Upon triggering Chip War to attain semiconductor independence, China has been hoping to repeat the past success of A-bomb making. Despite the Soviet Union’s withdrawal of technical support from empowering China to make a nuclear bomb, within four years, China succeeded in detonating A-bomb in 1964. But with the swiping export controls of the USA on advanced chip-making technologies, will China’s semiconductor independence see the light, as A-bomb did?
To attain chip independence, China needs to overcome four significant challenges. The first three are state-of-the-art (i) process equipment, (ii) wafers, chemicals, and gases, and (iii) process integration and yield management expertise. And the fourth one is to have enough demand to reach the minimum efficient scale of highly capital-intensive plants.
So far, to attain independence, China has been after acquiring process equipment by importing. From lithography, deposition, and cleaning to testing equipment, China has been sourcing all kinds of high-end equipment from the USA and its allies. For example, American Lam Research, KLA, and Applied Materials are significant process equipment suppliers. On top of it, there are Dutch ASML, and Japan’s Tokyo Electron and Nikon, among many others. Japan and other American allies are sources of wafers, gases, and chemicals. Among the top 10 process equipment makers (05 from the USA, 04 from Japan, and 01 from Dutch); there is none from China. China’s Naura, AMEC, and Hwatsing are far weaker than the top 10 equipment makers. Besides, China has been after foreign experts for process integration and optimization, mainly carrying American passports.
For scale–domestic demand is good enough
Well, for dealing with the scale advantage of expensive plants, China is in a comfortable situation. China meets large domestic demand through imports. In 2020 alone, China imported $378 billion worth of semiconductors for assembling 35% of the world’s electronic devices. This dollar value translates to the import volume of integrated circuits (IC) amounted to over 635 billion units in 2021. Besides, Government’s generous fund can take care of the rest. So, the reality raises serious doubt about whether China’s Semiconductor independence will experience A-bomb-making success.
US strategy to prevent China’s semiconductor independence
Unfolding American strategy has been maintaining a high gap between the process nodes—keeping China behind. Currently, the target is to ensure China cannot move beyond 14nm while TSMC and Samsung keep working on 3nm process nodes. Due to recent restrictions, if the 14nm lithography machine fails, SMIC will be compelled to revert to 28nm; a technology node Taiwan put into mass production in 2011. Furthermore, USA’s restrictions ensure that China cannot import chips produced with the latest nodes. And Chinese companies are not even allowed to take foundry services from TSMC or others to print their designs on the wafer using a sub-10-nm process node.
To make sure that China lags, the USA has been rolling out a growing list of unverified companies. The recently released one on Oct 07, 2022 includes NAND flash memory chip maker YMTC and 30 others. These are the companies that the United States could not complete on-site visits to determine whether they can be trusted to receive sensitive technology exports from the United States. As a result, a growing list of unverified companies is banned from obtaining US technology without a license.
Unlike previous sanctions targeting companies like Huawei, SMIC, or specific chips from US suppliers such as Nvidia, the updated export control measures have seemingly been limiting every Chinese semiconductor company’s maneuverability. Several Chinese Think Tanks are expressing that there was no question now that the US was imposing blanket sanctions to stop the development of advanced chips in China. According to them, “a war is declared.” Experts are under the impression that China’s most important semiconductor companies will be “destroyed, damaged, or circumscribed” by the latest US measures.
China’s semiconductor independence mission in perspective
So far, China’s investment in chip fabrication plants, the import volume of semiconductors, and the USA’s regulatory measures in preventing technology access have been dominating China’s semiconductor independence discourse. Unlike many other missions, China’s semiconductor independence is about catching the moving target. Winning a performance race through discoveries, inventions, and precision engineering highly matters. Success will depend on the relative learning performance of China and the US-led blue channel. Assimilating and incrementally advancing existing technologies will not be sufficient. The capability of pursuing reinvention, forming a creative wave of destruction, will have a strong bearing. In addition to chip design and production process, China needs to pursue reinventions of existing products creating the demand for the next generation chips. To understand the unfolding of China’s semiconductor independence dynamics, we should draw a lesson from the history of invention and evolution—leading to the rise and fall of chip independence.
Although the focus is on production-level technologies and capacities, ultimately, China’s semiconductor independence will depend on its learning ability. The ability to pursue strategic possibilities amid uncertainties will largely determine success. Furthermore, success will also depend on component-level edges, which are being developed by many specialized firms distributed all across the world. For example, ASML‘s EUV machine has 100,000+ parts, developed and supplied by 4,000+ firms. Some components, like mirrors, are so specialized that there is only one supplier. Such specialization at every value chain layer demands well-managed distributed R&D for scientific discoveries, technological invention and advancement, and precision engineering. The journey to attaining the target may last over decades. For example, ASML’s EUV adventure took more than 20 years before releasing the first commercial lithography machine. Does China have such a track record in any industry?
