note Based on data from select companies; we select 2,712 companies from Bloomberg’s software and technological services sector; data as of July 2021.

R&D expenditure is a good indicator of a company’s innovation efforts. US companies spend more on R&D for system and application software than Chinese companies (Fig. 3.7). Developing system and application software requires more intensive investment than labor-intensive IT services. The US is the top spender on R&D for system and application software, followed by China and Europe. China and South Korea have higher R&D spending on IT services than other countries.

Chinse software companies lag in system and application software. Global markets for system software (operating system and database) and industrial software (computer-aided design [CAD] and EDA) are both dominated by overseas companies, mostly from the US. Chinese software providers are less competitive even in local markets. We believe it is important for China to catch up in these fields, so as to guarantee the security of its software value chain.

Having analyzed China’s position in the global value chain of the semiconductor and software industries, we believe it is crucial to enhance innovation inputs in the semiconductor industry to maintain advantages, and catch up in both system and application software to ensure security. In the following sections, we conduct in-depth analysis on the path of innovation in the semiconductor and software industries. Section 3.2 focuses on how to mobilize policy, capital, talent, and technology resources to promote semiconductor innovation. Section 3.3 provides a framework and targeted solutions for system and application software innovation.

3.2 The Path of Innovation in the Semiconductor Industry

3.2.1 The Dual Perspectives of Semiconductor Innovation

The Chinese mainland’s market for semiconductors is large and focuses on downstream intermediate products and terminal product assembly. We suggest companies from the Chinese mainland step up efforts to develop technologies in areas where they are weak for now, and may continue to innovate to catch up with the global leading players.

Incremental innovation is the general principal driving the development of the semiconductor industry. However, it is undeniable that radical innovation has triggered leapfrog development of some technologies, and moderately ahead-of-the-curve innovation is also necessary. There are two main directions for China’s semiconductor innovation. One is “More Moore’s Law”, referring to advanced processes of the entire value chain. The other is “More than Moore’s Law”, which is developing disruptive technologies in computing principles, materials, devices, computing architecture, and chip integration. This includes boosting investment in semiconductors with new structures and that are made with new materials and advanced packaging fields that have clearer outlook and narrower gap with overseas countries.

In the manufacturing field, China lags behind other countries and regions in advanced nodes and wafer manufacturing capacity. Although China can narrow the gap in wafer production capacity by increasing capital investment, it is unlikely to narrow the gap in the advanced processes in the short term since international competitors continue to innovate in this area. Due to the long development cycle, heavy capital investment, and winner-take-all characteristics in advanced nodes, an oligopoly is likely to be formed in the global market in some segments. Therefore, to narrow the gap with global top tier players, Chinese semiconductor companies can continue to develop wafer manufacturing equipment, materials, and EDA tools, along with sustaining investment in manufacturing capacity.

“More than Moore’s Law” focuses on potential disruptive technologies in the semiconductor industry. It refers to technologies and products that are based upon or derived from circuit design and system algorithm optimization and the use of new materials rather than on simply adding more transistors to a chip. At present, the semiconductor process node has reached 5 nm (mass production). Foundries have started to work on the 3 nm node or below. However, advancing nodes alone cannot fully meet market demand for better chip performance and more complicated functions. We expect innovative technologies to drive further improvements in the semiconductor industry in the post-Moore era from the four aspects below:

The area of computing principles includes quantum computing, photon computing, and neuromorphic computing using quantum action law, photon action law, and neuromorphic information processing law to replace classic electronic computing and NOR characterization computing. In theory, these can be used with certain types of algorithms to increase computing efficiency.

