The dollar's fraying edge.
Reserve share is slipping; the dollar's trading role is not. The honest read is erosion at the margin, not collapse at the core — a multi-decade relative decline in official reserves, set against still-overwhelming dominance in invoicing, FX turnover and safe-asset demand. The live risk is a catalyst-driven discontinuity, not a smooth glide.
Still #1 by ~37 points over the euro (20.25%) and a 30-year low — but ~92% of the 2025 quarterly drop was the dollar weakening in price, not central banks selling. Adjust for valuation before reading any single quarter.
Settling is not reserving. A country can pay for a cargo in yuan without trusting the yuan enough to save in it. Russia settles ~99% of its China trade in local currency — but that is sanctions-forced substitution, not a reserve shift. Most "de-dollarization" headlines confuse the two.
Dollar share, by function
Conflating these numbers is the single most common error in the debate. The dollar is ~57% of reserves but ~89% of FX trades — a currency can lose reserve share (a stock-allocation choice) while gaining transactional centrality (a network-liquidity fact).
| Function | USD share | Type | What it means |
|---|---|---|---|
| FX turnover | ~89.2% | current | On one side of ~9 of every 10 trades — and rising. The dollar is the FX pivot. |
| Trade invoicing | ~40% | current | USD+EUR together >80%; far above the US ~11% share of world trade. |
| Official reserves (COFER) | 56.77% | current | The one function where the decline is visible — down from ~71% in 1999. |
| SWIFT payments | ~50% | current | Dollar and euro dominate the messaging layer; RMB under 3%. |
| Trade finance | >80% | current | The deepest invoicing-adjacent moat. |
| Intl. banking claims | ~55-60% | current | The eurodollar / offshore-dollar system — the stickiest layer. |
Sources: BIS Triennial Survey 2025 (FX turnover); IMF COFER, Mar 2026 (reserves); ECB invoicing feature, Jun 2025; Federal Reserve 2025 dollar report (SWIFT, banking); IMF Gopinath, May 2024 (trade finance). All current but lagged to their release cadence.
Why the share is sliding
Six distinct mechanisms push the dollar's reserve share down, and collapsing them into one "the dollar is dying" story is the central analytical error of the debate. Three are deliberate portfolio choices; three are reactions to geopolitics and US policy. They move on different clocks and point the same way — but only one is central banks actually selling dollars, and it is the smallest.
The dollar's raw allocated-reserve share fell from 57.79% (Q1 2025) to 56.32% (Q2 2025) — but at constant exchange rates it would have been 57.67%, so ~92% of that quarterly drop was the dollar weakening in price, not central banks selling (IMF valuation blog, Oct 2025). Of a ~1.5-point headline fall, only ~one-tenth of a point was an actual allocation change. A weaker dollar mechanically shrinks the dollar slice of a fixed basket — like a falling stock shrinking its weight in an index nobody traded. Read any single COFER quarter without this and you mistake dollar weakness for de-dollarization.
With valuation set aside, here are the six mechanisms in the order that matters — the data-hygiene point first, then the deliberate diversification it is so often confused with, then the forces of geopolitics, policy and price.
Valuation, not selling. The correction above, restated as the first driver: the largest apparent mover is not a flow at all — a weaker dollar shrinks its slice of a fixed basket without anyone selling (~92% of the Q2 2025 drop).
Diversification for its own sake. Reserve managers spread into many small currencies, not one successor. "Other currencies" reached 6.13% of allocated reserves (Q4 2025, up from 5.61%) — more than doubled since 2021 — with the Swiss franc roughly quadrupling to ~0.8%. The destination is a long tail of Australian and Canadian dollars, not the euro or yuan.
Sanctions and seizure fear (post-2022). Freezing ~US$300bn of Russian central-bank assets turned a tail risk into a standing line item: dollar reserves are entries in US-reachable institutions, switchable off by policy [DIR]. The fingerprint is in surveys — 73% of central bankers expect the dollar's share to fall and 43% plan to add gold, both record readings. Intent, running well ahead of capacity.
Multipolarity and political signaling. BRICS+ supplies the desire for a non-dollar layer ahead of the capability. Russia and China report settling ~99% of ~US$240bn (2025) bilateral trade in local currency [DIR] (official Russian source — a political claim, not audited data). But this is settling, not reserving: paying in yuan is not trusting it enough to save in it.
Structural gold accumulation. The clearest non-fiat expression of the same impulse — sanction-proof, no issuer. Central banks bought 863.3t in 2025 against a 473t 2010–2021 average, lifting gold past the euro to the #2 reserve asset — part quantity, part price (gold ~27% of reserves at end-2025 prices vs ~16% at end-2023). Full treatment in the cross-asset chapter.
