Evaluating Chip Overcapacity

Geopolitical and technological competition with China over semiconductors has focused overwhelmingly.

By Emily Benson, Catharine Mouradian, and Pau Alvarez-Aragones

A topic of growing prominence in the transatlantic policymaking arena is to what extent Chinese overcapacity of legacy (“mainstream”) chips is or will become a problem and, if it is a problem, what to do about it. European Commission vice president Margrethe Vestager said at a CSIS event on the sidelines of the U.S.-EU Trade and Technology Council (TTC) summit in January 2024 that the transatlantic alliance will “follow very closely what happens with the production capacity and the global market on mature technology.”

The European Union is considering pursuing an information-gathering process similar to the Bureau of Industry and Security’s Mature Node Semiconductor Assessment, which surveys companies’ mature-node supply chains. The April 2024 meeting of the TTC similarly emphasized that both the United States and European Union are gathering more information about the current state of play of the global legacy chip marketplace and that the parties seek to institutionalize early-warning alert systems, which they regard as having been useful in the context of Chinese export restrictions on gallium and germanium. It is clear that the transatlantic alliance is seeking more information and intends to move jointly if it determines that China’s actions warrant a trade tool response. The first step, however, is agreeing on the nature of the problem, and that is far from clear.

“Legacy” Chips

Geopolitical and technological competition with China over semiconductors has focused overwhelmingly on the most advanced semiconductors. The October 7, 2022, export controls identified chips under 16 or 14 nanometers (nm) as critical to national security. The guardrails established by the CHIPS and Science Act separately identified the national security importance of blocking funds from reaching “foreign entities of concern,” seeking to halt expansion of funding for recipients’ facilities in China. The guardrails create an exception for existing legacy semiconductor manufacturing or new facilities producing legacy chips. However, the United States is broadening the scope of critical semiconductors to include both leading-edge chips as well as legacy or “foundational” chips. This commentary uses “foundational,” “mature,” “legacy,” and “mainstream” interchangeably so as not to favor one or the other since each term has a unique set of political connotations.

Legacy chips, defined in the CHIPS and Science Act as 28 nm or larger, cover a wide array of chips and uses. They are embedded in cars, military equipment, refrigerators, energy and telecommunications infrastructure, phones, televisions, and other consumer electronics—making them a vital cornerstone of the U.S. economy. However, even though they are essential to thousands of products, they have not been the focus of recent investments. Before the CHIPS and Science Act passed, “less than $1 of every $6” of private semiconductor investment was intended to support legacy chips. Only $2 billion of the $39 billion in the CHIPS and Science Act was originally earmarked for mature nodes.

Chips Priorities  

Similar to advanced chips, the United States lacks major manufacturing capacity for legacy chips. In terms of global foundry capacity, China and Taiwan hold 80 percent of the manufacturing capability for chips from 20 to 45 nm and 70 percent for chips from 50 to 180 nm. The U.S. government astutely left the CHIPS and Science Act’s goals ambiguous, believing that artificially identifying hard policy targets would constrain its ability to adapt to a dynamic and changing geoeconomic environment.

Given accelerating fears about potential overcapacity of legacy chips from China, this flexibility has provided the Department of Commerce with added agility to revert funds to legacy nodes. The CHIPS and Science Act stipulates investments of at least $2 billion in legacy chips, although Secretary of Commerce Gina Raimondo has confirmed the department’s intention to enlarge legacy chip investments. The CHIPS Program Office has so far announced funding for four companies. The first three rounds were slated for projects related to mature chips. The Department of Commerce awarded a first grant of $35 million in December 2023 to BAE Systems’ Microelectronics Center, which produces legacy chips for the U.S. military. The department awarded the second grant of $162 million to Microchip Technology for added capacity of critical legacy-node chips. GlobalFoundries is also receiving $1.5 billion to support both legacy and current-generation chip manufacturing. More recently, the department awarded Intel $8.5 billion for several projects at the leading edge.

Chinese Overcapacity Concerns

Fears are accelerating about China’s potential weaponization of legacy-node chips. These fears are primarily threefold: (1) that China could create a dependency and cut off supply, creating both military and broader economic vulnerabilities; (2) that price competition would undercut the value of chips subsidy packages in partner countries; and (3) that Chinese chips could contain backdoors and should thus be excluded from military uses.

Many long-time trade experts fear that Chinese overcapacity will eventually become a problem, similar to what has transpired in steel and solar markets. What is less certain, however, is whether this perceived inevitability is likely to be problematic on a short or long timeframe—or whether it will become a problem at all, particularly if Chinese domestic demand is able to soak up most of the excess capacity. In recent CSIS conversations with industry leaders and U.S. and foreign government officials, there is no consensus on the risks of overcapacity, but there is broad recognition of parallels with other supply chains.

From 2008 to 2015, China fostered new projects and expanded production capacity in strategic sectors such as steel, aluminum, and chemicals. The former head of the European Union Chamber of Commerce in China noted, “China is always enticing industries to grow. The system breeds overcapacity.” The chair of the American Chamber of Commerce in China predicted in early 2024 that widespread industrial overcapacity will most likely materialize again, particularly as Chinese businesses look for other streams of revenue as the real estate sector falters.

China’s quest to reduce its reliance on foreign chipmakers underpins China’s bid to add more legacy semiconductor manufacturing capacity in 2024 than all other countries combined. China anticipates adding 13 percent additional forecasted capacity this year, with 18 semiconductor fabrication facilities (called “fabs”) slated to launch in 2024. The Chinese government also provided nearly $1.8 billion in direct subsidies in 2022 to 190 Chinese chipmakers, along with loans to advance China’s “self-sufficiency” strategy.