As for materials and devices, compound semiconductors (e.g., GaAs, GaN, SiC, and Ga₂O₃) have advantages of wide bandgap, high thermal conductivity, and high radiation resistance. These semiconductors enjoy notable advantages over silicon-based (Si) semiconductors when used in high-speed, high-frequency, and high-power applications. Carbon-based devices (e.g., graphene and carbon nanotube) have the advantage of high electron migration rate and can theoretically work at a rate of nearly 200 times higher than that of silicon-based devices. Flexible devices (e.g., carbon nanotubes and ZnO) can theoretically be better adapted to applications in the field of flexible electronics. New types of memory (e.g., phase-change memory, ferroelectric RAM, magnetic core memory, and resistive random-access memory) have the advantages of high reliability, fast access speed, and low power consumption compared with traditional memory such as DRAM, NAND flash, and NOR flash.

Regarding computing architecture, RISC-V has the advantages of being open source and featuring simple architecture and modular design. RSIC-V is actively promoted in the Internet of Things (IoT) and other related fields. With the advances in AI technology (especially the emergence of compute unified device architecture [CUDA] technology), heterogeneous architectures have now become widely used. The integration of storage and computing (i.e., resistive random-access memory) combines the current two basic functional units of computer storage and computing into one unit. In theory, it can form a better coupling with AI algorithms (i.e., neural networks).

Regarding chip integration, advanced packaging technologies such as Chiplet, system in packaging (SiP), and 3D stacking are important trends for the packaging industry.

3.2.2 Industrial Policies Have Supported Semiconductor Innovations

Industrial policies have a profound impact on the semiconductor industry around the world. The majority of countries and regions involved in the semiconductor industry chain have actively supported the development of the semiconductor industry. Different industrial policies have been implemented at the different stages of the semiconductor industry, including the formation of industrial alliances; the creation of industrial clusters; and the promotion of industry-university-research integration, fiscal and tax incentives, direct investment, and support for technology transfer. Given the current stage of development and external environment of China’s semiconductor industry, we believe the industrial policies of other countries and regions may offer some examples of successful cases for China.

Based on the formal incidence³ of support measures and transfer mechanism,⁴ the Organization for Economic Co-operation and Development (OECD) categorizes the support measures of governments around the world in the semiconductor value chain, and derives a two-dimensional matrix.⁵ Using this matrix as a research framework, the OECD has found that government support for R&D is one of the most common forms of state intervention in the semiconductor value chain. Less common is for governments to intervene directly in the production of semiconductors, either through direct ownership of semiconductor companies or by exerting strong influence on the decisions of local companies.

Tax incentives are the most common support measure for global semiconductor industries. The OECD pointed out that R&D tax incentives have become an important way to increase the attractiveness of the national research ecosystem.⁶ Russia, Israel, and the US rank as the top three in the world in terms of direct government funding for R&D as a percentage of GDP, and France, Belgium, and Ireland rank as the top three in terms of tax support for business R&D as a percentage of GDP (Fig. 3.8).

Fig. 3.8

Source OECD, R&D Tax incentives database, CICC Research

Direct government funding and tax support for business R&D as a percentage of GDP by country and region (2006 vs. 2016).

Government support mainly comes from the fiscal budget. The OECD’s analysis results for 21 large companies operating across the semiconductor value chain indicate that total global government support has exceeded US$50bn over 2014–2018. This comprises support provided through government budgets, and below-market borrowings and equity investment.

Budgetary support mainly targets R&D, capex, and revenue. Most budgetary support targets R&D of semiconductor vendors. This is consistent with the trend of the semiconductor industry needing a large amount of R&D investment. Governments also provide fiscal support for capex of companies that involve asset-heavy operations, such as Taiwan Semiconductor Manufacturing Co (TSMC), Vanguard International Semiconductor Con (VIS), and other wafer foundries. In addition to targeted subsidies for R&D and capex, governments also support enterprises by reducing or exempting corporate income tax.

3.2.3 Three Drivers Behind China’s Semiconductor Innovation

The semiconductor industry emphasizes innovation in R&D, technology iteration, and business model. There are three key drivers behind: Capital, talent, and technology.

