Executive Summary

The United States and China have each built the capacity to inflict serious economic damage on each other, and neither has used it in full. Washington controls the semiconductor technology on which China’s ambitions depend. Beijing, through its near monopoly on rare earth processing, controls inputs that the West’s defense and advanced manufacturing cannot do without. Most analysts judge that neither side will be able to escape these dependencies for several years. 

This paper argues that the resulting mutual vulnerability, the Cold War logic of Mutually Assured Destruction (MAD) where “destruction” here should read as economic disruption, now constrains both sides and prevents further decoupling. That equilibrium does not ensure stability. The baseline is several years of “unstable stability”: a world less integrated than the one we lived in for decades, less decoupled than many fear, and punctuated by flare-ups that resolve through de-escalation rather than victory or settlement.

The paper maps this contest across two stacks: the real economy stack of computing, semiconductors and rare earths (economic MAD, or EMAD) and the financial stack of dollar dominance and digital finance (financial MAD, or FMAD). 

On the real side of the economy, an analysis of global earnings calls, run through large language models (LLMs), shows this mutual vulnerability already coloring the corporate discourse.

On the financial side, a form of Financial MAD (FMAD) has existed for some time. The U.S. dollar dominance and the power to cut entities out of the global U.S. dollar payment system are a first-strike capability. China’s Treasury holdings and the exposure of U.S. interests to its regulatory reach are the second.

Digital finance could tip this balance either way, and not along the same trend line. Portable U.S. dollar-denominated digital assets, the path the U.S. has chosen with the GENIUS Act, could deepen U.S. dollar dominance and erode China’s capital controls, tilting the balance toward Washington and inviting more fragmentation. A drift of transactions onto offshore rails like Tether could do the opposite, costing Washington the enforcement power behind sanctions and, counterintuitively, restraining it.

Both the U.S. and China, meanwhile, run internally inconsistent policies. The U.S. is closing its door to trade while keeping an open financial account, and China is doing the reverse. These are configurations that are ultimately unsustainable and could erode the very arsenals that sustain the equilibrium.

For corporate leaders, the variable that matters most is time. Two clocks are running at once. The first is structural and slow: neither side can escape its key dependency on the other for several years, which makes mutual vulnerability a reliable planning baseline. The second is unpredictable and can move fast: the policies that hold that mutual vulnerability in place are internally inconsistent, and each side’s own contradictions could end the calm sooner than the structural clock implies.

The practical implication is that the multi-year horizon is enough time to plan around but no promise of globalization-era stability: firms should use it to decide which dependencies to reduce first, and on which side of the divide, sequencing by where they sit on the fault line. With the end-date uncertain, the payoff comes from adapting early to the new MAD reality.

For investors, the open question is valuations. National security is now an overriding government priority, yet it is unclear whether markets have fully priced in a security risk premium across asset classes and geographies. Markets have tended to read tariffs, digital finance, and financing costs as separate events, while they are in fact interconnected.

For policymakers, especially in Europe and the middle powers that have mostly been bystanders to the rewriting of the global order, the contest has created space. Now the question is whether and how to use it. Middle powers need not pick one bloc and stay in it. Europe, above all, has the economic weight to matter and the leverage (e.g., single-market access, the power to write rules) to stop being a target for pressure from both sides. They can either act with urgency or spend the next decade as collateral damage.

Over the next several years, the winners will be those who treat the MAD equilibrium as a condition to plan and optimize around, managing the risks and seizing the opportunities, rather than a disruption to wait out.

Technology, Globalization, and the New U.S. Policy Regime

The U.S. and China have each acquired the capability to inflict material economic damage on each other, but neither has used it to its full extent yet. The U.S. controls access to the semiconductor technology on which China’s technological ambitions depend. China, through its near-monopoly on rare earth processing, controls inputs without which the West’s defense and various segments of advanced manufacturing cannot function. These dependencies will likely persist since most analysts assess that neither party can resolve them for several years despite ongoing efforts to derisk. This paper argues that the resulting mutual vulnerability, analogous to the Cold War logic of Mutually Assured Destruction (MAD), now constrains both sides and limits further economic decoupling. This is the equilibrium that corporate and policy leaders will need to navigate for the next several years, and this paper sets out to map. [1]

The U.S. is now pursuing two sets of broad economic and financial policies in parallel: “border” measures including immigration restrictions, tariffs, and restrictions on exports and foreign investment in strategic sectors; and a decentralized, private-sector oriented approach to digital finance, as exemplified by the recent passage of the GENIUS Act for stablecoin regulation. The new U.S. policy regime is the product of three factors that compounded each other: the eventual limits of the post-war global economic and financial bargain, the failure of the convergence bet after China joined the WTO, and a domestic political backlash against globalization that has become more vocal after the global financial crisis and COVID pandemic (Box 1).

---

---

This paper asks whether technology and trade, which have driven the policy shift, will continue to amplify fragmentation going forward or instead start to dampen it and whether blockchain-powered digital finance will reinforce or offset these dynamics in the real economy. The paper examines these questions across two stacks: the real-economy stack of computing, semiconductors and rare earths and the financial stack of dollar dominance, digital finance, and the public and private plumbing through which money moves across borders.

The answer to these questions depends on whether either side of the Sino-American economic conflict holds asymmetric coercive power arising from weaponized interdependence (Farrell and Newman, 2019). In the one-hegemon world described by Hirschman (1945), weaponized technology amplifies fragmentation forces and leads to decoupling. However, in an alternative two-hegemon world in which both sides hold credible retaliatory capabilities, as Box 2 documents, technology and trade can instead contain fragmentation by setting limits on how much decoupling can be imposed on the other — akin to nuclear deterrence during the Cold War (Schelling, 1960 and 1966).[2]

The paper finds that a two-hegemon equilibrium is the scenario that evidence from the recent corporate narrative and policy chronology supports. Drawing on survey evidence from global earnings call transcripts implemented through large language model (LLM) analysis, it shows that mutual vulnerability has started to color the corporate discourse in 2025. Furthermore, a review of the recent tariff and technology confrontations between China and the U.S. suggests that the EMAD equilibrium is indeed likely constraining further fragmentation and economic decoupling. The paper’s baseline is that this equilibrium will hold for the next several years, but it will be fragile — punctuated by frequent flare-ups — and could suddenly break, pointing to several years of unstable stability, not a return to the peace and stability of the globalization era.

The paper also examines how digital finance can interact with this new MAD equilibrium. A parallel, but less symmetric form of “balance of financial terror” (Summers, 2004), or “financial MAD” (FMAD; Ferguson, 2023), has been in place for some time. U.S. dollar global dominance and the ability to exclude entities from the international payment system through SWIFT sanction, or the U.S.-based correspondent bank network, are a first-strike capability, but China’s significant Treasury holdings and the exposure of U.S. economic interests to China’s regulatory reach are the second strike. Additionally, Beijing has been quietly building an increasingly effective countersanction arsenal since the beginning of the Trade War in 2018.[3]

The paper argues that digital finance could alter this equilibrium in either direction. In one direction, the development of portable and interoperable U.S. dollar-denominated digital asset markets could deepen dollar dominance within the West and undermine China’s efforts to internationalize the RMB. Such developments could also erode China’s system of capital controls, weakening its ability to absorb economic shocks, even if ring-fenced by the country’s Central Bank Digital Currency (CBDC) or digital RMB. This would tilt the balance in favor of the U.S. and likely result in additional fragmentation, as a weaker Beijing means a lower cost to Washington of escalating on the rungs where the U.S. holds leverage.

In the other direction, the balance could also shift in China’s favor. If a fragmented approach to digital finance regulation and geopolitical fractures within the West were to push transactions toward offshore unregulated rails, like Tether, Washington would risk losing the financial enforcement capacity that has long underpinned sanctions. Perhaps counterintuitively, this would further contain fragmentation by constraining Washington’s freedom to escalate. The implication is that the next several years of U.S.–China economic competition will not unfold along a single trend line. Instead, it will move with technology, regulation, and geopolitics — often at the same time and not always in the same direction.

