When the world's at stake,
go beyond the headlines.

In 2025, the ongoing contest between the United States and China moved from tariff skirmishes to a confrontation over strategic choke points.

Washington expanded semiconductor export controls under the Biden administration, limiting China’s access to advanced chips, design software, and lithography tools. The Trump administration tightened these rules on Sept. 29, extending them to foreign affiliates. Beijing retaliated ten days later with new licensing requirements on rare-earth oxides, metals, and magnet products. The escalation jolted global markets and forced emergency consultations in Busan, South Korea, where both sides agreed to a tentative one-year suspension of their measures. The pause quieted financial markets but altered none of the fundamentals. Both sides understood they would revisit the one-year freeze in 2026.

This 2025 export control showdown can be examined through one question: If both sides had pushed their choke points to the limit, whose leverage would have lasted longer? Measured by five dimensions — durability, replaceability, precision, feedback, and sustainability — America’s semiconductor choke point cuts deeper and endures longer than China’s rare earth ban. This contrast reveals a broader logic of modern economic statecraft: Power now depends less on who can create choke points than on who can sustain them.

The Rare Earth Burn: China’s Single-Use Weapon

To see the logic of renewable versus one-off leverage in practice, it is useful to begin with China’s rare earth pressure and the short, front-loaded power it provides. In the first dimension of durability, China’s leverage is limited because it is most effective at the start and depends on immediate disruption. Price spikes prompt governments to intervene, reduce investment risk, and ease the flow of private capital, making extraction and refining projects that were previously too costly suddenly attractive. After the 2010 China–Japan dispute, for example, rare earth prices rose sharply, with some estimates suggesting a 10-fold increase. Tokyo diversified suppliers and built permanent capacity through public–private partnerships and overseas investments. Europe displayed similar learning after Russia’s full-scale invasion of Ukraine, replacing most Russian energy imports in less than two years. These cases show how coercion creates markets and how state intervention speeds adaptation far beyond normal market conditions.

According to the International Energy Agency, China dominates refining for 19 of 20 strategic minerals, holding roughly 70 percent of global capacity. This imbalance was long recognized, but cheaper and reliable imports dampened efforts to build alternatives. China’s use of the rare earth choke point has forced a shift in the West, elevating upstream extraction, midstream refining, and downstream magnet production to strategic priorities. Canada is already conditionally approving facilities to refine samarium and gadolinium in Kingston, Ontario. New capacity is emerging across allied economies: Arafura’s Nolans Project in Australia integrates mining with on-site processing and separation, Vital Metals’ Nechalacho mine and Saskatoon facility combine extraction with early-stage refining, REEtec’s plant in Norway provides midstream separation, and Japan is expanding magnet-grade production through firms such as Shin-Etsu Chemical and Hitachi Metals. If these projects come online at planned capacity by the early 2030s, they could significantly blunt China’s current advantages. These emerging projects are being reinforced by parallel international coordination. China’s actions have accelerated efforts through the Quad, the U.S.–Australia Critical Minerals Partnership, and the G7 Minerals Security Partnership, now shifting from planning to operations by pooling stockpiles, sharing data, and aligning finance.

The second dimension, replaceability, concerns the structural difficulty of substituting a choke pointed input, and in this space the United States holds a stronger capacity to diversify suppliers and develop alternatives than China does. The West holds substantial rare earth resources: The U.S. Geological Survey identifies more than 200 U.S. mineral districts with significant rare earth enrichment and estimates U.S. reserves at about 1.8 million metric tons, while Australia and Brazil each report tens of millions of tons in reserves. Processing and refining technologies are also widely available. Solvent extraction, ion-exchange separation, and oxide conversion were developed in the West decades ago and remain commercially replicable outside China. Many of the barriers to scaling, from local opposition to insufficient public funding and weak private incentives, stem from the fact that Chinese imports have been cheap, reliable, and predictable. Under extreme pressure, however, investment can move rapidly, as shown by the synthetic rubber program of the 1940s, the energy build-out of the 1970s, and Operation Warp Speed in 2020.

The effectiveness of substitution also varies. Those with established extraction and midstream bases in allied economies such as lithium, nickel, and light rare earths like lanthanum and cerium can expand quickly once investment and permitting align. By contrast, heavy rare earths such as dysprosium and terbium, and niche inputs like scandium or tellurium, expand more slowly because of complex separation chemistry, small markets, and near total dependence on Chinese processing. There are also built-in buffers during the transition to new sources. The Defense Logistics Agency maintains stockpiles of rare earth oxides and alloys sufficient to sustain select defense manufacturing for several months, which can be released under Title III of the Defense Production Act. Automaker Hyundai and its affiliate Kia hold about one year of rare earth and magnet inventory for electric and hybrid production, while Tata Motors reports adequate stocks as it diversifies supply lines.