3.2.3.1 Capital: Chinese Companies Narrowing the Gap with US Companies in R&D Expense Ratio, But Still Lag Behind in Terms of Total R&D Spending

Average R&D expense ratio of Chinese companies has risen from 5% in 2010 to about 10% in 2020, versus over 15% for the US companies during this period (Fig. 3.9). Meanwhile, China’s total R&D spending on electronics and electrical equipment, technology hardware, software, and computers was lower than that of the US in 2015–2020 (Fig. 3.10).

Fig. 3.9

Source Wind Info, Bloomberg, CICC Research

The gap in the average R&D expense ratio of listed semiconductor companies between China and the US is narrowing.

Fig. 3.10

Source The EU Industrial R&D Investment Scoreboard, CICC Research

China saw faster growth in R&D spending in electronics and electrical equipment than the US, but R&D spending on technology hardware and software and growth were slower than those in the US in 2015–2020.

Industry funds can play a role in guiding the semiconductor industry. Investment in the semiconductor industry can be classified into four categories. Investing in mature products has low barriers to entry and low risks, while investing in projects in their incubation stages has high barriers to entry and high risks. Investing in industrial clusters requires large and continued investments in infrastructure, which has high barriers to entry and high risks, but can bring high long-term investment returns and indirect improvement to the regional economy and supporting industries. In addition, investing in basic science has limited direct capital returns in the short term, but it contributes to the development of the upstream stages of the semiconductor industry. However, basic science is a necessary but not sufficient condition for enhancing the competitiveness of the semiconductor industry.

Reasonable division of labor between public and private capital can improve the effectiveness of investment. The central government can coordinate planning of local governments to form the optimum industrial structure and provide guidance in talent education and basic science. Local governments can formulate plans based on the existing regional conditions. Public capital should invest based on market-oriented practices as a market participant. Private sector should avoid repeated and blind investment.

3.2.3.2 Talent: Quality and Structure of Expertise Could Be Improved Despite the High Number of Professionals in China

China’s semiconductor industry has more employees than that of the US, but Chinese expertise focuses more on design than on manufacturing at the current stage. Compared with the US and other countries and regions with mature semiconductor industries, China has a larger number of employees in the IC industry, and the industry is growing faster. As of end-2019, the number of people directly engaged in the IC industry in China was about 512,000, a YoY increase of 11.04%, of which the total number of people in the design and manufacturing industry was 353,000, versus 277,000 in the US.⁷ In 2019, the number of employees in China’s semiconductor design industry and manufacturing industry was 181,200 and 171,900, versus 92,000 and 185,000 in the US.⁸

The attractiveness of the domestic semiconductor industry still could be improved. In 2019, only 13% of China’s college graduates majoring in semiconductors entered the semiconductor industry, and this figure was 55% for the 28 universities that have exemplary microelectronics colleges.

Upside potential in the number of foreign employees in China’s semiconductor industry. US citizens accounted for just 59% of senior employees in the US semiconductor industry in 2012–2016, with the remainder mainly from India, China, South Korea, and other countries (Fig. 3.11). In 2020, Chinese citizens accounted for 87% of senior employees in listed semiconductor companies on China’s STAR Market, and the US citizens accounted for 9% (Fig. 3.12).

Fig. 3.11

Source CSET, CICC Research

Number of high-skilled workers in the U.S. “electronic components and products” industry by place of origin, 2012–2016.

Fig. 3.12

Source Wind, CICC Research

Nationality of directors and senior executives of Chinese semiconductor companies listed on the STAR Market in 2020.

China has a larger number of highly educated professionals in the semiconductor design and manufacturing fields than the US, but still has room for improvement in terms of the international ranking of corresponding universities. Some 39% of employees in the semiconductor design industry have master’s degrees or above in China, and 43% have bachelor’s degrees, both higher proportions than in the US. Among semiconductor manufacturing employees, 32% have bachelor’s degrees and 20% have master’s degrees or above in China, versus 19% and 8% in the US (Fig. 3.13). According to the QS World University Rankings 2021, 26 of the top 100 universities in electronics are in the US and eight are in China. Based on the 2019 Academic Ranking of World Universities (ARWU) released by ShanghaiRanking Consultancy, eight of the world’s top 10 semiconductor-related universities are in the US (Fig. 3.14).