Finally, the paper highlights that both sides are running internally inconsistent policy configurations in trade and finance: the U.S. closing international trade while keeping an open financial account, and China doing the reverse. These policy mixes are ultimately unsustainable and could weaken both countries’ own policy arsenals, adding uncertainty and instability to the global economic environment.

The Real-Economy Stack: How Each Side Built Its Arsenal and Deployed It

Today’s semiconductor industry is a highly specialized yet globally integrated value chain. As more capable chips became harder to manufacture and the costs of building advanced fabrication plants soared, the industry shifted away from the vertically integrated model of Intel toward a horizontal structure. Today, companies like Nvidia design chips but do not own factories, while geographically concentrated suppliers handle manufacturing, assembly, and testing.  

This specialization resulted in a very small number of firms at each node. The U.S. controls two critical upstream layers. First, U.S. firms Synopsys and Cadence hold a near-duopoly on the Electronic Design Automation (EDA) software; without them, no firm could design a modern chip. Second, American fabless firms, like Nvidia, Qualcomm, and AMD, dominate chip design and the intellectual property that protects it, producing the blueprints that the rest of the chain manufactures.[4]

Downstream, the Netherlands and Germany supply the equipment layer: ASML holds a near-monopoly on EUV lithography machines, relying on Germany’s Zeiss and Trumpf for the ultra-precise optics and lasers at their core. Japan dominates the adjacent materials and wafer-processing equipment layers, with Shin-Etsu Chemical, SUMCO, and Tokyo Electron as the leading firms. Taiwan’s TSMC and South Korea’s Samsung and SK Hynix then translate all of this (designs, tools, equipment, and materials) into actual silicon. TSMC dominates leading-edge logic fabrication, while Samsung and SK Hynix dominate memory.

China’s position in the semiconductor value chain is asymmetric. It is vulnerable where it matters most for its technological ambitions, yet indispensable in ways that constrain how far the U.S. can force it out of the value chain. China is dependent on the leading edge of chip design and production, likely still 5–10 years away from producing the EDA software or EUV lithography equipment needed for technological independence from the U.S., according to the latest publicly available assessments.[5] This dependence is the foundation of U.S. export control leverage (Miller, 2022).

However, China is critical in other parts of the value chain. To begin with, China is the world’s largest consumer of semiconductors, accounting for roughly one third to one half of global semiconductor demand, depending on the measure and year. For U.S. fabless firms, China is a major revenue source. This creates mutual dependence: China needs the chips, but the U.S. firms need the market access to generate revenues which in turn supports cash flow to invest.

Furthermore, China is also one of the world’s largest hubs for outsourced semiconductor assembly and testing, taking processed wafers and turning them into finished, packaged chips ready for devices. This is a less technologically demanding node than fabrication, but essential and difficult to relocate on a large scale in the near term.

Finally, China has significant capacity in older-generation chips, which represent the bulk of the global demand in cars, appliances, and industrial equipment, creating a vulnerability for the U.S. and its allies that was exposed by the Covid pandemic.

China’s rare earth position operates on a different layer of the computing stack and constitutes a more immediately critical second-strike capability in the tech war. China controls more than 60% of global rare earth mining and 90% of the processing capacity, dominating the supply of gallium and germanium. These materials are critical for the manufacturing of radio frequency chips, power electronics, and intensively used in the U.S. defense industry among many other sectors. This lever cannot directly affect the firms at the core of the U.S. technological advantage, but it is a more diffuse node in the global manufacturing network and very difficult to substitute at scale, even in the medium term. Available estimates suggest that meaningful processing independence is at least half a decade away, even under optimistic investment scenarios.[6] 

This extreme degree of specialization and concentration emerged from market logic during the globalization era rather than policy design. Yet, it has created chokepoints that the U.S. and China have now chosen to weaponize as instruments of geopolitical pressure, as Iran and the U.S. have weaponized the Strait of Hormuz (Box 3).

The Policy Arsenals

The U.S. and China have each built export-control regimes that exploit the chokepoints each country dominates (Appendix A). The U.S. arsenal targets advanced computing technology. Its principal instrument, the Foreign Direct Product Rule (FDPR), extends U.S. jurisdiction to any foreign-made good produced using American technology or equipment. This policy reach is effective because every modern chip depends, at some stage, on U.S.-origin design software or manufacturing tools. Deployed initially against individual Chinese firms in 2019, the controls broadened to industry-wide restrictions in 2022. They became multilateral when Japan and the Netherlands imposed parallel licensing requirements in 2023 and were further tightened in 2025.[7]

China’s counter-arsenal targets rare earths and critical minerals. Its leverage rests on a different premise: regardless of where rare earth ores are mined, processing is overwhelmingly concentrated in China, making border controls sufficient to restrict global supply. Beginning in 2023, with licensing requirements on gallium and germanium, Beijing has progressively expanded restrictions to graphite, rare earth extraction technology, and magnet-grade materials, following a tit-for-tat cadence that mirrors the U.S. escalation. The defense sector is the most acutely exposed. U.S. military systems depend on rare earth magnets with no short-term domestic substitute.[8]

Both arsenals can impose immediate costs on the opponent but also generate side-effects on their own firms, and both triggered a derisking response that is eroding the effectiveness of the chokepoint over the longer term. This pattern comes with direct implications for the stability of the EMAD equilibrium.

Economic MAD: A New Equilibrium

Hirschman’s (1945) framework has been the dominant paradigm to interpret the Sino-American trade and technology war. It was recently formalized in Clayton, Maggiori, and Schreger (2026a) as a large country (in this case the U.S.) that exploits economic relationships to create asymmetric dependence. This view presumes that American dominance of semiconductor design and equipment is a unilateral leverage with no credible Chinese counterpart. However, China’s near monopoly in rare earth processing is a mirror-image chokepoint on a different layer of the global manufacturing supply chain.

The weaponization of this critical input has resulted in a structural shift away from a single-hegemon world toward an equilibrium with bilateral deterrence grounded on mutual vulnerability. The paper terms this, by analogy with nuclear strategy, Economic Mutually Assured Destruction (EMAD), consistent with the two-hegemon model of Clayton, Maggiori, and Schreger (2026b). In this new EMAD setting, deterrence does not require symmetric capability; rather, like in the standoff in the Strait of Hormuz (Box 3), it requires that the weaker party can credibly impose high enough costs on the attacker (Schelling, 1960, 1966).[9]

These two paradigms generate distinct testable predictions. In Hirschman’s single-hegemon world, (a) the U.S. escalates unilaterally while China accommodates or makes asymmetric concessions, as in the case of the U.S. and Europe or the U.S. and Iran through February 2026; (b) China’s retaliations remain symbolic rather than operationally credible; and (c) the corporate narratives reference only U.S. policy actions, with no measurable impact from Chinese policy actions.

Under EMAD, an equilibrium akin to cold war MAD, (a) both sides escalate and demonstrate credible and comparable retaliatory capacity; (b) mutual stand-downs prevent a full decoupling (i.e., forcing China out of the tech value chain or major disruptions to the West manufacturing capabilities) but the arsenals’ architectures remain intact rather than being dismantled. This implies neither a return to the pre-war order of open borders nor a descent into Cold War blocs; and (c) corporate narratives register a growing bilateral signal: Chinese export controls appear alongside U.S. restrictions, concentrated in the semiconductor and critical minerals sectors, even if the Chinese signal remains smaller in magnitude. The evidence that follows tests these predictions.

Policy Chronology and Global Corporate Narrative Evidence

The first two predictions, retaliatory escalation and mutual stand-down, can be tested against the trade and technology war policy chronology. The detailed sequence of U.S. and Chinese policy actions from January 2025 through May 2026 is reported in Box 2.

