The “Choke Point” Struggle Between Rare Earths and Semiconductors: Insights from Over a Decade of Diverging Paths Between China and Japan

After the Diaoyu Islands dispute in 2010, Japan regarded “reducing dependence on Chinese rare earths” as a national task. More than a decade later, the statistical proportion of “raw materials from China” has indeed decreased, but the dependence at the value chain level has not weakened; instead, it has become deeply coupled with China in the higher value-added segments of magnets, motors, and complete machines. In 2024, China is expected to export about 58,000 tons of rare earth permanent magnets, while holding approximately 90% of the global market share in the separation refining and magnet manufacturing stages; if the licensed “valve effect” tightens, the delivery rhythm and costs of the entire chain will be immediately rewritten.

In contrast to rare earths, semiconductors are often referred to as the “jewel in the crown of industry,” seemingly possessing greater “long-term choke point” power. However, when examining the practices of China and Japan over the past decade, the answer is not simple. Semiconductors are a complex system with multiple links and multiple centers: design (EDA), manufacturing (various process nodes), equipment and materials, packaging and testing (OSAT) are distributed across multiple countries, with the “chokeable” links mainly concentrated in EUV lithography and top-tier EDA, while the vast majority of global applications use mature processes. The United States, the Netherlands, and Japan indeed hold key control in cutting-edge segments, but this control does not equate to a “master valve” over the entire industry chain. On the contrary, in recent years, China’s rapid expansion in mature processes and packaging has significantly diluted the possibility of a “complete choke point.” According to public estimates from the Semiconductor Industry Association (SIA) and the U.S. Bureau of Industry and Security (BIS), China’s share of mature process (28nm and above) capacity has increased from about 19% to approximately 33%, and investment is accelerating; policy departments also predict that the global share transfer in the mature process field will continue to occur.

Why, after a decade, does Japan have a stronger real dependence on China in the rare earth/magnet chain, yet has not formed a “reverse choke point” in semiconductors? The core issue lies not only in “whether it can be replaced” but also in where the scalable demand centers and large-scale trial-and-error spaces are located. The explosive growth in demand for high-performance NdFeB from electric vehicles, wind power, and industrial control has led to the simultaneous emergence of demand and supply centers in China, naturally concentrating the shortest chain of magnets, motors, and complete machines in China. The IEA’s “Global Electric Vehicle Outlook” indicates that in 2024, electric vehicle sales in the Chinese market will account for nearly half of domestic passenger car sales, continuing to raise global penetration rates; this forces Japanese motor/magnet companies to form a structure of local production—local consumption—spillover in China, thereby “welding” the value chain focus in China.

At the same time, Japan does not possess a complete set of “single-point choke” chips against China in semiconductors. Japan does have advantages in materials and equipment (wafers, photoresists, deposition/etching/cleaning equipment, etc.), but the equipment export controls introduced by Japan in 2023 adopt a case-by-case licensing approach, coordinating with the U.S. while avoiding a “one-size-fits-all” strategy, which is precisely due to industrial interdependence: the regional revenue structure of leading Japanese manufacturers is highly exposed to the Chinese market. Tokyo Electron (TEL) expects that in the 2024 fiscal year, the proportion of revenue from China will rise to about 44%, and it will remain at a high level in the 2025 fiscal year; for such companies, a complete closure would mean self-harm. This is also a key constraint for Japan’s policy choice of “adjustable valves rather than permanent gates.”

From the perspective of “choke point forms”, the essential differences between rare earths and semiconductors determine the effectiveness of policy measures:

• Rare earths/magnets are a “high concentration + low substitutability + local consumption” physical choke point. Separation refining and magnet manufacturing are highly concentrated in China, making short-term substitution difficult; while the downstream of magnets (motors, drive assemblies) has formed a super-large market in China, once the licensing rhythm changes, multiple industries globally will simultaneously “lack magnets.” In 2024, the IEA explicitly states that rare earth supply is among the lowest in geographical diversification, and delays or denials of licenses for China’s rare earth magnet exports are sufficient to change the rhythm of the global industrial chain.

• Semiconductors are a “multi-link + regional advantages + demand can be layered” system choke point. Cutting-edge (EUV/top-tier EDA) can be tightened, but the vast majority of end products (automotive MCUs, power devices, industrial and home appliance controls, and some generations of communication basebands) can be processed on mature nodes; while the packaging/testing segment accounts for the largest share globally in mainland China and Taiwan, with local OSAT leaders (such as JCET) firmly in the top three worldwide. Therefore, control is segmentally effective: it can precisely choke high-end chips in a few chains such as HPC/AI, but it is difficult to choke the broad industrial applications as a whole.

