An international currency works like a network protocol: its value comes not from being intrinsically superior but from everyone else already using it. A Chilean exporter billing a Korean buyer will often settle in dollars — the peso–won market is thin, while both currencies have deep dollar markets. Network effects create a positive feedback loop, and the equilibrium tends naturally toward a single dominant player.
Dollar dominance rests on four mutually reinforcing layers: reserves (roughly 56%–57% of allocated FX reserves, IMF data for 2025; the peak around 2000 was about 71%), invoicing, funding, and clearing — most cross-border dollar transactions must pass through banks located in the United States, so money with no American connection still falls under US jurisdiction. Hence the "exorbitant privilege" named by Giscard d'Estaing: having your currency held as the world's reserve means persistently cheap funding. Hence also the Triffin dilemma: supplying the world with your currency requires running sustained deficits, and those deficits erode confidence in it.
For unipolarity (steelman): a single unit of account minimizes transaction costs and maximizes liquidity, and in a crisis there is an identifiable lender of last resort — in 2008 and 2020 the Federal Reserve opened dollar swap lines to other central banks, stabilizing global dollar funding. For multipolarity (steelman): handing the financial core to one sovereign means that state's fiscal choices and legislation spill over as everyone's cost; multipolarity is less efficient but preserves an exit option.
Size is only a necessary condition: Japan was the world's second-largest economy for decades and the yen never became dominant — depth and predictability are what matter. Nor are the benefits one-sided: global demand bids up the currency, which depresses the competitiveness of the tradables sector.
Day 9 covered geographic chokepoints. Finance has chokepoints too, and they are more concentrated: the US banking system that dollar clearing must pass through, and SWIFT, the Belgium-headquartered messaging network (it moves no money, only standardized "who pays whom how much" instructions — but the world's banks are all plugged into it). A star topology makes the central node a control point by default — a by-product of efficiency, not a deliberate design.
What turns this into a tool is secondary sanctions: not merely barring domestic parties from transacting, but penalizing third-country parties that deal with the target, usually by cutting off their dollar clearing. For an international bank that is close to a shutdown, so it steers clear of sanctioned entities even where its own law permits the business. The real coercive force comes not from the extraterritorial reach of law but from private institutions' self-avoidance.
| Case | Mechanism | Costs and blowback |
|---|---|---|
| Iran and SWIFT | In March 2012 SWIFT disconnected sanctioned Iranian banks under EU Regulation 267/2012; access was partly restored after the 2015 nuclear deal took effect, then cut again in November 2018 | Trade and humanitarian payments were badly obstructed — and every country was shown that the pipes can be severed |
| Russian central bank reserves frozen | From February 2022, roughly $300 billion of Russian central bank reserves were frozen within G7 and EU jurisdictions, and several Russian banks were removed from SWIFT | Even another state's central bank reserves can be frozen, which changed reserve managers' risk models everywhere |
| The EU's counter-move | The EU has a blocking statute and set up INSTEX in January 2019 to route around the dollar and SWIFT | Only a handful of transactions were ever completed — a government can permit a firm to trade, but cannot absorb its risk of losing dollar clearing |
For use (steelman): between diplomatic protest and military action states have long lacked instruments; financial sanctions apply real pressure without direct casualties, can be targeted at specific entities, and can be lifted at any time. For restraint (steelman): their efficacy rests entirely on others voluntarily keeping their money inside your system, and the more often they are used, the stronger everyone's incentive to build backup channels; moreover the pain tends to fall on ordinary people rather than decision-makers, and the empirical record on whether sanctions change behavior remains contested. Both sides share one judgment: this is a form of power that depreciates with use.
"Being cut off from SWIFT means exiting the global economy" — SWIFT is a messaging standard, not money. Disconnection raises costs, but parties can fall back on direct correspondent-bank links. What is genuinely crippling is the loss of dollar clearing and the worldwide self-avoidance that secondary sanctions induce.
First, a clarification: the balance on your phone is not central bank money — it is a commercial bank's liability. Actual central bank money is available to the public only as cash, and as cash use shrinks the public is losing its direct access to it. A CBDC moves cash into digital form, issued and backed by the central bank itself.
Two routes are routinely conflated. Retail CBDCs face the public and concern financial inclusion, payment-system resilience, a counterweight to private payment monopolies — and, directly, privacy. Wholesale CBDCs circulate only among financial institutions. The geopolitical weight sits with the wholesale variety: today's cross-border payments pass through a chain of correspondent banks, slow and expensive, with every link subject to oversight; if two central banks swap tokenized balances in their own currencies on a shared ledger, settlement can in principle take seconds without touching any third country's clearing system — which is precisely where this connects to the previous card.
In favor (steelman): payments are critical public infrastructure, and full privatization breeds new monopolies and single points of failure; once cash disappears, the public should retain a means of payment that does not depend on any commercial institution's solvency. Against (steelman): a central bank issuing directly to the public puts banks' deposit-intermediation function at risk — in a crisis, one-tap transfers would amplify rather than dampen a run; and a state-operated ledger, whatever the intent behind it, creates an unprecedented degree of transaction visibility, so whether the boundary holds depends on that country's accountability mechanisms rather than on the technology.
A CBDC is not a cryptocurrency: the latter pursues decentralization and trustlessness, the former is a centralized state liability and need not even use a blockchain. Nor will it quickly unseat the dollar — the dollar rests on the depth of its reserve assets and legal predictability, and swapping the pipe is not the same as swapping what flows through it.
"De-dollarization" actually bundles three separate things, and conflating them guarantees a muddled argument. ① The settlement layer — bilateral trade shifting to local-currency settlement; the lowest bar, and genuinely growing. ② The reserve layer — central banks cutting dollar holdings; a much higher bar, since it requires substitutes of comparable scale, and the world is short of safe assets matching the Treasury market in size and liquidity. ③ The systemic layer — the emergence of an alternative global unit of account, which requires network effects to flip wholesale.
Once separated: settlement-layer change is real, the systemic layer has no evidence yet, and the reserve layer is moving slowly. Note also that a sizeable share of quarterly moves in reserve shares comes from valuation effects rather than active buying or selling.
The erosion camp (steelman): the freezing precedent is now written into reserve managers' risk models and cannot be unwritten; the issuer's debt trajectory and political cycles keep chipping away at the "risk-free" label. And the case does not require a successor to exist — a system can slide from "one pole" into a "disorderly multipolarity" without ever passing through a new hegemon. The inertia camp (steelman): flipping network effects requires a coordination point, and no currency currently combines market depth, free capital movement and legal predictability, so states that genuinely want to cut exposure find nowhere to go. The two camps disagree far less about the facts than about the interpretation.
The most common is a layer-crossing fallacy: taking a settlement-layer headline ("two countries announce local-currency settlement") straight to a systemic-layer conclusion. The second is assuming some currency must take over — a fragmented multipolarity would not necessarily be fairer: it weakens the dollar's privilege while also weakening the unified crisis backstop.