---

---

On the one hand, in 2025, the policy narrative shows that China matched U.S. escalation with retaliatory measures of comparable scope, culminating in the October 9 export control package (a direct response to the September 29 issuance of the BIS Affiliates Rule). This demonstrated credible and effective retaliatory capacity, consistent with EMAD prediction (a), somewhat similar to what happened in the Iran War (Box 3). A month later, at the Busan bilateral meeting, both sides suspended their new measures but did not revoke them, leaving the arsenals’ architecture intact, consistent with prediction (b). On the other hand, these patterns are not consistent with the Hirschman hypothesis of unilateral escalation met by accommodation.

The third prediction, a shift in corporate narratives from a unilateral to a bilateral signal, requires a different kind of evidence, which Figure 1 provides. This shift, which unfolds over the course of 2025, is consistent with the EMAD prediction (c) above but inconsistent with a Hirschman baseline of one-directional geoeconomic pressure. The evidence in Figure 1 shows that U.S.-led restrictions dominate the global corporate narrative from 2019 to late 2024. Chinese export restrictions are visible starting in late 2024 and quickly become salient through 2025. By the end of 2025, the gap between the impact of the two policies on the corporate narrative falls to single-digit percentage points, as each side weaponizes the supply chain rung where it holds maximum leverage.

Three features of the two policies’ footprint in the global corporate narrative stand out. First, Chinese export controls surface at the end of the sample but scale fast. Policy 2 is effectively absent from the narrative through most of 2024, registering only after China’s December 2024 export ban to the United States on gallium, germanium, antimony, and superhard materials. It then climbs into the high teens, peaking just below 20% of total discourse after the April 2025 controls on seven medium-to-heavy rare earths and before the October 2025 MOFCOM package is announced.

Second, by the end of the sample, the gap between the two narratives closes, and not merely because the denominator grows. U.S. restrictions lead by roughly 40 percentage points at the end of 2024. A year later, the two series sit within single digits of each other: the first sustained narrowing since the chip war began. The convergence reflects China’s policies gaining share while U.S. policies lose it, a shift that appears in 2025’s better-populated quarters — not an artifact of a thin sample. This is the signature of a bilateral dynamic, in which each side weaponizes the rung of the supply chain where it holds most leverage. The shift tracks the policy chronology in Box 2 on both sides and survives the robustness checks reported in the Appendix.

Third, the EMAD total narrative itself (the denominator of the two shares — the grey shaded area) is informative. This total includes side effects reported by firms in the imposing countries, e.g., U.S. firms such as NVIDIA hurt by U.S. export restrictions on China. These mentions are included in the quarter totals but not in the two-policy series because of the sign restrictions that sharpen the definition of the policy footprint. It is notable that the evidence in Figure 1 supports the EMAD hypothesis even without considering the costs to the U.S. and its allies of implementing their export restriction policies. The Appendix reports the results including these side effects and shows that the results are robust in other respects of the analysis.
 

Technology and Trade: From Amplifiers to Constraint on Fragmentation

With this transition from a world with a single coercive hegemon to one in which two hegemons are mutually vulnerable, weaponized technology shifts from amplifying fragmentation to limiting it. This is not because either side voluntarily chooses restraint, but because each discovers that the other can inflict comparable pain. The standoff in the Strait of Hormuz between the U.S. and a far weaker Iran offers a parallel. As long as each side can impose economic damage on the other, both have an incentive to negotiate rather than risk extremely adverse outcomes (Box 3).

Of course, these incentives remain in place only as long as each country can maintain its economic leverage. The EMAD equilibrium could break down if the U.S. achieves rare earth independence or if China can domestically produce chips and other high-tech items comparable to those produced in the West. Both are taking action along these lines to extend their own advantage and diminish the effectiveness of the economic weapons in the other’s arsenal. Nevertheless, as noted earlier, such actions will likely require concerted effort over several years before either country can meaningfully reduce its critical dependencies on the other.

EMAD can prevent Cold War-era decoupling, but it also incentivizes low-intensity competition over standards and regulation, subsidies, talent, and third-country pressure. The conflict migrates to different areas; it does not end. We can expect both sides, over time, to seek advantage along specific dimensions including finance, the critical lever to which the paper turns next.

---

---

Digital Finance and Geopolitical Fragmentation

Will digital finance dampen or amplify international economic and financial fragmentation driven by geopolitics? How do borderless digital finance and economics, which is by definition “physically bordered”, interact? What does that interaction mean for the stability of the EMAD equilibrium described above? Finally, what is the role of government and the private sector in determining this balance?

Blockchain-based digital finance (for the purposes of this paper, both private payment instruments such as stablecoins, public central bank digital currencies (CBDCs), and the broader tokenization of financial assets) is inherently borderless, yet it is coming of age at a time when the world is drawing new hard borders. Digital finance was born out of distrust for centralized financial systems and government involvement in them, promising portable services with 24/7 settlement, decentralization, and minimal or at least transparent government intervention. That being said, in a world where globalization is effectively ending, it needs to reckon with the reality of geopolitical fragmentation and less coordinated and potentially incompatible national regulations.

Depending on the assumptions made, digital finance can dampen, amplify, or leave fragmentation unchanged by acting on the new EMAD equilibrium through the three channels summarized in Table 1 below. Digital payments like stablecoins, or tokenized assets more broadly, cannot on their own overcome international economic fragmentation driven by geopolitical forces. At the most fundamental level, international trade integration drives financial integration (e.g., Obstfeld and Rogoff, 2000).

However, research has also shown how trade invoicing and asset denomination can be jointly determined (e.g., Gopinath and Stein, 2021). Also, technology adoption in the financial system can have feedback effects into the real economy, including long-term growth (Jiang, Rebucci and Zhang, 2025). In principle, therefore, it is possible to think of mechanisms such as the three discussed below through which digital finance can push fragmentation in opposite directions or simply bring about more economic and financial uncertainty.

The evidence previously discussed on the emergence of a new EMAD equilibrium is retrospective and quantifiable. The conclusions here about digital finance’s role in shaping international economic fragmentation, the interaction between the two stacks, and the role of government and the private sector are prospective and more speculative, resting on scenario analysis and suggestive evidence rather than tested hypotheses. They should be taken as illustrative of possible outcomes rather than as forecasts or established findings.

The Baseline Scenario: Financial MAD

The status quo in the international financial system can be interpreted as a form of Financial MAD (FMAD), which has existed for some time.[11] Undoubtedly, the U.S. possesses “first-strike” capability. Its arsenal includes fiat dollar hegemony, SWIFT control, and correspondent banking gatekeeping being its arsenal, and it is increasingly deployed through sanctions, most recently on Russia and Iran. The arsenal now even helps to enforce technology restrictions.

China has long been recognized as having “second-strike” capability through its U.S. Treasury holdings, albeit a very costly one to deploy. China also hosts huge U.S. sunk investments such as Apple’s supply chain, Tesla’s Shanghai gigafactory, and financial institutions including banks, securities firms, and asset managers. Selective regulatory harassment, licensing delays, or asset freezes on U.S. firms operating in China is a more credible retaliatory instrument than a large-scale sale of U.S. Treasuries. This is exactly China’s tactic, already used repeatedly and increasingly so. China is also continuing to internationalize the Renminbi (RMB). It is building a parallel cross-border RMB settlement architecture with CIPS as the clearing rail and the e-CNY, tested through mBridge, as the digital-currency leg. The internationalization effort has been aided by an aggressive official finance program and a network of swap lines.

Table 1. Three Channels Through Which Digital Finance Could Affect Fragmentation

Note: Each channel acts on the U.S.’s hand relative to China’s, and fragmentation follows.

Digital Finance Regulatory Fragmentation and What Private Markets Can (and Can’t) Do About It

Geopolitical fragmentation is hardening into regulatory fragmentation, with asymmetric and difficult to anticipate implications for economic fragmentation. The U.S., China, and Europe are building three different architectures without any formal regulatory or technological coordination (Ahmed, Clouse, Natalucci, Rebucci, and Sun, 2025). The U.S. government backs privately-issued stablecoins under the GENIUS Act and has shelved retail central bank digital currency (CBDC). China has banned crypto and stablecoins and is building a state-controlled system around the e-CNY. Europe runs a dual system: regulated stablecoins under the Markets in Crypto-Assets Regulation (MiCA), alongside a digital euro. These are not three “private versus public” choices. They are three different public-private architectures, each shaped by a different view of who should run the rails.

