Executive Summary

In September, 2020, the government of China announced the National Integrated Circuit Industry Investment Fund Phase II in the amount of $28.9 billion, which nearly doubles investment compared to the earlier Phase I. According to its manager, the fund could reach $47 billion, with a focus on etching machines, film deposition, test and wafer cleaning equipment. 

The goal is to make China self-sufficient in semiconductor production starting with the 28 nm CMOS node, the most widely-used fabrication node, or feature size, for integrated circuits (ICs).  China can meet most of its semiconductor needs by making ICs at this node, and it appears that this plan can succeed. 

Background

China needs reliable sources of semiconductors to continue producing and exporting electronics.  Producers of electronic products in mainland China, unable to get crucial semiconductor chips or fearing that they would be unable to get them in the future, have pushed for domestic China sources of semiconductors as a matter of survival.  On top of this, the Chinese government regards self-reliance in semiconductors to be of strategic importance in order to protect and maintain growth in manufacturing employment.

China buys about 36 percent of the $400-plus billion in semiconductors sold globally every year, and re-exports about half of these semiconductors inside finished electronic products such as PCs, TVs and cellphones.  Semiconductor producers in mainland China have a global market share of about 5 percent, while US companies have a share of 47 percent, according to the Semiconductor Industry Association (SIA).

The escalating trade war between China and the U.S. has brought the world to a new reality, a technology cold war.  The U.S. has extended its “Entity List” of blacklisted companies from Chinese military equipment producers to large Chinese producers of commercial products including makers of telecommunications equipment and the semiconductor foundry SMIC.  The U.S. has forbidden U.S. companies from buying from or selling to blacklisted Chinese companies, and has told U.S. allies that they must do the same or risk trade sanctions. 

The COVID-19 pandemic has caused additional disruptions to international trade, underlining the importance of companies having multiple sources of supply that include local sources.

We have argued against the technology trade war in another piece, but it is now a reality.  Much of the damage to the electronics industry supply chain has already been done, and the world has to live with it. 

Analysis

China’s Response to the Trade War

Even before Phase II of the semiconductor investment fund, the government of China had announced the China Manufacturing 2025 (MIC 2025), Internet Plus (IP), and Next Generation Artificial Intelligence Development (AIDP) plans.  The trade war between the U.S. and China seems to have really escalated in May of 2020 with harsh new sanctions against Huawei, spurring Phase II of the semiconductor investment plan and efforts by China to become self-reliant in 28 nm chip production as soon as possible. 

According to the Information Technology and Innovation Foundation, a pro-trade group, China will probably spend $1.4 trillion on semiconductor self-reliance over the next decade, far exceeding the Phase II spending announced so far. 

Critical Need: Semiconductor Fabrication Equipment

For complete self-reliance in semiconductor production, Chinese semiconductor fabrication facilities (fabs) need domestic suppliers of semiconductor equipment, domestic field support, and domestic suppliers of maintenance parts and starting materials such as silicon substrates and gases.

China has more semiconductor fabs than any other country, and has approximately 30 new fabs in construction or on the drawing boards. Last year, the global industry trade association SEMI predicted that China would spend $14 billion on semiconductor equipment in 2020, making China the largest buyer of such equipment. However, the U.S. has imposed export controls on the sale of U.S. design tools and manufacturing equipment to semiconductor producers in China. 

Photolithography equipment is probably the most difficult to produce.  No domestic Chinese-made machines are yet capable of lithography for 28 nm CMOS.  Photolithography machines typically cost about $50 million each, and make up about 25 percent to 30 percent of fab equipment costs.  In addition, photolithography operations typically make up about 50 percent of semiconductor manufacturing time, so the machines are critical in determining fab throughput and cost per wafer and per die.

Photolithography uses challenging and expensive technology.  Photolithography at 28 nm requires machines using a powerful argon-fluoride excimer laser to produce 193 nm deep ultraviolet light, plus immersion lithography, which illuminates the wafer while immersed in a transparent fluid to alter the index of refraction and bring fine lines and features into focus.   

Chinese fabs have had notable success in producing LEDs, low-power processors, sensors, discrete semiconductor devices, and in assembly, packaging, and test (APT).  However, China remains behind in multi-core processors and memory devices, semiconductor design tools and equipment, especially at the smaller, leading-edge process nodes needed for the highest levels of integration in SoCs (system on chips) for cellphones, for example.  China also lags somewhat in analog / mixed signal ICs and RF (radio frequency) front-end components for cellphones such as power amplifiers and RF filters.  Full self-sufficiency in all of these will require China to produce even more advanced semiconductor equipment than what is required at the 28 nm CMOS node.

It remains to be seen whether European and Japanese equipment suppliers will follow the U.S. lead, or will risk jeopardizing ties with Washington to continue to sell equipment to China.  Equipment suppliers around the world are pushing back against the sanctions, as these companies stand to lose billions of dollars in sales.

The good news for China is that Shanghai Microelectronics might announce its first 28 nm-capable machine as early as December of 2020 according to news stories.  The company has not confirmed the date officially, but it is certain that there is intense interest in such a machine, which should allow China to produce its own 28 nm ICs with little or no reliance on companies outside of China. 

US Reactions to Phase II

Regrettably, the latest investments in semiconductor independence will fan mistrust of China in the U.S., where politicians, foreign policy, and defense analysts see China’s investment in 5G, AI, quantum computing and semiconductors as part of a plan for world domination, and a threat to U.S. technology leadership.  These fears are overblown, but they persist and have intensified of late.

Semiconductor suppliers in the U.S. have over the past decades outsourced production to avoid the high cost of building new, state-of-the-art fabs.  Today, only a handful of U.S. suppliers still produce their own chips in house, most having gone fabless except for producing older chip designs.  

As a consequence of this, the U.S. has begun its own efforts at semiconductor independence led by the Defense Advanced Research Projects Agency (DARPA), with its $1.5 billion Electronic Resurgence Initiative (ERI) program. Congress has gotten involved by cooking up the CHIPS for America Act (Creating Helpful Incentives to Produce Semiconductors), a proposed $17-plus billion bill that aims to revive U.S. chip production, fund R&D, and secure supply chains using investment and generous tax incentives.

Although the Trump Administration can be blamed for imposing crude trade policies that have caused collateral damage, sanctions against China now have bipartisan support, and will not simply go away with a new administration after the 2020 U.S. presidential election.  The best China can hope for right now is better-considered, more predictable U.S. trade policies going forward.

The Best Path Ahead

China will attain its goal of independence at the 28 nm node, and probably within two years, which should reassure customers of China’s electronics industry.  However, attaining complete production independence across all semiconductors, including leading-edge ICs, will depend on some easing of U.S. export controls on design tools and equipment especially.  This can only happen if China and the U.S. engage in good-faith negotiations that lead to improvements in mutual trust and trade policies.

The central challenge to both the U.S. and China is that the semiconductor industry is truly global in nature; no single company or country can produce most efficiently without selling into global markets, and efficient production requires specialist companies in various parts of the supply chain located in other countries. 

So, it is a balancing act. China can still benefit from semiconductor help and selling into the markets of the outside world, even while securing its own semiconductor supply base.  And, the U.S. will still need China’s markets, strong APT and electronics assembly and production capabilities, even while seeking to protect what it deems as sensitive intellectual property and technologies.  We hope a good balance can be found.

You can download a PDF copy of this blog here.