The third dimension is precision, wherein China’s choke point is inherently imprecise because its leverage comes from the sheer scale of its rare earth refining and separation capacity rather than from targeted control. If Beijing imposes a rare-earth ban, domestic supply tightens and prices rise. For instance, China’s neodymium-praseodymium oxide benchmark soared approximately 40 percent in August 2025 after a shipment disruption. This is because export controls tighten supply and trigger speculation and rationing inside China, as traders hold back material and processors delay sales in anticipation of shortages. This reduces domestic availability and drives up spot prices. Because China’s own magnet producers, motor manufacturers, and electronics firms rely on the same refined oxides, alloys, and magnet inputs that a ban would restrict, they must pay more for the very materials they depend on. In other words, a rare earth ban forces China’s midstream manufacturers to absorb much of the shock they intend to impose on foreign buyers.

The fourth dimension is feedback, as each use of a choke point weakens its impact, a pattern seen in the United States where Chinese pressure has driven stronger bipartisan support for alternative supply chains. Under both the Trump and Biden administrations, the Departments of Defense and Energy have funded and expanded projects such as MP Materials’ Mountain Pass separation facility in California, Lynas USA’s refinery in Texas, and Energy Fuels’ monazite operation in Utah. The Trump administration has even taken this further by acquiring equity stakes or issuing major loan and exploring agreements in key critical mineral and technology firms, including Lithium Americas, Trilogy Metals, Vulcan Elements, and USA Rare Earth. The Trump administration has also explored agreements with Saudi Arabia on critical resources, broadening its search for non-Western funders and financing. This alignment did not emerge organically. It was driven in part by China’s own history of abrupt export controls, licensing shifts, aggression toward  Taiwan, and maritime coercion in the South China Sea, all of which highlighted the risks of concentrated dependence.

The fifth dimension is sustainability, capturing the internal limits on maintaining a choke point over time, as shown by the environmental and social sustainability pressures now confronting China’s refining system. Decades of acid leaching and waste dumping in Inner Mongolia and southern provinces have contaminated water and soil, provoking protests and tighter regulation. The government’s Green Mining initiative now mandates costly remediation, tailings control, and centralized waste treatment, while its campaign against illegal mining has further reduced output. Rising energy, labor, and compliance costs are eroding China’s price advantage and narrowing the gap with Western producers, improving the competitiveness of cleaner operations abroad. Some analysts note that China is already extending its rare earth chain through investment in Myanmar’s heavy rare earth extraction and Vietnam’s magnet production, shifting environmental risks abroad and diffusing what was once a concentrated industrial base.

In sum, the five dimensions place China at a disadvantage and tilt the overall system toward sustained Western adaptation. Yet Beijing’s threats are not irrational. By threatening costs on the U.S. defense and economic industrial base, Beijing secured a reciprocal U.S. pause on the Bureau of Industry and Security affiliate rule. Domestic political incentives also shape its coercive calculus, since export controls are designed not only for foreign leverage but also for domestic audiences. Rare earth restrictions are both performative and transactional, useful for bargaining and signaling but ultimately corrosive to structural power once markets and adversaries adapt.

The Semiconductor Choke: Computer Power and Moving Goalposts

America’s semiconductor choke point exceeds China’s across all five dimensions. In the first dimension of durability, semiconductor controls now function as an industrial firewall that prevents Beijing from absorbing the learning loops of advanced chips, design software, and lithography tools. For decades, Beijing leveraged access to foreign markets to acquire know-how, reverse engineer designs, and build national champions. American telecommunication firms localized production in China during the 1990s and 2000s, transferring expertise that allowed state-backed companies like Huawei and ZTE to replicate their architectures and later displace them in domestic and foreign markets. A similar process unfolded in the electric vehicle sector when Chinese suppliers in Tesla’s ecosystem rapidly absorbed production methods. By 2024, automakers such as BYD and XPeng had matched, and in some areas surpassed, Tesla in output and price competitiveness. In other words, unlike earlier sectors where openness enabled China to internalize foreign expertise, the semiconductor hardware layer is being ring-fenced to block that process.

The second dimension is replaceability, evident in China’s ability to build data centers but not to equip them with frontier processors, forcing firms to rely on export-compliant chips bound by U.S. performance limits. Commerce Department data and independent analyses show that U.S. and allied firms control roughly 90 percent of global semiconductor manufacturing equipment and about 92 percent of overall supply chain value, underscoring Chinese dependency on American technology. Companies such as Alibaba Cloud, Baidu, and Tencent have postponed major training runs and must acquire thousands of additional processors simply to maintain previous performance levels. This is the reason why a gray market has emerged, with distributors in Singapore and Malaysia diverting restricted hardware toward China through shell companies and falsified documents. Even if limited quantities slip through, the volume is too small to offset the shortfall in high-end chips. According to a recent Center for a New American Security working paper, between 10,000 and several hundred thousand advanced graphic processing units were likely smuggled into China in 2024, with a median estimate of about 140,000 units, amounting to only 6 to 10 percent of China’s total AI computing capacity.