Multilateral attempts to build a workable architecture for cross-border tokenized payments exist, but they have themselves split along the same fault lines. The Bank for International Settlements (BIS) Innovation Hub’s Project Agorá, launched in 2024, brought together seven central banks: the Federal Reserve Bank of New York, the Bank of England, the Bank of Japan, the Eurosystem (through the Banque de France), the Bank of Korea, the Bank of Mexico, and the Swiss National Bank, in addition to more than 40 private financial institutions, to test a multi-currency unified ledger for wholesale cross-border payments using tokenized commercial bank deposits and central bank reserves. The prototype phase concluded in May 2026. The project is now advancing to real-value testing, with the Bank of Canada joining as an eighth member.

On the other side, Project mBridge (which was originally a BIS-affiliated experiment in CBDC) continued as a China-led initiative after the BIS withdrew in late 2024, with the People’s Bank of China, the Hong Kong Monetary Authority, the Central Bank of the UAE, the Bank of Thailand, and the Saudi Central Bank as core members. The result is two parallel multilateral tracks: one Western-aligned and dollar-anchored, one China-aligned and increasingly oriented toward sanctions-resilient settlement. Both are technically viable, but they sit on opposite sides of the regulatory, governance, and currency-anchor fault lines that define the bordered financial fragmentation this paper maps. Reconciling them would require resolving the very contests over sanctions reach, payment-system control, and which currencies anchor settlement that the EMAD equilibrium has crystalized. Multilateralism in cross-border digital finance is therefore not absent; it is simply tracking, rather than mitigating, the geopolitical divide.

The question, then, is whether private rails can stitch back together what diverging public policy is pulling apart. The answer to this question has two parts. Within the Western bloc, there is precedent of private-sector led initiatives that help bridge the gap. For example, after Sarbanes-Oxley, foreign firms adopted U.S. governance rules voluntarily to keep access to American capital by importing U.S. standards without a treaty. Common token standards and International Swaps and Derivatives Association-style legal documentation routinely paper over regulatory gaps. The same dollar gravity could make stablecoin and tokenization standards converge on U.S. rails over time (Ahmed, Clouse, Natalucci, and Rebucci, 2026).[12] However, most of these precedents date from the 2000s, when broader policy cooperation was the default. That backdrop is gone. Today, within-bloc convergence depends on private firms moving ahead of governments. That is possible but not at all guaranteed.

Across the U.S.-China fault line, the private sector is unlikely to play a meaningful role. Private actors cannot bridge gaps that China or the U.S. make illegal to cross. Beijing has banned domestic crypto activity and will not accept U.S.-issued stablecoins on its rails, while the U.S. has shelved CBDC. Commercial coordination cannot override that. The regulatory perimeters are becoming the borders between the U.S. and China for digital finance.

The implications for fragmentation are straightforward. Within the Western bloc, private rails could help deepen dollar dominance, strengthen Washington’s hand, and push fragmentation up, though this outcome is uncertain because it depends on private firms moving ahead of governments. Across the U.S.-China fault line, regulatory perimeters are becoming hard borders that commercial actors cannot bridge. The overall direction of fragmentation would then depend on the two channels examined next: where sanctioned dollar activity may migrate and whether China’s capital controls will hold.

How Geopolitics and Geoeconomics May Fracture the Dollar Ecosystem

One of the risks of financial fragmentation along geopolitical fault lines is to push digital payments toward offshore unregulated instruments to escape the sanction risk of sitting inside the U.S. dollar rail. Unregulated dollar stablecoins such as Tether, while carrying greater counterparty risk, offer some insulation from U.S. jurisdictional reach, a feature whose value rises in a fragmented world. As fragmentation events unfold, users seem to migrate from regulated onshore U.S. dollar instruments to unregulated offshore ones, without abandoning dollar denomination, arguably seeking protection from the reach of U.S. sanctions and asset freezes.[13] The empirical implication is that geopolitical events and fragmentation more generally may not push the financial system off the U.S. dollar rail. Instead, it could fracture the dollar ecosystem into regulated and unregulated tiers.

This hypothesis can be tested by examining how the relative market capitalizations of offshore, unregulated U.S. dollar stablecoins, such as Tether and onshore, and de facto regulated U.S. dollar stablecoins such as Circle have changed around fragmentation events. Figure 3 seems to support the idea that following such events, stablecoin demand shifts offshore, affecting the instrument composition of U.S. dollar global demand without necessarily undermining dollar dominance itself.

If transactions by targeted entities migrate to unregulated offshore rails that sit outside the U.S. regulatory perimeter, the U.S. position in both financial and economic MAD could weaken. The hit to U.S. financial leverage is direct. There is also an indirect effect stemming from the loss of enforcement capacity of technology export controls. The tools that enforce sanctions and export controls on the traditional dollar rail do not automatically reach offshore stablecoins, like Tether or other crypto assets.

For example, Entity List screening, applied at SWIFT and correspondent-bank chokepoints, cannot intercept transactions that don’t flow through those chokepoints. Anti-Money Laundering and Know Your Customer (AML/KYC) checks, enforced by the U.S. Treasury through banks and exchanges, reach Tether only when users convert into or out of regulated money, and only when the wallet owner can be identified. The U.S. Treasury could ask stablecoin issuers to freeze wallets as it recently did in the case of Iran (Box 3). Then, offshore issuers have a choice whether to comply or not, but sanctioned actors hold funds in wallets that cannot be easily traced back to them. The result is that the U.S. dollar can remain the unit of account, but the U.S. control of the plumbing may weaken. The global reserve currency can survive while U.S. financial leverage diminishes.

This shift may dampen fragmentation by weakening Washington’s reach over both sanctions and technology export controls. It would, however, inject uncertainty into the global financial system by strengthening China’s hand even without any action by Chinese authorities, as the Iran case shows (Box 3).

Digital Finance’s Challenge to Beijing’s Wall of Capital Controls

China’s capacity to absorb economic pain from a trade or technology war depends in part on its ability to control cross-border financial flows. Capital controls can keep a trade shock as an economic and political problem rather than allowing it to become a financial crisis. They are the buffer that helps Beijing accept short-term economic damage without losing macroeconomic control. If digital finance erodes that buffer, China’s pain threshold would likely drop and its position in the EMAD would weaken, potentially leading to more U.S.-led fragmentation.

The threat is slow, not imminent, but it is real. China has suppressed private digital finance activity on the mainland while permitting it in Hong Kong. East Asia’s share of global crypto transaction value (driven primarily by China) fell from around 30% in 2019 to under 10% by 2022 (Chainalysis, 2023). That being said, crypto has been the easy case, since these assets are speculative with no stable unit-of-account function. Banning them is politically cheap because no one loses access to something really needed.

Dollar-denominated stablecoins and tokenized dollar assets are  different propositions. They offer ordinary savers a portable, hard-currency store of value and potentially a genuine portfolio diversification tool. These are precisely the functions that Chinese capital controls interfere with. If tokenized U.S. dollar assets achieve broad retail adoption in the West, the latent demand inside China will not come from speculators. It will likely come more from households seeking to protect savings against inflation, currency depreciation, and domestic financial risks such as the bursting of stock and housing market bubbles —  a far larger and politically more sensitive user base to police.

How that pressure may play out is difficult to anticipate. Beijing has strong tools, e.g., the Great Firewall, supervised domestic payment systems, banking-side identification, and has shown willingness to crack down on middle-class wealth management strategies when it judges the costs are worth paying. The likely response is increased and tighter enforcement before considering any opening up to international digital finance. However, over the next decade, the cumulative pressure may steepen the trade-off to a point at which the costs of policing a far broader and more complex web of permissions and prohibitions become larger than the costs of at least partially opening the capital account to private digital finance.