Looking at cross-chain coupling: semiconductors themselves are not “without chokeable minerals.” Gallium/germanium and other key materials were included in China’s licensing management in 2023, and graphite will be implemented with licensing at the end of 2023, directly affecting adjacent industries such as power devices, optoelectronics, and battery anodes; high-precision motors/lasers/magnet components in semiconductor manufacturing equipment are also influenced by the rare earth chain. Rare earths and semiconductors are not two parallel tracks, but rather a strategic network with multiple intersections.

Breaking down the “choke point strength” into three dimensions can more clearly answer the question of “can semiconductors choke more than rare earths?”:

1. Concentration and substitutability: The single concentration of rare earth separation/magnet manufacturing is extremely high, and short-term substitution technology paths are limited; semiconductors are concentrated in cutting-edge and dispersed in mature processes. In terms of “overall choking ability for broad-spectrum industries”, rare earths/magnets are stronger.

2. Expansion cycle and trial-and-error space: China’s capacity expansion in mature processes and packaging is faster than the global demand growth rate, combined with a super-large domestic market, forming large-scale trial-and-error and rapid climbing; cutting-edge segments are restricted by EUV/EDA, but system-level engineering (chipletization, advanced packaging) and “lateral innovation” at the architecture/algorithm level can still release some performance space. In terms of “diluting choke point expansion speed”, the non-cutting-edge areas of semiconductors are more buffered.

3. Geographical demand and industrial locking: The large-scale commercialization of new technologies such as electric vehicles, energy storage, and photovoltaics mainly occurs in China: electric vehicle sales and penetration rates are the highest, and over 80% of photovoltaic manufacturing capacity is concentrated in China. The co-location of demand centers and supply centers makes the rare earth/magnet—motor—complete machine present an internal circulation, making it increasingly difficult for external policies to change the de facto availability pattern.

Returning to the experiences of China and Japan, why after ten years of Japan’s “de-reliance,” has there been a statistical “reduction” but a real-world increase in dependence? The answer lies in the calibration of statistics and geographical reconfiguration:

• At the raw material level, by signing contracts and investing in places like Australia, Japan has reduced the proportion of “direct imports from China”;

• At the capacity level, key manufacturing links from alloys to magnets to magnetic components to motors have been outsourced or established in China and ASEAN, and then cross-border settled as components/complete machines, creating a facade of “diversified sources” while essentially being close to China’s super cluster;

• At the market level, China has become one of the largest demand centers for motors and complete machines, with local consumption + local matching forming the optimal solution for returns/risks.

  Policy-wise, starting in 2025, China will further upgrade licensing and cataloging for rare earths/magnets, forming a more flexible “valve governance” for the chain; Japan, on the other hand, adopts licensing rather than embargoes in semiconductors, balancing its exposure to revenue from China with industrial security demands. In contrast, the magnet chain is more like a natural fit for “valve control,” while semiconductors resemble a “segmented control” system engineering.

Therefore, “can semiconductors choke more than rare earths?”—for top computing power and specific dual-use devices, yes; for broad-spectrum industries and large-scale production of complete machines, no. In the track where performance/cost optimization can be achieved using mature processes + advanced packaging + system engineering, market scale and manufacturing clusters determine who can better “choke” the delivery rhythm and unit costs. Currently, this advantage is more concentrated in China: the penetration rate of electric vehicles, the full-chain cluster of photovoltaics/batteries, and the accelerated expansion of packaging and mature processes, combined with the structural control of rare earths and magnets, make the reality of “seemingly de-reliance, but actually deeply embedded” a common reality for Japan and more industrial countries.

Looking to the future, what is truly important is not the slogan of “de-reliance,” but the identification of segmented choke points and the design of combinable redundancies:

• View cutting-edge semiconductors as a “long-wave buffer + technical breakthrough” issue, gradually narrowing the gap through multinational cooperation and local alternatives;

• View rare earths/magnets as a “valve governance + industrial agreement” issue, establishing predictable mechanisms in licensing, recycling, and HRE-lean/HRE-free processes;

• At the system level, use advanced packaging, heterogeneous integration, and power semiconductors as engineering solutions to reconstruct the triangular balance of “performance—cost—availability.”

The historical crossroads of the past decade between China and Japan provides a lesson in “structure”: statistical de-reliance does not equate to real disengagement; rather, the combination of demand centers + manufacturing clusters + valve governance is the decisive factor in the “choke point strength” in the real world.