US fiscal and institutional credibility shocks. The catalyst channel — episodic, but the strongest evidence the tail is real. The Treasury's own Borrowing Advisory Committee described April 2025 as a period when simultaneous declines in US equities, Treasuries and the dollar led some investors to question Treasury safe-haven status and even dollar reserve status. When the issuer's own debt managers write that down, the tail is no longer hypothetical.
USD 56.77% (a 30-year low, still #1 by ~37 points) · EUR 20.25% (the only credible fiat #2) · "other" 6.13% (up from 5.61%, doubled since 2021) · RMB 1.95% (far below its trade weight — capital controls bind). Total global FX reserves US$13.14tn. Diversification shows up not as a rival gaining 14 points, but as the long tail and gold quietly thickening.
Sources: IMF COFER Data Brief, Mar 2026 (shares; from Q3 2025 the unallocated bucket is dropped and composition imputed — 10.75% imputed in Q4 2025, so pre-Q3-2025 shares are not directly comparable). IMF valuation-effects blog, Oct 2025 (the 92% correction). World Gold Council FY2025 and 2025 survey (gold tonnage, ~20%-by-value, the 73%/43% readings). ECB International Role of the Euro 2026 (gold price decomposition). US Treasury TBAC 2025 (April-2025 episode). Russia-China settlement is official-Russian-source and labelled [DIR].
The alternatives, and their limits
Ask "what could replace the dollar?" and the question is mis-posed. A true successor would have to clear five bars at once: an open capital account, a single deep safe asset at scale, a working settlement function, jurisdictional trust, and enough collective belief that everyone else will switch too. Each contender clears some handsomely and fails exactly one — and the failures are not the same failure, which is precisely why they do not add up to a successor.
So the frame is not a horse race with a likely winner but a field in which every runner is lamed in a different leg. Each card below takes the genuine strength first, then the exact binding limit.
The eurostrong, lawful — and capped20.25% of reserves · 28.5% of FX turnoverhigh
The only credible second fiat: deep, liquid, rule-of-law, and the default home for diversification that doesn't go into gold. It held 20.25% of allocated reserves (Q4 2025) and one side of 28.5% of FX turnover (ECB, Apr 2025), and the ECB's expanded EUREP facility now supplies backstop euro liquidity to non-euro-area central banks.
No single euro safe asset at Treasury scale. The bloc issues sovereign debt through nineteen national treasuries of differing credit quality, not one homogeneous bond a manager can sell instantly in a crisis — and behind that, no fiscal union to manufacture one. Its constraint is plumbing the union has chosen not to build.
The renminbisuperpower trade, closed account2.94% of payments · 1.95% of reserveshigh
No challenger has more commercial gravity. China is the world's largest trading nation and has spent a decade building the rails to settle trade in yuan — CIPS, swap lines, and a digital yuan that settled its first cross-border oil cargo with Saudi Arabia in September 2025 [DIR]. Inside the aligned bloc, where it faces no competition, it works.
You cannot freely hold or move it. Capital controls are the hard ceiling — a policy the system depends on. Despite that vast trade base, the renminbi is just 2.94% of SWIFT payments (Nov 2025), 1.95% of reserves, 8.5% of FX turnover and under 2% of trade invoicing. Convertibility, not capability, is what binds.
Goldperfect store, cannot pay~20% of reserves by value · now #2high
The one reserve asset no government can freeze, print or default on — no counterparty, no issuer, no jurisdiction. It is the post-2022 hedge of choice: central banks bought 863.3t in 2025 (vs a 473t 2010–2021 average), lifting gold past the euro to the world's #2 reserve asset by value.
Gold cannot settle a payment — no yield, illiquid for daily trade, too volatile to anchor invoices. And its ascent is part price: the ECB puts gold at ~27% of reserves at end-2025 prices but ~16% at end-2023 prices (Treasuries ~22% vs ~26%). "Gold passed Treasuries" is real rotation amplified by a rally — a flight to a store of value, not a means of exchange.
The BRICS "Unit"a design, not a currency40/60 prototype · no summit commitmentlow
The political will is real and an engineering sketch exists — a gold-anchored prototype (40% gold, 60% a currency basket) floated in a pilot dated 31 October 2025 [DIR], with the New Development Bank already lending in local currencies. As an internal unit for settling member trade balances, the concept is coherent.