The paradox is that if this point were to be reached, it could lead to more fragmentation, not less. More permeable Chinese capital controls mean a wider and deeper reach of the U.S. dollar system and a stronger U.S. hand in the new EMAD equilibrium. However, a China with a lower pain threshold is also more likely to escalate sooner using the tools where it holds leverage such as rare earths and other critical inputs, to compensate for a weakening financial buffer, which means a more volatile EMAD equilibrium.

Will Digital Finance Amplify or Dampen Fragmentation?

The assessment rests on the assumption that fragmentation is a U.S.-led process and China’s capacity to resist geoeconomic pressure (via the EMAD logic developed in Section II) may limit how far it goes. Factors that strengthen the U.S. hand may make Washington freer to escalate. Factors that weaken it may force restraint.

One possibility is that none of the three channels has any material effect. That is, private rails fail to bridge Western regulatory divergence, sanctioned activity stays inside the U.S. perimeter, and China’s capital controls hold. In that case, digital finance does not move the EMAD status quo. The more interesting case, and the one this section addresses, is what happens if one or more of the channels gain traction.

The three channels do not all push in the same direction. Two channels may work in tandem to push fragmentation up by strengthening Washington’s hand: private rails help to deepen dollar dominance within the West, while pressure on China’s capital controls weakens Beijing’s buffer. A third channel, migration of sanctioned activity to offshore instruments like Tether, would run the other way, weakening Washington’s enforcement reach and, counterintuitively, dampening fragmentation.

Both outcomes leave fragmentation contained by EMAD but could move the equilibrium somewhat in different directions. The first outcome, on net, pushes the equilibrium toward more fragmentation, with Washington’s hand strengthened by private rails inside the Western bloc and by pressure on China’s capital controls. The second outcome pushes the opposite way, with U.S. enforcement reach over sanctions and export controls weakened, dampening fragmentation but raising uncertainty because we have never lived in a world where Washington does not have the ultimate upper hand. The two outcomes are clearly not symmetric.

Can the Two Stacks Remain Segmented?

Can the U.S. and China sustain their current trade and finance configurations? Both basic national accounting and economic theory suggest that international trade and finance are jointly determined. Yet the U.S. and China are each attempting to separate them. Neither configuration is likely to be sustainable, and the resulting internal contradictions can feed back into the MAD equilibrium by raising U.S. financing costs on one side, and by capping RMB internationalization on the other.

The U.S. is attempting to close its trade border through tariffs to reduce the deficit on the current account while keeping the capital account open, and is pursuing a decentralized, private-sector approach to digital finance.[14] Tariffs, nevertheless, can raise the cost of financing those imbalances by reducing what foreigners can buy with the dollar payoffs from U.S. assets, and foreign demand for those assets would fall (e.g., Obstfeld and Rogoff, 2000). If sustained, the U.S. approach could therefore raise the cost of financing both government deficits and the ongoing private AI capital expenditure cycle that current equity valuations assume. Financial repression, including through captive stablecoin demand for Treasuries, can contain these costs but only temporarily (Ahmed, Josephson, and Rebucci, 2025).

Trade restrictions and financing costs are connected, but the connection runs counter to the stated U.S. policy goal of reducing U.S. current account imbalances. Tariffs can raise the cost of financing the trade deficit without narrowing it, because the imbalance is driven by fundamental forces (e.g., savings, investment, productivity, and global demand for dollar assets) that tariffs have proven unable to change.

China faces the opposite configuration (also discussed in Section 3.4), aiming to remain open to trade while keeping the capital account closed. An internationalized RMB must eventually be a convertible RMB to rival the U.S. dollar, as trade invoicing and financial intermediation are mutually reinforcing (Gopinath and Stein, 2021). Foreign exporters accept RMB receipts more easily if they can be converted into liquid assets without friction, and capital controls prevent that. Currently, CIPS volume and swap-line networks expand the technical reach of the RMB without delivering free convertibility into other reserve currencies. This bounded internationalization caps the financial leg of China’s deterrent. Beijing’s economic leverage continues to operate mostly on the real-economy side, with the financial side remaining asymmetric and the dollar system still dominant, even as its enforcement reach erodes. Beijing can technically expand RMB reach without convertibility. Nevertheless, it cannot deliver the market depth that would let banks fund trade in RMB at dollar-scale, or the network effects that would make the RMB a truly global currency. Without those, the RMB cannot rival the dollar.

Both sides are running policies that potentially undermine their own arsenals. Tariffs can raise U.S. financing costs, while capital controls cap RMB internationalization. That makes the planning horizon around the new MAD equilibrium more uncertain. Its durability is vulnerable to self-inflicted damage in addition to the technological breakthroughs on either side discussed in Section II.

Conclusions: A Multiyear Window and What to Do With It

The U.S. and China have each built the capability to inflict material economic damage on each other and have shown they are willing to use it short of full deployment. The November 2025 suspension of the chip war was not a settlement, any more than the Liberation Day tariff retreat. Both were mutual recognitions that a further escalation would hurt both sides. The technological constraints each country faces point to several years of unstable stability: a world less integrated than the one we lived in for decades and less decoupled than the one many fear, punctuated by flare-ups that resolve through de-escalation rather than definitive victory or settlement.

Decisionmakers who plan for either corner, restored U.S. hegemony or clean Cold War-era-style decoupling, will likely be disappointed. The unipolar post-WWII order is over. China cannot replace it, and America cannot restore it. What this paper portrays is a persistent contest between two powers strong enough to deny each other victory but too exposed to seek it at full force.

Over the next several years, EMAD is the framework the private sector can use to assess risks and opportunities. The imperative is to build resilience across supply chains, mindful of the geopolitical fault lines and which side of them a firm operates from. National security has become an overriding priority for governments, and financial markets have barely begun, if at all, to price a security risk premium across asset classes and geographies.

Beyond that horizon, the restraining effects of today’s mutual vulnerabilities cannot be assumed in a world where technology evolves at fast pace. Rare earth processing can be built in the West as a COVID vaccine was brought to market against prevailing expectations; China could close the semiconductor gap; and digital finance could tilt the MAD balance. The window could also close from within through a resolution of the policy contradictions each country faces, rather than any technological breakthrough on either side.

What to watch? The window stays open only while the underlying dependencies hold. Six indicators can track whether they are eroding:

• China’s chip stack — progress on domestic EDA software and advanced lithography. Genuine independent capability here would erode the U.S. side of EMAD.
• Western rare earth processing — technological advancements and capacity coming online at commercial scale would erode China’s side.
• Stablecoin market cap and where it settles — growth on onshore dollar rails deepens dollar dominance and pushes fragmentation up. Migration to offshore instruments like Tether erodes U.S. enforcement and pushes it down but makes the financial system more unstable.
• Digital-finance regulation — convergence across the U.S., EU, and China versus continued divergence. Divergence is what gives offshore rails their opening.
• RMB convertibility — any real capital-account opening would lift the cap on China’s financial deterrent. Continued controls keep it bounded.
• Energy security on both sides — for the U.S., LNG export capacity is the buffer to monitor, as the Iran episode has shown. For China, watch how much energy it still has to import by sea, and how fast pipelines, stockpiles, and home-grown power cut that dependence.

For corporate leaders, the critical variable is time. EMAD defines the environment to plan around, but the multi-year horizon is a baseline, not a promise of globalization-era calm and stability. The task is how to manage under that baseline: diversification calibrated to the fault lines and to the side a firm sits on, prioritizing which dependencies to reduce first and on which side. The window is not fixed, as each side’s own policy contradictions could end the calm sooner than the baseline horizon implies. So resilience built on the assumption of a fixed multi-year horizon of globalization-era stability is itself a losing bet.

For investors, the open question is valuations. National security is now an overriding government priority, yet it is not clear that markets have written a security risk premium into asset prices. Two questions are unavoidable, posed here rather than answered.