The third dimension is precision, reflected in U.S. semiconductor controls that target specific performance thresholds, architectures, and tools in ways that directly constrain China’s technological capabilities. Thresholds for interconnect speed, chip density, and training computing power determine exactly which processors are restricted and which remain export compliant. This selectivity forces Chinese firms to operate below the technological frontier while still permitting sales of downgraded processors that tie them to American hardware and software ecosystems. These export compliant chips are slower and less capable, yet they remain superior to domestic Chinese alternatives, ensuring continued dependence and preserving U.S. visibility through licensing and ecosystem control. It also discourages Beijing from abandoning U.S. architectures fully, delaying the emergence of an independent computing stack. This move was backed by the Trump administration’s decision to permit sales of NVIDIA’s H200 chips, a product two generations behind the Blackwell and Rubin series. The result is a two-tier system: Export controls restrict access to the technological frontier while permitting limited sales below performance thresholds, maintaining leverage through both scarcity and dependency.

The fourth dimension is feedback, in which U.S. chips keep advancing through successive generations, powered by global demand. These purchases channel revenue back to U.S. suppliers, who reinvest it in the next generation of chips, software tools, and architectures. Public supercomputing data shows that the United States already holds roughly five times the aggregate training computing power of Chinese systems, and the continued reliance on American chips reinforces this gap. According to Epoch AI, the computational performance of leading U.S. systems has doubled roughly every nine months, and U.S. firms also dominate the global supply of high-end  graphic processing units and the software tools that underpin AI design and training. In effect, China’s adaptation funds the very innovations that keep it dependent, creating a feedback loop that tightens the choke hold the more China attempts to work around it.

The fifth dimension is sustainability — reflected in how America’s semiconductor advantage is anchored in institutional and financial depth, from its big tech firms and venture pool to its private equity ecosystem — that allows the choke point to be maintained without letting up. Despite a lead in energy, China faces sustainability constraints of a different kind, and in late 2025 this became even more visible when Washington approved restricted H200 exports to selected Chinese customers with a fee attached. This action forced Beijing to weigh both the benefits of continued access and the political and security risks of deepening reliance on American high-performance chips. Beijing has signaled its intention to limit access to these, but these efforts remain constrained by the absence of viable substitutes and by broader economic pressures, including rising unemployment, economic stagnation, and pressure on household incomes, which make it costly for firms to invest in unproven domestic options or accept performance losses. In effect, China’s attempts to reduce reliance impose significant costs on its own firms and society, revealing the burdens that accompany any effort to catch up under constrained conditions.

Semiconductor controls endure because they form a self-reinforcing system in which containment blocks China’s full absorption of American technology, limited replaceability constrains its computing capacity, precision sets clear performance thresholds, feedback loops compound U.S. leverage with each generation, and temporal asymmetry results in further pressure on the Chinese economy. Still, sustaining this architecture of control depends on allied coordination and U.S. leadership on export regimes, incentives, and subsidies.

The Futures of the Burn and the Choke

This assessment forecasts the effectiveness of both sides’ export bans if fully implemented. Three outcomes are possible, two of which sustain Western advantage over China. Each reflects a reconfiguration of the five dimensions outlined above.

The first is the enduring choke, in which Western alignment on refining, fabrication, and innovation converts short term disruption into a lasting architecture of technological dominance. Beijing’s rare earth leverage contracts into a local industrial tool, while Washington’s semiconductor controls renew their force with each innovation cycle. Partial decoupling is the second possible outcome. Western firms retain command of design software, fabrication standards, and global compute capacity, while China remains reliant on mid-tier chips and legacy tools. The third, reversal shock, is a temporary setback in which allied execution falters or political cohesion erodes, allowing China limited recovery. Yet such gains would be cyclical rather than systemic. Across all scenarios, rare earth coercion burns out while semiconductor leverage compounds over time.

Alvin Camba, Ph.D., is lead scientist and director at Lyvi AI. He is also a senior research fellow at Associated Universities Inc. and a nonresident fellow with the Atlantic Council’s Indo-Pacific Security Initiative at the Scowcroft Center for Strategy and Security.

**Please note, as a matter of house style War on the Rocks will not use a different name for the U.S. Department of Defense until and unless the name is changed by statute by the U.S. Congress.

Image: Craig Fritz via Sandia Labs