The obstacle is political, not technical — and decisive. A common currency needs a common central bank, a unified authority, and the trust to surrender sovereignty; China and India are rivals with a contested border. The BRICS Rio 2025 summit declaration contained no common-currency commitment [DIR]. A research prototype and a press narrative; not a state that exists.
CBDC rails (mBridge)real plumbing, corridor scale~$55.49bn lifetime · vs $9.6tn a daymedium
Proof an alternative rail works: by linking several central banks' digital currencies on a shared ledger, mBridge settles cross-border payments without SWIFT or a dollar correspondent — bypassing the chokepoint sanctions exploit. From $22mn of pilot value in 2022 to ~$55.49bn cumulative across 4,000+ transactions by late 2025.
Corridor erosion, not replacement — and a rail is not a reserve asset. That ~$55.49bn of lifetime volume compares with ~$9.6tn of FX turnover every day, so one day of the dollar FX market is ~170× mBridge's entire history. The e-CNY is ~95.3% of volume, so it inherits the renminbi's capital-control ceiling; the BIS stepped back in October 2024, leaving no live feed.
Dollar stablecoinsnot a rival — the opposite~99% dollar-linkedhigh
Here to correct a category error. Stablecoins are filed under "dollar alternatives" because they are crypto-native and bank-bypassing — but ~99% of stablecoin market capitalisation is dollar-linked, backed predominantly by short-dated US Treasury bills. They are private dollarization in a digital coat.
They cut the other way. Far from an exit, dollar stablecoins extend the dollar's reach — exporting access to inflation-prone economies and creating a fresh, structural bid for Treasury bills. The most visible "alternative" in the retail imagination is, mechanically, one of the dollar's newest demand channels. It belongs here only to be ruled out.
A switch of reserve currency is a coordination problem, not a quality contest. The dollar is sticky because everyone uses it because everyone uses it — a unilateral exit is costly even when a collective exit would be rational. Breaking that needs a focal point: one obvious alternative everyone can switch to at once. The field has none — each rival fails a different necessary condition, so there is no Schelling point a coordinated move could land on.
| Candidate | The test it fails |
|---|---|
| Euro | Unified safe asset — no fiscal union behind a single bond at scale. |
| Renminbi | Open capital account — controls are the policy, not a bug. |
| Gold | Medium of exchange — a store of value that cannot settle a payment. |
| BRICS "Unit" | Political precondition — no common central bank; a design, not a currency. |
| CBDC rails | Safe asset — solve payment access in one corridor, supply none; inherit the RMB ceiling. |
| Stablecoins | Not an alternative at all — dollar-extending, the opposite. |
Sources: IMF COFER Mar 2026 (reserve shares) · BIS Triennial 2025 (FX turnover, daily volume) · ECB International Role of the Euro 2026 (euro indicators, EUREP, gold price-vs-quantity decomposition) · SWIFT RMB Tracker Dec 2025 · World Gold Council FY2025 · Atlantic Council (mBridge volume, e-CNY share, BIS Oct-2024 step-back) · Federal Reserve 2025 dollar report (stablecoins). The BRICS "Unit" prototype and Rio-2025 read, the e-CNY oil corridor, and the fiscal-union interpretation are [DIR]; no live feed exists for mBridge volume post-October 2024.
Who is actually de-dollarizing
It is bifurcated, not global. The measured fall in the dollar's reserve share is driven almost entirely by a non-aligned bloc being pushed — or pushing itself — out of the dollar, while the allied core stays firmly, voluntarily attached. Treating "the world" as one de-dollarizing actor is the single biggest framing error in the debate. Five cohorts move at five speeds, for five reasons, and one large counter-current runs the other way.
Read the cohorts as a forced vanguard, a strategic builder, the hedgers, the gold-buyers, and the anchors — not as a single tide. The same headline ("dollar share at a 30-year low") is real and misleading at once: genuine at the margin, driven by the bloc with the strongest motive, and offset at the core by allies who bought Treasuries in size. Direction is real; breadth is not.