• Can the productivity payoff expected from AI arrive before the financing and energy costs of the capex come due? The fragmentation bill (capex, dual-sourcing, energy) is being paid now, including in terms of inflationary pressures, while the AI payoff will likely materialize over a number of years if the history of technology-driven financial boom is of any guide.
• Is the second “China shock” priced? Markets are pricing the disinflation from Chinese overcapacity in EVs, batteries, solar, and mature-node chips. Depending on whether an AI oligopoly will prevail or competitive pressure (including from Chinese providers of AI services) will compress markups will have huge implications for earnings, margins, and valuation multiples.

For policymakers, especially in Europe and the middle powers that have largely been bystanders and collateral damage to the ongoing rewriting of the rules of global order, the contest has opened space. The question is whether and how to use it. The Iran episode has shown that going at it alone carries direct and indirect costs for the U.S., and whether multilateralism returns, and in what form, is an open question. The more durable answer, as Carney has argued, is that middle powers, like individual firms, need not pick one bloc and stay in it.

In that framework, bloc perimeters become issue-dependent: chips with one side, payments with the other, energy with both. The same trade-diversion logic now extends to financial assets and settlement rails. Europe, above all, has the economic scale to matter, so it does not need to remain a bystander. It does not have chokepoints but has leverage via single-market access, regulatory rulemaking, and, in lithography, a position today absorbed into U.S. enforcement rather than wielded. They can choose to spend the next decade closing that gap or remaining a target for coercion from both sides. Europe needs to act; urgency is the key word.

The time-window is open but neither permanently nor in self-sustaining manner. The competitive advantages will go to those who treat unstable stability as a condition to plan around, including managing risks and benefiting from opportunities, rather than as a state waiting for settlement.

Appendices

A. The Policy Arsenals and Their Deployment in the China-U.S. Trade War

This appendix provides details on the policies implemented by the U.S. and China and their recent deployment.^[15]

A.1 The U.S. Arsenal: Export Restrictions on Chip Technology

The pillar of the current U.S. semiconductor export control policy is the Foreign Direct Product Rule (FDPR). The FDPR is a provision of the Export Administration Regulations (EAR) in the U.S. Code of Federal Regulations. The EAR applies to foreign-made goods that contain more than a de minimis share of U.S.-origin content, and the FDPR extends that jurisdiction to certain foreign goods produced using U.S. technology or equipment, even if the final product contains no U.S.-origin content. The FDPR’s effectiveness rests on the fact that the production of any modern chip depends at some point on U.S.-origin design software or manufacturing equipment.

For example, when TSMC severed its relationship with Huawei in 2020, the decision rested on TSMC’s dependence on U.S. EDA design and its unwillingness to jeopardize access to its U.S. suppliers. Additionally, in 2024, the financial system was formally enlisted as a second layer of monitoring and enforcement of technology export restrictions, when the Bureau of Industry and Security (BIS) issued guidance to banks warning that processing transactions linked to EAR violations carries its own liability.

The deployment of U.S. technology export restrictions has escalated gradually. The initial phase, running from 2019 through 2021, was firm-specific. The BIS Entity List targeted Huawei, SMIC, and their affiliates, restricting their access to U.S.-origin technology and components on national security grounds. In October 2022, the Biden administration’s BIS rules imposed industry-wide controls on China to slow its progress in advanced chip design and fabrication, where China’s lag behind the frontier is currently estimated at roughly 5–10 years, depending on the layer.

By 2023, the unilateral U.S. export restriction regime became multilateral. Japan and the Netherlands implemented their own licensing requirements on advanced semiconductor manufacturing, extending the reach of U.S. controls to ASML and Tokyo Electron. The policy was further refined in 2024 and 2025, closing loopholes while granting selective exceptions—to cushion the impact on U.S. firms like Nvidia and, increasingly, as bargaining chips in the broader trade war negotiations with Beijing.

Difference‑in‑Difference estimates of the U.S. technology export restrictions on China’s imports in Anderson and Jin (2026) indicate that successive waves of export controls from 2020 to 2024 produced heterogeneous effects across restriction waves, countries, and products. The most impactful wave was the 2022 BIS measures, when advanced logic chips, high‑end memory, and leading‑edge manufacturing equipment exports to China significantly declined, especially from the U.S. and South Korea. In contrast, during that period, Chinese imports from Japan, Germany, and the Netherlands surged as Chinese firms front‑loaded purchases of lithography equipment ahead of new licensing rules, similar to the frontloading response to tariffs seen in the U.S. Consistent with these patterns, firm‑level evidence in Albertson, Howard, and Sarzosa (2026) documents substantial financial strain among affected Chinese firms.

Nonetheless, the economic costs of these policies to U.S. firms in the value chain are also sizable. Crosignani, Han, Macchiavelli, and Silva (2026) finds that export controls to China reduce revenue, profitability, and employment among exposed suppliers. The policy has also had longer-term strategic side effects. U.S. export controls may have accelerated Chinese domestic semiconductor investment, triggered by a wave of state-directed R&D spending, according to evidence based on earlier implementation of these types of restrictions (Liu, Makarin, and Wen, 2025). The U.S. technology restrictions have also incentivized the weaponization of rare earths. By making China’s technological dependence a point of leverage, the U.S. also made it a visible point of vulnerability, providing Beijing with the motive, the justification, and the blueprint to build a mirror-image retaliatory framework.

A.2 China’s Counter-Arsenal: Export Controls on Rare Earths and Minerals

China’s leverage over global rare earth supply did not begin weaponized. For much of the globalization era, China managed its rare earth resources primarily to protect the domestic industry. In 2012, the U.S., the European Union, and Japan filed a WTO dispute, and the WTO panel found the system discriminatory in 2014, forcing China to dismantle it. China removed the quotas and duties in 2015 and shifted toward domestic production controls and licensing.  

The current framework rests on the Export Control Law (ECL) enacted in 2020, which introduced an extraterritorial dimension similar in ambition to the U.S. EAR and the FDPR. Just as the FDPR’s leverage rests on the universal requirement of U.S. EDA software and chip design, China’s leverage rests on its near-monopoly in refining and separation — and within that, on heavy rare earth separation and permanent-magnet production, the segments where capacity outside China remains thinnest and which the IEA identifies as the main bottleneck to diversification (IEA, 2026).

China’s rare earths export restrictions are enforced differently from U.S. technology controls. The non-substitutable stage of production — processing — takes place on Chinese soil, regardless of where the minerals are mined. Australian, African, and American miners ship their raw materials to Chinese refineries because no comparable processing capacity exists elsewhere. Border control is therefore sufficient since any shipment of refined rare earth materials leaving China now requires a Ministry of Commerce (MOFCOM) export license. Like in the U.S. case, firms that circumvent the system risk losing future licensing.

In 2023, China first used the ECL as a basis for introducing licensing requirements for gallium and germanium — two materials used in compound semiconductors for radar, satellite, and 5G components. Controls on high-purity graphite, essential for electric vehicle battery anodes, and restrictions on technology for rare earth extraction, separation, and smelting followed later that year. In April 2025, China expanded rare earth export controls in response to U.S. tariffs, requiring licenses for seven medium/heavy rare earth elements. The escalation demonstrated both the vulnerability of global markets to the Chinese refining chokepoint and China’s credible enforcement capacity.

The pandemic-era semiconductor shortage of 2021–2022, which affected automotive, defense, and consumer electronics production when geographically concentrated chip fabrication was disrupted, provides indirect evidence of how supply chain dependencies translate into economy-wide vulnerability (Shivakumar and Wessner, 2022).

China’s post-2023 rare earth restrictions have had similar effects across the same sectors through deliberate policy rather than an exogenous shock. Gallium, germanium, and magnet-related controls have created supply delays, import declines, and production bottlenecks across European and American manufacturers, with scenario analyses and network modelling suggesting that a severe or prolonged interruption could propagate output losses well beyond directly exposed industries (ECB, 2025). Qualitative assessments of the April 2025 restrictions reach similar conclusions about the breadth of sectoral exposure and the risk of cascading production disruptions (CSIS, 2025a). The defense exposure is the most acute. U.S. military systems depend on rare earth magnets for precision-guided munitions, radar, and submarine propulsion and navigation, with no short-term domestic fix available (CSIS, 2025b; CFR, 2026).

