Amid growing geopolitical tensions, the Chinese government has made clear its aim to achieve techno-independence, denying rivals such as the United States the ability to leverage international supply chains and exclusive intellectual property to restrict Chinese freedom of action. While recent estimates suggest that China’s domestic semiconductor industry is still at least a decade behind its international competitors, China need not surpass the global state of the art to significantly diminish foreign leverage.
As Elsa B. Kania and Lorand Laskai have written, Military-Civil Fusion (MCF) encompasses “everything from efforts in big data and infrastructure to logistics and national defense mobilization.” In a pattern witnessed in many countries, the strategy aims to create synergies between the private and defense sectors. Generally, these synergies consist of making private sector technology more easily “spun-on” to defense applications as well as “spun-off” from defense-related research to private enterprises.
While this typical pattern may one day appear in the Chinese semiconductor industry, China's leading firms and military efforts today significantly lag their global competitors. MCF and efforts at indigenous or independent innovation appear to be fueling industrial advances that are not yet globally competitive but help Chinese industry level up while freeing some military capabilities from reliance on supplies subject to foreign manipulation. In turn, the military is a reliable customer for these otherwise less competitive products.
Semiconductor industry development in the context of the Chinese government's broader MCF strategy shows how industrial policy can support otherwise uncompetitive actors in service of specific strategic goals.
From spin-on and spin-off to step-up
China’s semiconductor industry sits several generations behind the leading-edge of innovation. Massive investments are required to close that gap, particularly in chip manufacturing. Therefore, the ability of the defense sector to spin-off anything cutting-edge—or of the private sector to spin-on anything superior to foreign offerings—is limited. Chinese companies have attempted to acquire foreign leading-edge technology through forced technology transfer agreements, intellectual property theft, and talent poaching.
However, China’s efforts to acquire know-how from abroad have so far failed to spark genuine domestic innovation. Instead, MCF and semiconductor efforts have found some positive synergy by applying defense procurement spending to prop-up China’s fledgling semiconductor companies that are globally uncompetitive but key to the PLA’s defense needs. Military and government contracts allow these companies to remain afloat. Over time, this support can allow these companies to acquire the experience and scale necessary to find a competitive niche in the expanding semiconductor industry and eventually compete globally. Rather than spin on or spin off, MCF is helping domestic semiconductor firms step up.
Changsha Jingjia Microelectronics Co., Ltd (Jingjia Micro, 景嘉微) illustrates the step-up possibility in China’s semiconductor industry. Originally a military-focused graphical processing unit (GPU) design company, Jingjia Micro used its early support from the government to develop several generations of GPUs for deployment in radars and satellites. In 2019, Jingjia Micro leveraged its acquired know-how to enter the civilian market for GPUs. According to the company, Jingjia’s first commercial line of GPUs, the JM9 series, has performance comparable to top-of-the-line foreign chips from 2016.
This is a considerable improvement over previous designs, even though Jingjia only recently announced completion of the design with final testing and mass production still pending and the viability of the chip on the commercial market still unknown. It also increases the possibility that China will complete its first competitive indigenously designed GPU processor—a major and necessary achievement given the importance of GPUs to deep learning and other types of advanced computing.
Cambricon Technologies (寒武纪) is another example of the step-up possibility. One of China’s leading developers of specialized hardware for cutting-edge AI applications, Cambricon is a sizable beneficiary of Chinese government procurement contracts. The Chinese government sees AI as a core component of its goal to achieve technological dominance as well as an opportunity to leapfrog the United States in military modernization efforts. Cambricon was initially spun out of the Chinese Academy of Science, a government-funded research institution, and received initial seed funding from state-backed financial institutions. The company’s relationship with the Chinese military is opaque. However, it counts defense-related firms like supercomputer maker Sugon, which was blacklisted by the U.S. Department of Commerce in 2019, among its biggest clients. The company’s 2019 prospectus revealed that its main source of revenue is from government and state-owned clients.
The lifeline that the PLA and the defense base provides fledgling semiconductor companies is akin to the traditional formula for industrial-defense policy, whereby firms are propped up on national security grounds. MCF offers an additional layer to this formula: government funding not only helps support companies, but also assists them in eventually attaining commercial viability. While it remains to be seen whether companies like Jingjia Micro and Cambricon will ever become profitable commercial actors that compete with market leaders, MCF gives them a fighting chance.
Autonomy now: out with the old, in with the outdated
Support for non-competitive semiconductor companies is not an act of charity on behalf of the PLA. Rather, it helps fulfill another objective of semiconductor policy: reducing reliance on foreign chips. The Chinese government has staked the ambitious goal of achieving self-reliance in semiconductor production. While China is far from achieving complete semiconductor autonomy, military autonomy—replacing foreign chips in military applications—is much more attainable and strategically urgent. Attaining military self-sufficiency would allow China to utilize its military to advance its foreign policy interests without the fear of facing a foreign supply shock. It would also guard against the informational security risk associated with foreign chips in critical communication networks.