Fig. 3.13

Source Wind, ACS 2019, CSET, CICC Research

Highly educated professionals account for a larger proportion in China’s semiconductor design and manufacturing industries than the US in 2019.

Fig. 3.14

Source QS, CICC Research

Number of Chinese and the US universities ranked among the world’s top 100 in electronics by QS Consultancy (2021).

In addition, US professionals are more experienced than their counterparts in China. Most employees are aged between 35 and 50 in the US, and 25–35 in China. It takes time for Chinese professionals to accumulate experience. We suggest that China pay more attention to foster senior professionals and create a sound environment for innovation and entrepreneurship.

3.2.3.3 Technology: China’s Academic Research is Gaining Momentum; International Cooperation is Deepening

China’s academic research in the field of semiconductors has achieved positive initial results; number of papers rising steadily in recent years. Compared with the US, EU, and other developed countries and regions, China is a newcomer in academic research in semiconductors, and it is lagging far behind in terms of the number and quality of papers produced. However, the number of papers in which Chinese authors have participated has steadily increased thanks to supportive policies and rising investment.

Number of papers in cooperation with other countries and regions continues to grow. SNV data shows that number of papers jointly published by authors based in the US, EU, and China rose YoY from 1995 to 2020 (Fig. 3.15). In addition, the proportion of jointly-published papers has remained high. At present, Chinese scholars mainly cooperate with scholars in the US and EU. International cooperation has become an important feature of academic research in the semiconductor industry. Moreover, the number of PCT patent applications in the semiconductor field in China continued to increase from 2000 to 2020, especially after the launch of the STAR Market in 2019.

Fig. 3.15

Source Stiftung Neue Verantwortung, CICC Research

Number of China’s papers on semiconductor in cooperation with other countries and regions. Note Data comes from IEDM, ISSCC and VLSI.

3.3 Solutions to Innovation of System Software and Application Software

3.3.1 Innovation Schemes in the Software Sector

Software innovation can be both vigorous (e.g., unsolved problems and brand-new platforms) and gradual (demand-based upgrades) in terms of the innovation model. The major incentive for the innovation of system software lies in the high pricing secured by monopolies. For example, companies such as Microsoft and Oracle enjoy high profit margins thanks to their monopolies. In markets for IT services and application software, however, new entrants can take market share from their predecessors by offering value-for-money products. Thus, it is important for frontrunners to leverage low pricing that they secured through economies of scale to disadvantage small rivals. Unlike the innovation in system software and new IT architectures that originates from R&D at universities and large software companies, innovation in application software and IT services relies more on feedback and actual user needs.

In terms of commercialization, users can acquire either open or closed source system software at a much lower cost than if they develop the software themselves. Commercially, IP and monopoly guarantees profitability of closed-source software, while open source software is more about the commercialization of public knowledge. We believe that the performance and anecdotal reputation of products and the quality and stability of services are important factors affecting the profitability of application software and IT services.

System software innovation is vigorous, and uncertainty exists over the length of the R&D period and the potential of market size. Prices for such software must be minimized to maximize social benefits. Hence, developing system software—closed or open source—requires extensive and sustained investment. Innovation of application software, however, happens gradually, and it is driven by market demand, making application software a better choice for venture capital. Investment in software projects from which returns do not match the risk needs more policy and funding support from the government (Fig. 3.16).

Fig. 3.16

Source CICC Research

Innovation models in the software industry.

Innovation of system software and application software can be achieved in following ways. For system software, we believe open source technology provides a reasonable solution to system software innovation. Given developed countries’ first-mover advantage in system software, we think it is difficult for China to adhere to a closed-source roadmap and start from scratch. The global adoption of open source technology offers domestic system software companies an opportunity to catch up with their foreign counterparts. For application software, we believe that participation in the setting of industry standards is a decisive factor for market position in the application software sector. Chinese companies are excluded from setting standards for industrial design software such as CAD, CAE, and EDA; therefore, we believe the setting of standards will be a major focus of China’s application software innovation in the future.

