China’s Mature Semiconductor Overcapacity: Does It Exist and Does It Matter?

Introduction

There is an ongoing debate amongst U.S. and allied policymakers about the overcapacity of foundational semiconductors in China. The answer to this question is complex, and experts have reached different conclusions when estimating supply and demand. Rather than waiting for a definite answer—which might not be available for many years—policymakers should be prepared to answer three questions:

  1. Is future “overcapacity” the right question to discuss? Or should the focus rather be on long-term strategic overreliance as well as economic harm caused in the short term by nonmarket practices?
  2. Once both long- and short-term policy goals have been defined, do governments have sufficient market transparency tools and expertise to monitor risks and act timely?
  3. Are policy instruments available to address strategic overreliance and unfair nonmarket practices?
     

Foundational Chips Are the Backbone of Modern Economies

Foundational semiconductors, also referred to as “legacy chips” or “mature chips,” are generally defined as semiconductors produced at manufacturing nodes 22 nanometers (nm) and larger. While advanced semiconductors are predominantly used for cutting-edge computer applications in datacenters and consumer devices, foundational semiconductors are the backbone of modern economies.

A modern car contains hundreds of foundational chips, controlling everything from engine electronics and seat adjustments to the infotainment system. No hospital or industrial facility functions without them—in fact, any electrical device with an on/off switch depends on foundational semiconductors.

The supply chains for foundational semiconductors are more varied than for advanced semiconductors. Design and wafer manufacturing takes place in the United States and allies such as Europe, Japan, South Korea, Taiwan, and Singapore. China is rapidly expanding its own industry, as technology, design, and manufacturing expertise have long been more widely available. Foundational semiconductors (except for dual-use technology and products) have not been the focus of recent export control restrictions to the degree advanced semiconductor technology has.

Some types of foundational chips have been available in the market for many years, leading providers to continue to invest heavily in research and development (R&D) for further innovation. In addition to large R&D expenditures, long-term commitments to capacity expansion are needed to fulfill the growing demand for foundational chips throughout the decade. The U.S. Chips and Sciences Act of 2022, the European Chips Act of 2023, and similar undertakings in Japan and Korea recognized this and provided government support to strengthen and expand high-tech industrial ecosystems.

China made semiconductors a centerpiece of its industrial policy a decade ago, as part of the “Made in China 2025” agenda. Since then, massive investments and subsidies have been provided, and China made impressive progress towards its goal of self-reliance.

With Chinese industrial overcapacity discussed at the highest level, U.S. and allied governments now need to assess whether foundational semiconductors will follow the path of the photovoltaic and electrical vehicle industries—and if so, what to do?
 

No Expert Consensus on the Threat of Overcapacity

Several think tanks and analysts published their findings in the spring of 2024. The analytical approaches differed greatly: some were more quantitative, and others used predominantly qualitative arguments. Not all authors took a clear position but generally acknowledged the complexity of the question. A consensus on the question of overcapacity did not emerge.

For example, a report by the Institute Français des Relations Internationales (April 2024) uses one leading Chinese semiconductor manufacturer, Semiconductor Manufacturing International Corporation (SMIC), and its regional end market split, as a proxy for Chinese foundational semiconductor supply. The demand analysis focuses on three applications (smartphones, automobiles, and the Internet of Things). While the report’s title suggests that overcapacity fears are “unfounded” it warns that “as China increasingly fulfills its own demand, global prices may fall from gluts of Western chips crowded out of the PRC.” The author concedes, “Whether China’s domestic downstream demand for the chip end-uses . . . continues to expand will decide whether the PRC does or does not flood the global mature-node chip market.”

An analysis by Rhodium Group (May 2024) takes a different methodological approach and considers domestic supply in China more broadly: it looks at current and planned wafer manufacturing capacity by process node and highlights Chinese semiconductor manufacturers dominating global capacity investments in the important 20–40 nm node. Demand estimates are not the focus of this analysis. It quotes the market researcher TechInsights with a self-sufficiency rate for the overall Chinese semiconductor industry of only 12 percent in 2023. This number, too, would not suggest a material overcapacity risk in the next years.

A commentary by CSIS (April 2024) uses mostly qualitative arguments. While the author is skeptical about claims of overcapacity risks in general, he ultimately concludes that “In determining if there are valid and sufficient reasons to believe that by 2030 there will be anything resembling overcapacity, governments must be very careful and fully understand the complexities of the industry, on both the supply and demand sides.”

Of the reports which do call out an oversupply risk the one by Bernstein Research (May 2024) is the most analytical. Using sophisticated investment, supply chain, and productivity models for supply, and projected sourcing power by Chinese original equipment manufacturers (OEMs) for demand, the research concludes that “It would take another c. 4 years for China to ‘fully’ meet its own need in Mature Logic if China’s annual spending stays as high as [2023].”