Supplementary: Can the U.S., Japan, and the Netherlands encircle and “choke China”?

The core strategy of U.S.-Japan-Netherlands collaboration is a “combination punch” at both ends of cutting-edge computing and key equipment: the U.S. has set licensing and thresholds for advanced computing chips and semiconductor manufacturing equipment based on the two rounds of rules in October 2022 and October 2023, continuously updating loopholes with FAQs, notifications, and subsequent page updates; the Netherlands has limited ASML’s shipments and services to China for some EUV/high-end immersion DUV through a licensing list and case-by-case revocation; Japan has included 23 types of semiconductor manufacturing equipment in prior licensing, aligning with allies through a “list + fallback” approach. In terms of limiting the rise of cutting-edge capabilities, this encirclement has indeed shown effectiveness: the long-term blockade of EUV combined with the “valve effect” on some high-end DUV, and the performance thresholds and notification rules set for AI/HPC chips have significantly extended the cycle for China to acquire cutting-edge processes—chips—systems.

However, the question of whether it can choke the overall system does not equate to a simple “yes.” First, the layered structure of capacity makes “complete suffocation” engineering infeasible: the vast majority of global end products still operate on mature processes and power device tracks, and the equipment—materials—processes in these areas are more substitutable and expandable. Public reports from the Semiconductor Industry Association and official sources indicate that from 2015 to 2023, China’s share of mature processes increased from about 19% to approximately 33%, expanding at a rate higher than global demand; BIS’s public reports also document the marginal impact of China’s mature capacity expansion on prices and supply patterns. In this context, even if the cutting edge is “choked,” broad-spectrum applications are unlikely to be entirely locked down.

Second, the revenue structure of the market and equipment manufacturers limits the political feasibility of a “permanent closure”: ASML still regards China as its largest/most important single market in 2023-2024, and while licensing is tightening, the proportion of sales and services in China remains high; Japanese equipment leader Tokyo Electron (TEL) disclosed that the proportion of revenue from China in the 2024 fiscal year exceeds 40%, reflecting the tension between its commercial exposure and policy orientation. The further service and spare parts licensing from the Netherlands and the U.S. to ASML are also mostly based on limited scope and controllable impact, indicating that the policy focus is more like an “adjustable valve” rather than a “master valve.” This means that while the encirclement is likely to continue to hinder China’s cutting-edge leap, it is neither realistic nor feasible to achieve a stable consensus among allies to completely sever the manufacturing—packaging—system application across the entire spectrum.

Third, the geographical co-location effect of demand centers and large-scale trial-and-error spaces is redistributing the “choke point strength” to non-cutting-edge areas: as the incremental growth of electric vehicles, energy storage, and industrial control mainly occurs in China, the system engineering solutions of mature processes and advanced packaging are sufficient to stabilize the triangular balance of overall performance—cost—delivery. In such a structure, the U.S.-Japan-Netherlands encirclement can effectively delay China’s leap in cutting-edge technology, but it is difficult to fundamentally rewrite the availability pattern of the “China-centered mid-to-low/mid-high-end manufacturing and application ecosystem”; on the contrary, the longer the cutting edge is restricted, the more the capacity and ecology of the mature areas will further concentrate on the Chinese side.

In summary, the U.S.-Japan-Netherlands collaboration can continue to “tighten the throat” (cutting-edge nodes and AI/HPC), but it is difficult to “cut off oxygen” (manufacturing across the entire spectrum and broad-spectrum applications) in the long term. The realistic outcome will resemble a segmented balance: one end shapes the “long slope” of China’s cutting-edge path through EUV/high-end DUV/EDA thresholds and licensing systems, while the other end continues to be driven by the rolling expansion of demand for mature processes—packaging—complete machines in the Chinese market. Any judgment on “can it choke China” must measure the three-layer structure of cutting-edge—mature—system separately to avoid misreading “frontier delays” as “overall suffocation.”

References and data sources (excerpt): IEA’s interpretation of 2024 rare earth magnet capacity/export and concentration; Japan’s case-by-case licensing path for semiconductor equipment controls in 2023; assessments by SIA (2025) and BIS (2024) on China’s mature process expansion and share transfer; Tokyo Electron’s IR disclosure of revenue proportion from China; IEA’s “Global Electric Vehicle Outlook 2025” and “Photovoltaic Supply Chain” quantifying China’s demand centers and manufacturing clusters.