3.3.2 System Software Innovation: Embracing Open Source

3.3.2.1 Open Source Technology: A Hindrance or a Boost to Software Innovation?

Is open source technology a hindrance or a boost to the software innovation? The open source community offers a platform for developers around the world to develop software projects jointly. However, free open source software may squeeze the profit at commercial software companies, and discourage them from continuing to invest in software R&D.

Open source is good for innovation. Figure 3.17 presents the pros and cons of open source software, according to research by Bitzer and Schroder in 2005. Bitzer and Schroder⁹ concluded that innovation of open source software costs less than that of closed source software, and that the market shift from monopoly to competition can enhance the technological strength of developers of both open source and closed source software.

Fig. 3.17

Source Bitzer & Schroder (2005), CICC Research

Pros and cons of open source software.

Open source is applicable to software sector because of economies of scale. Marginal cost of data reproduction is almost negligible for software companies, meaning that increases in cost are limited when service to a new user starts. In addition, the software industry features non-competitive supply. The cost of replicating and transmitting data is almost zero, and data use by new users does not raise supply cost for software companies.

Which type of software can better adapt to the open source model? Open source is not an ideal option for application software. Despite the advantages listed above, open source software has its shortfalls. Most open source software is system software for IT maintenance, operating systems, and databases. Open source software is normally less competitive in terms of user interface (UI), product documentation, and usability tests. In addition, it often cannot satisfy business demand for recovery of work and high availability (a measure of software performance). The root cause is that open source software is made for and developed by users, and a large proportion of participants in the open source community are IT administrators. Hence, free open source application software is generally less user friendly than paid open source-based software or commercial closed source software.

International competition of open source community. Open source has an impact on international competition and supply chain security as well. Technically, the open source community are not bound by national borders, but developers are. This can lead to potential value chain security risks related to code management platforms, foundations, and licenses. The trend towards open sourcing of system software is already well established globally. To safeguard value chain security, it is important for China to encourage domestic developers to join this trend and establish a leading role in the global open source community. Github predicted that China will have the world’s most active open source developer communities by 2030.

However, currently, China’s role in the global open source community is extensive but not necessarily influential. To increase influence, it is important for Chinese companies to participate more in significant open source projects and donate more open source projects to the global community. If domestic developers are unable to exert influence on the global community, it is likely that China’s own open source community will attract fewer developers.

3.3.2.2 Open Source Technology is a Feasible Solution to China’s System Software Innovation

Open source software has become an important infrastructure for new-generation software. Open source software is released under a license whereby the copyright holder grants users the rights to use, study, change, and distribute the software and its source code. The sharing of source codes can help prevent repetitive code development. Open source is a revolutionary trend increasingly used in the software industry. Open source system software such as Linux (operating system), Java (middleware), and PostgreSQL (database) is becoming mainstream, and upgrades and innovations are proceeding significantly faster than are those of closed-source software. Surveys by Gartner and Sonatype¹⁰ show that 99% of global organizations are using open source software, and that 3,000 companies surveyed download open source software an average of 5,000 times per year.

Chinese system software developers are embracing open source ecosystem. US tech giants such as Google, Amazon, and Intel are taking part in developing open source projects and acquiring open source software companies. Meanwhile, Chinese software developers are also an important part of the international open source community, and multiple sources have been gradually opened since 2019, such as Huawei-led distributed database GaussDB, Linux distribution version of openEuler, terminal operating system framework OpenHarmony, and Alibaba-led cloud native relational database PolarDB and distributed relational database Oceanbase. Data from the China Academy of Information and Communications Technology shows that Alibaba had opened 2,172 sources as of September 2020, and Tencent and Huawei had both opened more than 150 sources, including Apache projects such as Dubbo and CarbonData (Fig. 3.18).