| Cohort | What they're doing | The hard number | Type |
|---|---|---|---|
| Forced vanguard — Russia | Sanctioned out of the dollar; settles almost all China trade in rubles and yuan. De-dollarization by exclusion, not choice. | ~90–99% local-currency; ~US$240bn 2025 trade | [DIR] |
| Strategic builder — China | Building the parallel rails: CIPS, mBridge, swap lines, e-CNY oil settlement. The supply side of any non-dollar future. | 1st digital-yuan oil trade Sep 2025; ~90% Iran oil non-dollar | [DIR] |
| The hedgers — the Gulf | Maximising optionality: Saudi let petrodollar exclusivity lapse (Jun 2024) and built yuan infrastructure — but oil is still overwhelmingly USD-settled and the riyal peg anchors it. | Petrodollar exclusivity lapsed Jun 2024; peg intact | [DIR] |
| The gold-buyers — broad EM | Diversifying into the one asset no one can freeze. "More gold and more optionality," not "sell all dollars." | Poland +102t (gold 28% of reserves), Kazakhstan +57t, Brazil +43t, Turkey +27t | WGC |
| The anchors — US allies | Not de-dollarizing. Europe, Japan, the UK and Australia stay deep in the dollar system and bought Treasuries in size, offsetting rival sales. | ~US$463.9bn USTs bought in 2025 | Bloomberg |
| Counter-current — private sector | Re-dollarizing. Households and firms in inflation-prone economies adopt dollar stablecoins — a private dollar expanding as the official one diversifies. | ~99% of stablecoins dollar-linked | Fed |
Note: Russia-China settlement (~90–99%, finance ministry 99.1%) and the ~US$240bn trade figure rest on official Russian-government statements; Iran ~90% non-dollar and the Sep-2025 China-Saudi digital-yuan trade are single-source — all labelled [DIR], direction not audited data. Gold tonnes are cohort purchases (WGC FY2025); the site-wide 2025 total is 863.3t.
Russia is the only true full exit, and it was pushed: ~90–99% of its China trade now settles in rubles and yuan [DIR], on ~US$240bn of 2025 trade. China supplies the rails — CIPS, mBridge, e-CNY oil — as a strategic builder, not a defector.
The Gulf and EM gold-buyers diversify without divorcing: Saudi let petrodollar exclusivity lapse (Jun 2024) yet still settles oil in dollars [DIR]; Poland (+102t), Kazakhstan, Brazil and Turkey bought gold — more optionality, not "sell all dollars."
US allies are not de-dollarizing. Europe, Japan, the UK and Australia bought ~US$463.9bn of Treasuries in 2025, offsetting rival sales — while the private sector re-dollarizes through stablecoins that are ~99% dollar-linked.
So the arithmetic of decline is a story about who is leaving, not about everyone leaving. Subtract the forced bloc and the picture is near-stable: the allied core is buying, the private dollar is expanding, and the diversifiers are reaching for gold rather than the euro or the yuan. The dollar is not being voted out — it is being walled off from the few with the strongest reason to build the wall.
Why the dollar stays sticky
The forces in the last three chapters all push one way — down. So why does the share fall in tenths of a point a year rather than collapse? The answer is not loyalty. It is structure — five reinforcing mechanisms that make a unilateral exit costly even for a country that wants one.
The deepest reason is the least intuitive: the dollar is held because it is held. Its value to any one user rises with the number of others who use it, so a single country exiting alone pays the full switching cost and captures almost none of the benefit — even when a coordinated exit would leave everyone better off. And as Chapter II showed, escaping that trap needs a focal alternative the field does not have — so network effects don't merely favour the incumbent, they lock it in.
Incumbency and network effects. A firm invoices in dollars because its suppliers, banks, insurers and hedging markets already are; its bank funds in dollars because its clients invoice in them. Each leg feeds the next, so the switching cost is not one decision but a whole web of them re-priced at once.
The liquidity moat — and it's widening. The dollar sits on ~89.2% of FX turnover (BIS, Apr 2025), up from 88.4% in 2022, while daily volume rose to ~US$9.6tn. The mechanism is the FX pivot: the world trades currency A to USD to B rather than directly, so even non-dollar trade manufactures dollar transactions. A displacement story predicts this share falling; instead it rose.
Rule of law — two flavours of risk, not risk-free vs risky. The 2022 freeze of ~US$300bn of Russian reserves [DIR] proved dollar reserves are weaponizable — but through a transparent, allied, rules-based process. The alternative store, the renminbi, sits behind an opaque capital-control regime that is its own seizure risk. Managers weigh jurisdictional risk they can read against risk they cannot.
A safe-asset pool with no rival at scale. The Treasury market is ~US$28tn+ of homogeneous, liquid collateral. Foreigners held ~US$9tn — 32% — in Q1 2025, and a record ~US$9.35tn (custodial) in Nov 2025, even as reserve share fell. At 32%, the foreign share is broadly comparable to euro-area, Japanese and UK government debt — normal participation in a market with no substitute its size.