As in the case of U.S. policy, historical evidence from China’s earlier 2010 rare earth export restrictions suggests that supply shocks of this kind trigger compensatory innovation in targeted economies. Alfaro, Fadinger, Schymik, and Virananda (2025) find that the 2010 episode produced a global surge in rare earth-related patenting and expanded downstream production outside China as firms developed input-saving technologies in response to the supply shock. Saving innovation alone cannot close the gap because the binding constraint is not knowledge but production capacity. Howard and Underwood (2024), using confidential U.S. Census firm-level data, estimate that sunk costs of switching critical-mineral suppliers run into billions per firm, and a CFR February 2026 assessment concludes that meaningful independence requires sustained investment at a scale not yet mobilized.

As with U.S. chip technology, estimates vary considerably. Here too the planning horizon runs to several years (5–10), the longest precisely for the segments China controls: heavy separation and magnets. For example, IEA (2026) estimates that even counting announced projects, capacity outside China is projected to meet only about half of mining demand, a quarter of refining demand, and under a fifth of magnet demand by 2035.

Appendix B. The Global Corporate Narrative on Export Controls

This appendix provides details on the LLM analysis of global earnings call reports in Figure 1 of the paper.[16]

B.1 Data and sample

The analysis draws on newly released data from the Global Capital Allocation Project (GCAP) Geoeconomics Monitor on the impact of geoeconomic pressure policies on individual corporations. This data comes from applying large-language-model classification to a broad sample of U.S. and internationally listed companies’ earnings call transcripts. For each transcript flagged as discussing export controls, a second-stage prompt identifies the country imposing the restriction, the country receiving it, the sector or product through which pressure is exercised, and whether the reported effect is positive or negative.[17]

The result is a large firm-quarter panel in which each flagged earnings call report is characterized by a small set of variables. The classified sample that the paper uses includes 1,970 firm-quarter observations from 2019Q1 through 2025Q4. Observations from earlier periods exist in the underlying GCAP data but are excluded because pre-2019 EMAD-relevant counts are too sparse for meaningful aggregation. The starting date is during the first Trump administration; the ending date is the last reporting quarter fully collected. The quarter date in the panel refers to the calendar quarter of the earnings call. Observations are labelled by this calendar quarter throughout.

B.2 Aggregation and subsamples

Countries imposing or receiving measures are grouped into three categories: the U.S. and its allies (USALL), China (CHN), and the rest of the world (ROW).[18] A measure imposed by a single group is assigned to that group; a measure imposed jointly by the USALL and CHN is assigned to both (a rare case, eight observations in total). Any other multi-group combination, or a measure with no identifiable imposing country, is coded as “Unidentified.” A measure received by multiple groups is assigned to each. A measure with no identifiable receiving country is likewise coded as “Unidentified.”

Measures are classified into three sectors. First, “Technology” covers semiconductors, advanced computing, electronic design automation, lithography, and related items (e.g., AI chips and servers, data-center hardware, and 5G/telecom equipment). Second, “Minerals” covers rare earths and critical minerals, excluding bulk commodities such as steel, aluminum, copper, and fertilizer. Third, “Other” covers all remaining products and sectors. Two small residual categories capture observations that fit both the Technology and Minerals criteria or cannot be assigned cleanly to either (“Ambiguous”), and observations that miss the information necessary to assign to a sector (“Missing”).[19]

From the sample of all 1,970 firm-quarter observations, the analysis builds two subsamples. One maps out the global corporate narrative on export restrictions (summarized in Figure B1 below) and uses 1,263 mentions covering single, identified imposing and receiving jurisdictions across the three sectors (Technology, Minerals, and Other). The other is the EMAD-relevant universe (the denominator in Figure 1 in the paper) that includes only 567 mentions of USALL technology controls on China and CHN mineral controls on USALL or ROW.

The EMAD-relevant universe is the union of the two sets: USALL technology controls on CHN and CHN mineral controls on USALL or ROW, totaling 567 firm-quarter observations. Relative to the large sample, it excludes the “Other” sector, measures imposed by ROW, and any Technology or Minerals measures that fall outside the two sets, e.g., U.S. technology controls received by a country other than China.

B.3 Defining U.S. and Chinese export control policies: intended effects, unintended benefits, and side effects

The baseline definition of U.S. and Chinese policies in Figure 1, Policy 1 and Policy 2, respectively, is narrow. It is restricted to export-control mentions that identify the policy’s intended negative effect on the receiving country’s firms and the unintended benefit for the imposing country’s firms. For example, for Policy 1, these are Chinese firms reporting negative effects of U.S.-allied technology controls and U.S.-allied firms reporting positive effects. For Policy 2, they are U.S.-allied or rest-of-world firms reporting negative effects of Chinese mineral-control policies and Chinese firms reporting positive effects.

These sign restrictions help identify a more accurate policy signal but significantly shrink the sample. The restricted Policy 1 series contains 211 observations, and the restricted Policy 2 series contains 21 observations across the plot window. The 2025Q1–Q4 quarterly counts of restricted Policy 2 are 1, 8, 6, and 4 observations. As a result, individual quarterly peaks should be read cautiously. The relevant result is the pattern and the multi-quarter sustained impact on the corporate narrative.

An alternative, a broader definition (Broad Definition in Panel A of Figure B2) drops these sign restrictions and counts every observation in the EMAD-relevant universe. We discuss robustness under this broader definition in section B.5 below.

B.4 The global corporate narrative on export restrictions

The Sankey diagram in Figure B1 visualizes the subsample of 1,263 observations with an identifiable sector, single-group imposer, and receiver country. Three features stand out.

First, when the U.S. and its allies impose a measure, nearly half (47 %) of U.S.-imposed measures are in technology and aimed at China (USALL × Technology × CHN mentions). Second, when China imposes, half of its measures are in minerals, and almost none are in technology, reflecting its specialization in the supply chain. Chinese measures are mostly aimed at the U.S. and its allies. That is a mirror image of the U.S. footprint on the corporate narrative, but the Chinese impact is an order of magnitude smaller than the U.S. flow.

Finally, the rest of the world represents only 4% of the imposing mentions but 10% of the receiving ones, mostly through other sectors. Mentions of measures in other sectors (neither Technology nor Minerals) account for half of the diagram. Overall, the picture is one of reciprocal but asymmetric weaponization of economic interdependence between the U.S. and China.

B.5 Robustness

In this section, we discuss the robustness of the inference made based on Figure 1 in the paper. The two main findings of mutual vulnerability consistent with the EMAD hypothesis survive several robustness checks.

B.5.1 Alternative policy definitions

The baseline definition of U.S. and Chinese policies in Figure 1 (Policy 1 and Policy 2, respectively) is narrow, leading to a small total cumulative mention count. An alternative, broader definition drops the sign restrictions and counts every quarter-firm observation in the EMAD-relevant universe. Figure B2 compares the results reported in the paper with those obtained using this broader measure.

The comparison shows that the results would be even stronger, as much of the Policy 1 impact on the corporate narrative under the broad policy definition has negative side-effects from the measure imposed. Removing the positive-and-negative-effect sign restriction leaves the timing and 2025 narrowing result intact.

Under the broad definition, Policy 1 is uniformly higher throughout the 2019Q1–2025Q4 window: 519 mentions compared with 211 under the narrow definition. The gap is driven mainly by U.S.-allied firms reporting negative effects from U.S.-allied technology controls — about two-thirds of the excluded observations — including upstream suppliers such as ASML, Applied Materials, Cadence, and Synopsys that lost Chinese sales following the BIS rulings. The narrow definition, which is the version plotted in Figure 1 of the paper, excludes these self-inflicted-loss observations, which are neither an intended effect nor a side benefit. The restricted share is consequently lower.