The author and scholar Chris Miller takes a clear position, too: In congressional testimony and a report by the American Enterprise Institute (June 2024), he finds that “China’s rate of building new fabs threatens the profitability of Western firms. Of the world’s fabs under construction, fully a third (measured by wafer capacity) are in China.” He also warns that “Current trends—above all, China’s vast subsidy campaign and nonmarket practices—imply that without policy changes, the U.S. manufacturing base will become more reliant on Chinese-made chips.”

Why could experts not reach a consensus, and where does this leave policymakers in Western capitals?
 

A Complex Question Both on the Supply and Demand Side

Focusing on supply first—semiconductor factories are large industrial facilities, their construction is typically announced by companies and local governments alike and projections of future capacities are readily available. However, three reasons make the supply projections difficult:

  1. Even after a new factory has been announced, companies will adjust capacity expansion to short- and mid-term market outlook. The softness of the semiconductor market in the first half of 2024 led companies to stretch or cancel planned investments—a fact that was also acknowledged by SMIC’s CEO in a recent earnings call with financial analysts: he indicated that current oversupply conditions would lead SMIC to take a more cautious approach toward capacity expansion.
  2. The productivity of Chinese foundational semiconductor fabs varies greatly. One industry executive classifies local capacity into three categories: (1) unproductive or even mothballed, (2) small and below Western efficiency (but with potential for future consolidation), and (3) globally competitive. For outside observers, it is difficult even to assess the current aggregate capacity at a national level, let alone project it 510 years into the future.
  3. The diversity of products—from microcontrollers to discrete power transistors, wireless transceivers to memory—and manufacturing technologies requires a granular market understanding
     

Both the U.S. Department of Commerce Bureau of Industry and Security and the EU Commission conducted industry surveys to assess the industry’s status quo, with feedback requested—mandatory in the United States and voluntary in the European Union—by April and September 2024, respectively. Findings have not been made public but should provide a starting point for a deeper understanding.

Assessing the Chinese semiconductor demand—as a starting point to observe excess capacity—is even more challenging.

China is the world’s largest manufacturer of microelectronics. Therefore, shipments of semiconductors into China do not reflect in-country demand. As the CEO of one European foundational semiconductor manufacturer stated publicly: “about half of [the products shipped to China] are processed and then re-exported to Western buyers.”

Using the demand of Chinese microelectronics manufacturers as a proxy is equally misleading for two reasons. U.S. and European manufacturers, too, are present in China to serve local demand. More importantly, though, looking at Chinese OEM demand simply pushes the discussion about overcapacity one step down the value chain: a (hypothetical) OEM with 100 percent global market share for a particular application would also demand 100 percent of the semiconductors for it—far from an accurate reflection of “domestic” demand.

An informed, granular demand analysis would have to be based on end-consumer demand, today and in the future, for microelectronic devices. Given the omnipresence of semiconductors in microelectronics, the author estimates that approximately 80 applications would have to be analyzed in detail for a full demand picture, and even just covering 85 percent of the estimated domestic market for foundational semiconductors would require insights into 30 applications, including 10–15-year demand projections.

Focus on Overreliance, Nonmarket Practices, and Preparedness

Instead of—or at least in parallel to—studying supply-demand models for foundational semiconductors, U.S. and European policymakers need to focus on three issues: overreliance, market distortion, and preparedness.

Overreliance: Overreliance on any one country or geography for a product as critical as foundational semiconductors can pose grave national and economic security risks. Establishing a baseline of minimum U.S. and allied demand for foundational semiconductors (and the supply to match) and ensuring transparent and resilient supply chains for critical applications is the more pertinent task. The United States’ recently proposed rule for “Securing the Information and Communications Technology and Services Supply Chain” provides an example of the latter.

Unfair practices by individual companies: One industry executive estimated that some Chinese foundational semiconductors are sold up to 30 percent below Western companies’ prices. This might be possible due to lower profit expectations of Chinese investors and WTO-compliant public funding but could also hint at unfair subsidies. U.S. and allied governments have trade tools at their disposal to address nonmarket practices, as the European Union’s targeted and nuanced duties against unfair subsidies on electrical vehicles from China show.

Policymakers need to be prepared: Lastly, European policymakers especially need to be prepared to act—trade restrictions and industrial policy will prevail for the foreseeable future. Semiconductors are going to remain a centerpiece of many countries” industrial policies and trade restrictions. Developing suitable and compatible toolsets for trade remedies amongst allies is as important as partnering and aligning industrial policies.

Andreas Schumacher is a visiting technology fellow in the Economic Security and Technology Department and the Scholl Chair in International Business at the Center for Strategic and International Studies in Washington, D.C.

The author is grateful to William A. Reinsch, senior adviser and Scholl Chair in International Business, for valuable discussions and suggestions on this paper.

Image
Andreas Schumacher

Andreas Schumacher

Former Visiting Fellow, Economics Program and Scholl Chair for International Business