Several characteristics of the military semiconductor market make it an easier target for self-sufficiency. First, the military market overall is much smaller than the domestic market. The PLA’s spending on semiconductors makes up only two percent of all military spending and is projected to peak at $853 million—a tiny amount compared to the more than $300 billion in semiconductors China imports every year. The smaller market means that military self-sufficiency can be attained with much lower production capacity; China’s foundries, the largest of which is the Semiconductor Manufacturing International Corporation (SMIC), do not have nearly the capacity required to achieve full domestic production, but SMIC could quickly reach production levels required for military autonomy.
Second, the lengthy development cycles for modern weapon systems and their long service lives mean that many of the processors used in military applications are often several generations behind the state-of-the-art. China currently has the capacity to produce chips for its night vision goggles, and both military and civilian vehicles utilize dated 28nm and 40nm nodes which SMIC can produce. China may not beat the United States to developing a leading-edge chip, but it can still reduce dependence on foreign supply by indigenizing older generations of chips already in use in existing military applications.
An MCF-driven ecosystem for semiconductors is already aiding this process of swapping out foreign chips. Prior to Jingjia Micro’s development of the JM5400 GPU, Chinese military drones and aircraft relied in part on M9 GPUs produced by ATI, a Canadian company owned by U.S.-based Advanced Micro Devices. The M9 chip was originally developed in 2002, and Jingjia was only able to match its performance in 2018. However, with the combination of SMIC and Jingjia, China’s ability to domestically design and produce GPUs for military use in order to “fully replace foreign chips such as the M9” is still a major milestone for the PLA.
Limited horizons: advanced research on narrow frontiers
The final area where MCF and semiconductors overlap is military-related research. Computing capacity is an important requirement for military R&D. Though not exclusively dedicated to military research, supercomputers are used in modeling sophisticated weapons such as nuclear detonations and hypersonic missiles. Unsurprisingly, the chips powering China’s supercomputers have become the object of geopolitical tension. In 2015, the United States blocked Intel from exporting Xeon Phi processors, which had powered some of China’s fastest supercomputers. China’s ability to weather that blow and quickly replace Intel chips with domestic alternatives is testament to the power of MCF in the semiconductor ecosystem.
That ecosystem involves an assortment of civilian-military labs that work closely with SMIC to iterate and test chip designs. China’s research on radiation-hardened integrated circuits—which are used to withstand the conditions of space—has relied extensively on the SMIC process design kit. An analysis of research papers published by the National University of Defense Technology (NUDT) shows that NUDT researchers “regularly use the SMIC process to manufacture their test chips.”
When the United States denied Intel’s export license in 2015, that ecosystem jumped into action. Thanks to work by the Jiangnan Institute of Computing Technology (江南计算技术研究所), China was able to release the Sunway Taihu Light the following year, which for a period of time was the world’s fastest supercomputer. The Taihu Light employed the SW26010, a processor based on an entirely domestic computing architecture developed by the Jiangnan Institute of Computing Technology. Though the Jiangnan Institute of Computing Technology claims civilian status, it is owned by the 56th research institute of the PLA. The blending of civilian and military institutes for the research of supercomputing allowed the Chinese government to advance its military research and import foreign technology for years.
In 2019, the U.S. Department of Commerce added the Jiangnan Institute of Computing Research to its Entity List. However, that has not stopped China’s progress on supercomputing. Building on their successful turn toward domestic production, China utilized its domestic design and manufacturing capabilities to quietly develop and deploy the world’s first two exascale supercomputers in 2021. MCF, with domestic semiconductor companies in the mix, has blunted the effects of sanctions on China’s military-based research, creating an important area of independent innovation.
While China may not be able to dominate any segments of the semiconductor market in the near-term, its ability to leverage state resources to find common interests between civilian and military semiconductor production has already increased the level of foreign policy autonomy the state enjoys. MCF-driven semiconductor investment has played a key role in increasing state autonomy by helping prop-up non-competitive firms, allowing for the replacement of outdated foreign made chips with “new” domestic processors in key military weapons, and permitting military focused research to proceed without the threat of foreign sanctions. Thus, even if China falls short of its grander aspirations for technological dominance and seamless integration between the military and civilian sectors, limited success in the fusion of semiconductor companies with military research and production can still meaningfully increase the Chinese government’s autonomy.
The views expressed here are the author’s alone and do not necessarily reflect those of the U.S. government or any part thereof.