Policy 2’s footprint is far smaller than Policy 1’s, under both definitions, and its two series track each other closely because the absolute difference is small (48 mentions under the broad definition versus 21 under the narrow one, a gap of 27 compared to Policy 1’s 308).

That gap is not the mirror image of Policy 1’s, however. It is driven mainly by USALL and ROW firms reporting that they benefited from Chinese mineral controls (14 observations), with only nine Chinese firms reporting negative effects from their own government’s policies, arguably few because of self-censoring. The broad Policy 2 series remains larger than the narrow one, especially in 2025.

The Broad–Narrow differential for Policy 1 therefore primarily reflects the imposer side effects of U.S.-allied technology controls on China: the self-inflicted losses of USALL suppliers. The smaller Policy 2 differential reflects a different mix, dominated by receiving-side firms reporting benefits rather than Chinese self-inflicted losses.

B.5.2 Denominator effects

Another concern is that the EMAD total denominator moves with the numerator. The relevant universe itself responds to the same export-control activity that affects the numerator. During 2024Q4–2025Q4, the denominator triples, from 17 to 57 quarterly observations.

However, the main findings can be confirmed against the absolute counts. The 2025 rise in Policy 2 is unambiguous in raw counts. The three quarters from 2025Q2–2025Q4 are the three largest restricted Policy 2 counts in the 2019Q1–2025Q4 sample, with 8, 6, and 4, respectively. Before 2025Q2, Policy 2 is nearly always zero, with only one observation each in 2023Q4, 2024Q4, and 2025Q1. The narrowing gap between the two series in 2025 reflects both the growth in the denominator and the rise of Policy 2, not a decline in Policy 1.

B.5.3 Ambiguous, missing, and unclassified observations

Table B1 provides counts and shares of all variables used in the empirical analysis. It begins with 1,970 firm-quarter observations over 2019Q1–2025Q4 and reports the sector residuals, country residuals, and multi-membership cases alongside the 567 EMAD-relevant observations used in Figure 1.

The sizable “Unidentified” imposer and receiver counts are mostly “Other” sector mentions (about three-quarters), where the firm discusses export controls without naming a relevant sector or counterpart country. They are not classification errors and do not enter the EMAD measures.

Ambiguously classified mentions represent a very small share of the total. Reassigning the 37 such Technology/Minerals observations to either sector changes counts by at most 14 observations but does not change the qualitative pattern in Figure 1 in the paper.

Finally, an additional concern could be the fact that the GCAP Geoeconomics Monitor has better coverage of U.S. and internationally listed firms with English-language analyst coverage. Chinese A-share firms and many smaller European firms may be under sampled. This may inflate Policy 1 counts relative to Policy 2 by construction. As Chinese firms are likely under-sampled, and the Policy 2 share still rises in 2025, the underlying Chinese mineral geoeconomic pressure is plausibly stronger than what our measure captures.

Table B1. Sample Composition, Residuals, and EMAD Policy Definitions

B.6 Replication

All figures and tables in the paper and this appendix are produced by an R script applied to the GCAP Geoeconomics Monitor data. The main script, “chips_and_tokens_pipeline.R,” applies the country and sector group definitions, computes the Policy 1 and Policy 2 series under both the Baseline (Narrow) and Alternative (Broad) definitions (called Reading B and A, respectively, in the code), and produces Figure 1 in the paper, the Sankey Figure B1 and Figure B2, and Table B1. The script also produces complete intermediate outputs — including the consistency-check log and other descriptive cross-tabulations not reported here. Reproduction requires R 4.2 or later and the CRAN packages listed in the supplementary README.

Fragmentation Events and Stablecoin Demand: A Small Event Study

This Appendix describes the stablecoin event study in the paper and provides event-level results. The analysis focuses on seven geopolitical or geoeconomic events listed below that may be associated with risk of triggering new or tighter U.S. financial sanctions or that might strengthen incentives of affected parties to leave the U.S. dollar-based international payment system. The event can therefore be associated with a higher value of operating outside the U.S. regulatory perimeter.

C.1 Event List

Given the more frequent occurrence of geopolitical and geoeconomics events in 2025 and early 2026, the December 18, January 3, and January 18 30-day post windows overlap. Consequently, the change in the market-cap ratio after these three events reflects the influence of previous events, particularly in the second half of the post horizon between day 15 and day 30 after the event. However, the results suggest that the strongest effects occur in the first 15 days following each event over which there is no event overlap.

C.2 Outcome Variable and Methodology

As an outcome variable, the study focuses on the ratio of Tether’s USDT market capitalization to that of Circle’s USDC. Tether is the largest USD stablecoin, domiciled outside the United States, and largely operating outside the U.S. regulatory perimeter. USDC is the second-largest USD stablecoin, headquartered in the United States, and has operated under a de facto regulated regime even prior to the passage of the GENIUS Act in July 2025. This ratio can be interpreted as a proxy for the relative demand for offshore and onshore dollar stablecoins, respectively. By focusing on the ratio of the two USD stablecoin capitalizations, the event analysis partials out the effects of common factors that could impact broader stablecoin demand across all issuers.

The width of the event window is 30 days before and after these seven events (denoted k, for a total of 60 days. In addition, a linear time trend is estimated on each of the seven 30-day pre-event samples based on the following regression:

\frac{\text{USDT}_{k,t}}{\text{USDC}_{k,t}} = \alpha_{k,\text{pre}} + \beta_{k,\text{pre}} \cdot t + e_{t}, \quad t=1, \dots, 30

where t is a linear time trend. Then, the market cap ratio for each event k’s entire 60-day window is de-trended with the corresponding fitted value:

\frac{\widehat{\text{USDT}_{k,t}}}{{\text{USDC}_{k,t}}} = \hat{\alpha}_{k,\text{pre}} + \hat{\beta}_{k,\text{pre}} \cdot t, \quad t=1, \dots, 60

Figure C1. Tether-Circle Market Cap Ratios Around Fragmentation Events

Note: Each panel shows the market cap ratio within a 60-day window around each of the seven geopolitical or geoeconomic events listed above (red, solid line). The gray solid line represents the estimate pre-trend. The vertical black, dashed lines indicate the event.

C.3 Results

Figure C1 plots the results. In five out of seven cases, the Tether–Circle market cap ratio moves sharply upward and away from its pre-trend right after a fragmentation event. The exceptions are Maduro’s capture and the onset of the war in Iran. In the case of the Iran War, the shift is delayed. The military operation started on February 28, 2026, and the ratio does not deviate from the pre-trend until about March 10, which coincides with a major escalation in the conflict when oil prices crossed $100 per barrel for the first time. The behavior of the market cap ratio after Maduro’s capture can be interpreted by noticing the uncertainty surrounding its geopolitical interpretation in its immediate aftermath, with U.S. administration initially giving conflicting signals on its plan for the post-Maduro regime. Another anomaly in the results is the reversion of the market cap ratio about two weeks after the event, which could reflect the progressively clearer U.S. position toward Venezuela rather than the stabilization in the dispute between the U.S. and Europe over Greenland.

The event study result shown in Figure 3 of the paper reports the de-trended cumulative change in the Tether-USDC market cap ratio, averaged over the seven geopolitical events, for each day within the window, with the day prior to the event date (t+h=-1) set to zero:

\Delta \overline{\text{Ratio}}_{t+h} = \frac{1}{7} \left[ \sum_{k=1}^{7} \frac{\tilde{\text{USDT}}_{k,t+h}}{\text{USDC}_{k,t+h}} - \sum_{k=1}^{7} \frac{\tilde{\text{USDT}}_{k,t-1}}{\text{USDC}_{k,t-1}} \right], \quad t+h = -30, \dots, 0, \dots, 30

where

\frac{\tilde{\text{USDT}}_{k,t}}{\text{USDC}_{k,t}} = \frac{\text{USDT}_{k,t}}{\text{USDC}_{k,t}} - \widehat{\left( \frac{\text{USDT}_{k,t}}{\text{USDC}_{k,t}} \right)}

---