Historical rise and fall of Semiconductor supremacy
After the invention of the Transistor in 1947, Bell Laboratory struggled to create commercial interest. Major American companies like RCA, GE, and Texas Instruments (TI) did show tepid interest. Ironically the transistor did not emerge as a strong substitute for vacuum tubes. Hence, RCA, TI, and other electronics companies did not find it suitable to replace vacuum tubes of consumer electronics products like Radios. Therefore, Bell Laboratories decided to license this primitive device so that competition in making further improvements would start. Consequentially, Bell licensed it to 40 companies in 1952 at $25,000 fees. To transfer the know-how, Bell arranged a nine-day Transistor Technology Symposium. Bell also arranged a tour for 100 representatives at Western Electric’s ultramodern transistor manufacturing plant in Allentown, PA. Subsequently, Bell compiled symposium proceedings as “Ma Bell’s Cookbook.”
Rise of Japan
Subsequently, this event seeded the rise of two semiconductor waves—one in Japan and another in the USA. Among the 40 licensees of Transistor, there was a tiny Sony from Japan. As Japanese companies did not have the option to do defense work, Sony led Japan embarked on using Transistor for consumer electronics. The journey began with Sony’s mission of reinventing the radio by changing bulky vacuum tubes with transistors.
Ironically, although Transistors were smaller, these were noisier than vacuum tubes. Furthermore, signal handling capacity was also low. Hence, Sony focused on its refinement to improve the quality and reduce the cost—making it suitable for the reinvention of all kinds of consumer electronics products. It’s worth noting that leveraging transistors for reinventing existing products led to the rise of Japanese firms, often destroying US and European counterparts. Consequentially, Japan became the largest producer of transistors, which it maintained till 1986. Furthermore, through the process, Japan developed a monopoly edge in several layers of the semiconductor value chain. Notable ones are wafers (Shin-Etsu), lithography (Nikon and Canon), test and process equipment (Tokyo Electron, Screen, Hitachi, and Advantest), chemicals, and gas.
Rise of the USA and Fall of Japan
America’s semiconductor industry got a kick start due to the departure of highly competent scientists and engineers (known as treacherous eight) from Shockley semiconductor. Unlike Japanese firms, American firms like Fairchild (formed by treacherous eight) and Texas Instruments focused on defense and space markets. They took advantage of the transistor in making onboard computer modules of US Airforce’s Bombers and NASA space craft’s lighter. The development of microprocessors, such as a single-chip digital logic unit like Intel 4004, also led to the expansion of the American semiconductor industry for the industrial control unit and personal computer making. However, despite it, America’s semiconductor industry was lagging behind the Japanese, making NEC the largest semiconductor device maker in the early 1980s.
Due to a few punitive measures of the USA to curb Japan’s dominance in semiconductors and the adoption of Intel’s microprocessor by IBM PC led to the rise of the USA’s dominance. Particularly, the exponential growth of PC diffusion led to the rapid expansion of firms in Silicon Valley, making Intel the global winner. American companies pursuing the integrated device-making (IDM) model became the globally dominant players. Through the process, the USA also established a strong position in semiconductor process equipment, electronic design automation software (Cadence, Synopsis, MG), and chip design. Notable US equipment makers are KLA, Applied Materials, Lam Research, Plasma-therm, and Tera Dyne. In addition to IDMs, NVIDIA, Qualcomm, and Broadcom are well-known for chip design.
Rise of South Korea, Taiwan, and Fall of the USA
Despite the growth of Silicon Valley with an apparent unbeatable edge, America started losing its silicon edge at the dawn of the 21st century. First, South Korea’s Samsung became the dominant player in DRAM. In this race, Samsung also accelerated its process nodes. Hence, upon rejection from Intel, Apple found Samsung as its partner for processing silicon for the iPhone processor.
On the other hand, Taiwan’s TSMC started to rise due to the growing demand for its 3rd party foundry services, overlooked by USA’s IDMs. In serving the rising demand for higher density and lower energy footprint of mobile handset processors, TSMC discovered an opportunity in R&D. Undertaking in-house research and transferring its outputs into profitable yield enhancement became its winning strategy. Furthermore, the iPhone’s computationally intensive multi-touch user interface scaled up processor demand. Besides, the growing popularity of handling images and video has led to the exponential demand for reducing the chip dimension down to sub 10nm.
Consequentially, TSMC’s ability of yield maximization through process R&D has led to TSMC becoming the preferred silicon processor. Its clients include Apple, Nvidia, QUALCOMM, AMD, and many others. As a result, TSMC, with a 5nm process node, succeeds in occupying 90% of the high-end processor-making market, pushing Intel to a distant 2nd. Besides TSMC, a strong cluster with chip testing and packaging, and design firms has formed in Taiwan–fueling monopolistic market power.
Rise and Fall of USA and Japan and Rise of Dutch’s ASML in Photolithography
In the semiconductor value chain, a great example of rise and fall has been in the photolithography layer. In the late 1950s, the American Bell labs and Defense research program opened the door to photolithography innovations. It was for the purpose of integrating multiple transistors on a single substrate as an integrated circuit. Subsequently, in 1961, American GCA demonstrated the first photolithography machine for integrated circuit production. Several American firms like K&S and Kasper Instruments also entered the market; but they started losing ground due to the superior incremental innovation performance of Japanese Nikon, Canon, and Hitachi. Hence, despite having financial support from USA’s SEMATEC, American firms in photolithography disappeared, leaving the market to Nikon and Canon duopoly.