President Xi Jinping declared technological self-reliance a top priority during his initial address at the 20th National Congress of the Chinese Communist Party that year, barely a month after the U.S. October 7 export controls, meaning the U.S. export controls could have pushed China to redouble its self-reliance strategy, already outlined in its 14th Five-Year Plan.

During his meeting with New Zealand’s then prime minister Chris Hipkins, President Xi emphasized that China’s self-reliance “is by no means to adopt a closed-door policy, but to better connect domestic and international markets.” However, as a new classified Chinese order called “Document 79” discloses, this open and connected self-reliance would specifically aim to “delete America” from Chinese devices.

For instance, Huawei is engaging China’s largest chip manufacturer SMIC, as well as Baidu, to promote the creation of 5 nm chips. However, it remains to be seen whether Huawei has sufficient capacity to absorb the market share of established firms such as Samsung or TSMC. Additionally, Chinese firms would still depend on key processes such as ASML’s lithography equipment, the most advanced versions of which have been barred from entering China. ASML’s less-advanced deep ultraviolet (DUV) lithography systems are also subject to Dutch export controls. Even if China stockpiled a sufficient supply of Dutch lithography machines, China would still rely on ASML’s tools—and the servicing of those tools—to produce additional advanced units at scale.

Additional Overcapacity Factors

Growing Chinese efforts to “de-risk” from foreign imports of legacy-node chips is not the only reason that could explain potential overcapacity problems. Another factor is overcorrection following supply chain disruptions during the Covid-19 pandemic. Even as early as 2022, it was assumed that increases in production and higher inventory would inevitably lead to a glut. Fitch Ratings warned in March 2023 that excess inventory would pose a greater risk to the industry than the October 7 export controls. Covid-19 supply chain crunches have exacerbated concerns about overcapacity and highlighted the U.S. government’s push toward building more resilient supply chains that are less susceptible to disruptions, whether or not they are geopolitical in nature.

The semiconductor industry also stands out for its cyclical nature. Since the 1950s, the chip industry has experienced a regular pattern of high growth followed by overcapacity and big busts for companies and investors. In the 1990s and early 2000s, chip producers expanded capacity in response to demand for consumer electronics, which then resulted in overcapacity when global demand for expensive products such as phones and computers decreased during wider global downturns. However, even with more consistent demand for chips, the industry remains prone to cyclical downturns. One reason is that once a fab increases capacity, it will often keep producing at that output level even if demand falls, contributing to price drops.

What the Data Show

A major hurdle to assessing whether legacy chip overcapacity is a problem now or if it will become problematic in time is the lack of available data. For example, there are no specific Harmonized Tariff Schedule (HTS) codes dividing chips into “foundational” versus “cutting-edge.”

The current industry standard regards the four-digit code 8541 as representing legacy chips and 8542 as advanced chips. This division has issues, particularly since the article descriptions for codes 8541 and 8542 do not align with the definitions of foundational and cutting-edge chips. Perhaps the most precise is the six-digit 8542.32, which covers memory chips. This then further specifies megabit storage capacity in eight-digit codes, which could be used to estimate how “advanced” chips are. HTS specificity would help produce more transparent trade data that would enable better policymaking from transatlantic leadership.

An additional issue in tracking foundational chip imports exists as many enter the United States not as shipments of individual chips, but as components of finished goods. This makes examining chip imports through HTS codes less useful, as it only captures a fraction of the legacy-node chips entering the country. Another major problem is how to define what constitutes a Chinese chip and whether or not a rules-of-origin calculation may be valuable in determining which tariffs to apply to which products. Given the complexities of applying new tariff rates to both chips and finished goods containing chips, any new tariff action would likely need to occur at the wafer level.

Trade Considerations

It is not a foregone conclusion that China will weaponize legacy chips, particularly if Chinese domestic demand remains high. Nevertheless, the United States and its partners are beginning to assess which trade tools their governments could use either preemptively or as a way to combat future overcapacity. The United States has many tools at its disposal, although none are particularly well designed to confront prospective problems.

Section 301 of the Trade Act of 1974 is intended to remedy injurious trade practices, the root cause of which are foreign government policies. Section 232 of the Trade Expansion Act of 1962, on the other hand, provides statutory authority for the United States to apply tariffs on national security grounds. The distinction between these two statutory tools is—simply put—whether the United States would invoke an economic security tool or a national security one. Neither of these remedies is particularly forward-looking, leading some to question their applicability in current circumstances.

Another option is to build a new regime that functions similarly to the Uyghur Forced Labor Prevention Act. This new regime could function on either a sectoral or a firm basis, which would invite distinct compliance considerations throughout the semiconductor ecosystem. This structure would essentially function to price in geopolitical risk and functionally serve as a resiliency tax.

Next Steps for Policymakers

Policymakers on both sides of the Atlantic need better access to better data. This places a burden on the private sector to share relevant data with policymakers so they can make more informed decisions. In a geopolitical and economic environment with so many unknowns—especially a political environment that makes preemptive trade remedies even more appealing—it is still best to operate with the most advanced knowledge set possible, even if that means waiting.

If the United States proceeds with any of these tools, it will be critical to consult sufficiently with allies in advance. For example, it is not difficult to envision the complexities that would arise if the United States were to implement a tariff on inbound items containing Chinese chips—for example, affecting a Korean vehicle with chips produced in partner countries and also in China.

Furthermore, a significant U.S. tariff increase could push Chinese legacy chips into other foreign markets, including allied economies, undercutting their own attempts to bolster domestic chip supplies. Any action taken will be best served by strong international buy-in. Institutionalizing information sharing may seem insufficient, but in reality, it is foundational to longer-term success.

---