Fig. 3.18

Source Company websites, CICC Research

Global new-generation system software mainly open source; now being rapidly promoted in China.

Open source is extending to more industries and fields. Open source technology is being promoted in industries other than the internet industry. Goldman Sachs open sourced its data-modeling program Alloy; petroleum giant ExxonMobil released the Standards Devkit development kit to the open source community to create a standard data interchange format in the oil and gas industry; leading retailer Walmart made its OneOps platform for cloud and application lifecycle management available to other retailers. The open source trend is extending to various industries and fields.

Swiftness in open source upgrades have become a major competitive advantage for system software. The open source community saves programmers the repetitive work of developing the same codes that are already available in the community, which contributes to faster code upgrades. The collaboration and sharing in the open source community can also attract additional software developers, which in turn improves the user experience. This makes it possible to realize the concept of “open source ecosystem by and for developers”.

Open source technology is likely to help Chinese software companies catch up with global frontrunners. The open source community creates a level playing field for global developers to access existing source codes and contribute their own codes. A growing number of domestic companies have been enhancing their influence in the open source community in recent years, and so have Chinese software developers.

Therefore, open sourcing is a possible solution to system software innovation. We believe that Chinese software companies can base their research and innovation on the existing open source framework, and develop their own user-friendly system software. The bulk of existing domestic system software is based on open source software—e.g., Linux, Java, PostgreSQL, Hadoop, and Spark.

3.3.3 Application Software Innovation: From Follower to Standard Setter

3.3.3.1 Standards in Economics and How It Relates to Innovation

To examine application software innovation from the perspective of standard economics, we first distinguish formal standards from de facto standards. Economists classify standards into formal or de jure and de facto standards. Formal standards are based on deliberations of standards-writing organizations or mandatory standards issued by governments, while de facto standards are set by interest groups comprising single or a few enterprises, and are widely accepted by the market.

Looking into standards in the software sector, formal standards in software sector mainly apply to coding, such as the international UTF-8 standard and China’s GB-2312 standard, while standards for application software are mainly launched by companies and are gradually accepted in the industry. Unlike system software, application software is used in specific downstream fields, and can itself be seen as a means of production. For application software companies, setting standards in niche markets can secure absolute advantage. Standards in the traditional industrial age are more about quality, while standards in the IT era focus more on compatibility and interfaces. In IT related sectors such as software, frontrunners that lead the setting of de facto standards usually have solid competitive advantages.

Software standards also have network externality. The format of application software files is a typical computability standard, e.g. Microsoft’s Document, Excel, and PowerPoint and Adobe’s PDF. Leading companies can establish first-mover advantage by being the first to set the de facto standards, to which late entrants would have to adapt.

The goal of standardization is to enhance efficiency and promote innovation. During the course of technological innovation, standards can also serve as a fair and open technology infrastructure, and a foundation for innovation-driven growth.

Monopoly over standard setting and network externality may impede innovation. A few software companies might leverage the network externality of standards to create solid competitive advantage for themselves. Leading companies may continue to monopolize the setting of industry standards, which in some circumstances would impede innovation (Fig. 3.19).

Fig. 3.19

Source David P A, Greenstein S, The economics of compatibility standards: An introduction to recent research, 1990, CICC Research

Standard setters typically possess a more advantageous position in the market.

Standards should be properly governed to ensure their positive impact on innovation. In order to weaken the monopoly over standard setting, the common solution is to establish governance organizations to constrain standard setters and properly manage the standards. In the application software sector, the PDF Association and Open Document Architecture (ODA) are typical independent standard governance organizations, and their members are mainly industry participants other than standard setters Adobe and Autodesk.

In addition, participation in setting industry standards is crucial for the security of the application-software value chain. Standards for the application software sector are set by leading providers along the value chain, a market position achieved by: Attracting large numbers of software developers by creating a comprehensive tool chain; making better use of existing programs and source codes; educating users via promotion among colleges and training agencies; recruiting distributors around the world to better service clients; and creating ongoing exchanges between developers and users (Fig. 3.20).