In retrospect, it’s pretty interesting that Arthur del Prado’s humble journey, which started in the early days of Silicon Valley, led to the rise of ASML. Upon long persuasion, Arthur succeeded in forming a partnership with Philips to form ASML. This new baby to pursue photolithography in the semiconductor industry was born in a wooden hut and entrusted to 31 people to take care of. Despite the initial struggle, due to strong vision and R&D management skills, ASML has succeeded in delivering a 13.5 EUV light source-based photolithography machine. Interestingly, ASML is the only company worldwide to deliver it, which is vital for sub-10 nm chip production. Ironically, due to its critical role in high-end chip production, ASML has been the center point of the US-China Chip war.
Monopoly at each layer of the value chain poses an insurmountable barrier
The 70 years long journey of the global semiconductor industry is marked by intense competition in leveraging science and precision engineering to profit from ideas. It has been a race to benefit from ideas in making transistors increasingly smaller and better. Consequentially, the race of specialization with proprietary ideas in offering better performance has led to monopolization across the value chain. Every layer of the semiconductor value chain has firms with a high level of monopolistic market power. For certain critical components, there is only one supplier. And such a monopoly state is not due to subsidies or regulations. Instead, monopoly has arisen due to winners’ superior scientific, engineering, R&D management, and commercialization capability. For example, ASML managed a highly uncertain journey in turning EUV possibility into a real machine.
Besides, no single country or company has high-end machines across the value chain. The specialization has been distributed across numerous firms and as high as 25 countries. Hence, attaining semiconductor independence is not an attainable target for any country—let alone China. Furthermore, China’s semiconductor independence mission faces an insurmountable barrier. One of the reasons has been that there is not even a single company in China with unbeatable global performance. Even upon taking over Taiwan, China cannot attain its independence target, as TSMC depends on the supplies of numerous firms of America and its allies.
China’s strategic move to attain semiconductor independence
So far, China’s strategy has been to provide generous funding. That money goes for acquiring foreign firms, importing equipment, recruiting foreign experts, fostering startups, and setting up domestic production capacities. Notable examples are SMIC and YMTC (costing $24b). An army of chip design startups, numbered 2,810 as of December 2021, appears to be after subsidies. But the success of the semiconductor business depends on developing in-house R&D capability. It’s essential to keep improving the performance to win the global race in delivering higher quality at less cost. TSMC, ASML, and many others are testimonies of this reality. Otherwise, the initial demonstration of the production by the import of equipment and expertise will lead to failure, as the winner takes all. For example, in the 1990s, there were 20 NAND flash memory makers. But the number has shrunk to 6 as poor performers could not stay in the race.
Unfortunately, it appears that Tsinghua University (securing the 16th position in Times’ global HE ranking), holding the Unigroup, has not succeeded yet in outshining the competition through R&D. On the other hand, startups are after subsidies. Instead, they should be focusing on winning the race through incremental advancement and creative destruction. Besides, there are allegations of corruption of the pot of money ($150 billion) given for seeding China’s semiconductor independence.
Hence, China’s YMTC, SMIC, HiSilicon, Naura, and others are challenged to outperform the competition through superior performance. Instead of subsidy and external sourcing-based demonstration, they should outperform the competition through excellence. The same reality will also determine the fate of the American Government’s massive subsidy-based program to regain the semiconductor edge.
Is it a strategic mistake?
Perhaps, a more intelligent strategy for China’s Semiconductor independence could have been developed. The precursor to production should have been to develop ASML-type companies in a few layers of the semiconductor value chain. For example, if China had a company offering better lithography machines than that of ASML, the scenario could have been different. To attain such competence, China must outperform western companies in incremental advancement and reinvention (e.g., electron lithography). Such development could have made the US-led blue channel dependent on China. But at the moment, China has none.
Due to high-level specialization and patent holding, a monopoly-like scenario exists accorss the value chain. Intense competition, making rise and fall, has resulted in such a situation over the decades. Hence, it’s an extremely difficult job, perhaps not impossible, for China to develop an edge in any layer. So far, China has money to buy equipment and expertise for integration and operation. Unfortunately, such competence is not good enough to attain independence. The challenge is not like A-bomb making: acquire components, assimilate know-how, integrate, demonstrate and replicate. Instead, China’s chip independence demands continued progression in outperforming the growing competition edge. Hence, China has no card to play to deter the USA from preventing its development of silicon edge–let alone attaining semiconductor independence.
- Semiconductor Value Chain–globally distributed ecosystem
- Semiconductor Monopoly Due to Winning Race of Ideas
- Intel Falling Due to PC and Mobile Waves
- ASML Lithography Monopoly from Sustaining Innovation
- Taiwan’s Semiconductor Monopoly – How did it arise?
- ASML TSMC Nexus Fuels Semiconductor Monopoly
- ASML Monopoly in Semiconductor — where is magic?