Fig. 3.20

Source CICC Research

Application software: importance of participating in the setting of industry standards.

Therefore, Chinese application software companies can participate in the governance of international standards. By joining international standard governance organizations, Chinese companies can exert greater influence in setting new technological standards. As one of the most important application software markets, China can be an important participant in improving the standard governance network. We believe that by joining the global market, Chinese companies can help diversify technology and product standards and advance global innovation.

We suggest that Chinese companies start by setting local standards in emerging sectors. Domestic companies can set local standards for application software, but it is difficult for these standards to compete with existing international de facto standards. Thus, we believe Chinese companies can start by setting local standards in emerging sectors such as AI, Internet of Things (IoT), and industrial internet. We expect domestic software companies to lead the setting of standards for these emerging sectors.

3.3.3.2 Standard Setting in China’s Application Software Industry

In terms of industrial software, companies in China are currently unable to participate in standard setting. Industrial software standards are based on software companies’ knowledge of downstream manufacturing technology. Industrial software needs to be continuously improved and adapt to special requests from clients in various industries. Overseas high-end manufacturing giants (e.g., traditional aviation, automotive, machinery manufacturing, and chips) have collaborated with foreign industrial software suppliers, leaving Chinese companies little chance to serve high-end manufacturers and participate in setting standards.

In comparison, overseas companies have absolute advantage in high-end industrial software and simulation software. The high-end industrial design and simulation software market (e.g., aviation design) in China remains dominated by foreign industry giants, while domestic industrial software is mainly used for low-end industrial design (Fig. 3.21). China is also relatively short of user-friendly simulation software, and is, therefore, susceptible to potential value chain risks.

Fig. 3.21

Source e-works, CIMData, CCID Consulting, CICC Research

China’s market for industrial design software.

Support from downstream sector is needed for Chinese industrial software companies to thrive over the long term (Fig. 3.22). Industrial software products cannot be improved without support from downstream manufacturing sectors. Besides enhancing their competitive strengths, more and more domestic high-end manufacturers began have started to adopt industrial software provided by domestic companies such as ZWSOFT and Empyrean. Nevertheless, we believe that the rise of China’s industrial software will take some time; it might be 5–10 years until Chinese industrial software companies participate in setting international standards. During the course of achieving this, support from downstream sectors is essential for domestic industrial software suppliers to make breakthroughs in innovation and to narrow the gap with their foreign peers.

Fig. 3.22

Source CICC Research

Support from downstream sector needed for Chinese industrial software companies to thrive over the long term.

Management software and vertical market software, however, are two rare domains in which domestic players are capable of setting standards. Local standards for management process and vertical market software are well-established in China, and Chinese companies have naturally developed their own sets of standards for business management processes with Chinese characteristics, with many clients requiring heavy customization and intensive services. We see no significant supply chain risk in this sector.

Setting local standards lays foundation for local market leadership. As extensive services are needed for vertical market software, local providers can respond more swiftly and offer prices that are more appealing than overseas providers. The domestic market for vertical market software is dominated by domestic companies (Figs. 3.23 and 3.24). However, this situation adds to the difficulties that Chinese companies have in going global. European and US enterprises had an early start in digital management, and some still dominate the high-end management software market. We believe domestic companies can begin their global expansion by tapping into developing countries.

Fig. 3.23

Source Qianzhan Industrial Research Institute, CICC Research

China’s ERP software market (2020).

Fig. 3.24

Source IDC, CICC Research

China’s hospital management system market (2020).

After nearly 30 years of development, some Chinese companies can participate in the setting of global standards for office software. We see limited supply chain risk in this sector. For example, Chinese company KingSoft has developed its own WPS Office software, which is compatible with domestic operating systems, and it started deployment in mobile and cloud technology three years earlier than Microsoft. As of June 2021, MAU of WPS Office PC and mobile versions exceeded 199 and 296 mn.¹¹