The Fed swap-line backstop. When global dollar funding seizes, only the Fed can supply unlimited dollars to foreign central banks — drawings peaked at ~US$585bn (2008–09) and ~US$450bn (2020). No rival has anything comparable. This turns the dollar from a peacetime convenience into a crisis necessity: an ally that diversifies at the margin still wants the swap line open when markets break.
| Safe-asset pool | Why it can't absorb global reserves at scale |
|---|---|
| US Treasuries | ~US$28tn+ of homogeneous, deeply traded, repo-able collateral; foreigners hold ~US$9tn (~32%). The benchmark every rival is measured against. |
| German Bunds | The euro area's closest analogue, but a fraction of the size — and no single euro bond, only nineteen national issuers of differing credit. [DIR] |
| Japanese govt bonds | Large in total, but largely held domestically; little free float for foreign managers seeking liquid size. [DIR] |
| Chinese govt bonds | Gated by capital controls — the holder cannot freely enter or exit, disqualifying the asset for crisis-sellable reserves. [DIR] |
| Gold | No issuer to default, but yields nothing and cannot settle a payment; a store of value, not a liquid collateral market. |
Russia and China can route nearly all of their bilateral trade in rubles and yuan [DIR] because two conditions hold at once: the political imperative is overwhelming, and the relationship is concentrated — two parties, one corridor, a shared motive. A global reserve asset clears a far higher bar: it must satisfy third-party liquidity, convertibility, hedging, legal and capital-account needs across unrelated countries, each of which must independently trust it enough to save in and sell instantly in a crisis. Bilateral settlement is a private toll road two neighbours build; a reserve currency is the public highway the whole world must be able to drive on without asking permission.
Sources: BIS Triennial 2025 (FX turnover ~89.2%, daily ~US$9.6tn) · Federal Reserve 2025 dollar report (foreign UST ~US$9tn = 32% Q1 2025; euro-area/Japan/UK comparability; swap-line peaks ~US$585bn 2008–09 and ~US$450bn 2020 — historical peaks, not current drawings) · Atlantic Council Feb 2026 (record ~US$9.35tn custodial, Nov 2025) · ECB 2026 (EUREP). The ~US$28tn+ scale and rival-market relative sizes are directional comparisons; the 2022 freeze (~US$300bn) and Russia-China settlement are [DIR].
The Treasury-demand link
This is where the reserve question stops being abstract and starts costing money. The chain is short and mechanical: as the dollar's official role erodes, the buyer who once took Treasuries regardless of price — a foreign central bank parking reserves — is slowly replaced at the margin by a private investor who buys only if the yield compensates. A price-insensitive buyer gives way to a price-sensitive one, and the price of that shift is a higher term premium.
But the honest read inverts the headline. The dominant force lifting the term premium is not de-dollarization — it is the United States' own fiscal trajectory, supplying bonds faster than the world wants to absorb them. De-dollarization removes a buyer; fiscal policy supplies the paper. The two do not cancel — they compound.
The buyer base, today. Foreign investors held ~US$9tn of marketable Treasuries — 32% of the total — in Q1 2025; domestic private hands 55%, the Fed 13%. The foreign share fell from almost 50% in 2014 but has barely moved since 2022, and now sits in line with the foreign-held share of euro-area, Japanese and UK debt. Less an exodus than a plateau.
The marginal-buyer shift, in one print. The Treasury's March 2026 TIC release shows it live: a net US$150.7bn inflow, made of a US$162.1bn net private foreign inflow against a US$11.4bn net official outflow. Officials stepped back; private money more than filled the gap. The bid did not vanish — it changed hands, from a holder indifferent to yield to one that demands it.
The price of that swap — the term premium. The extra yield investors demand to hold a long bond rather than roll short ones — the term premium — has risen ~70bps on the ACM 10-year measure since September 2024, to ~0.59–0.8%, its highest sustained level since 2011. On ~US$28tn of marketable debt, each sustained 100bps is roughly US$280bn of added annual interest as the stock refinances.
The shock that did most of it. Of that ~70bps, ~44bps arrived after the April 2025 tariff shock — the same episode in which equities, Treasuries and the dollar fell together and the Treasury's own Borrowing Advisory Committee recorded investors questioning Treasury safe-haven status. The premium doesn't drift up on de-dollarization alone; it jumps on confidence events. Attribution is genuinely contested [DIR].
The larger jaw. The 2026 federal deficit is running near 6.6% of GDP [DIR], forcing the Treasury to issue paper into a market with fewer captive buyers. More supply at a given demand curve means a lower price and a higher yield — the driver the US controls, and on most readings the dominant one.
The smaller, structural jaw. As official managers diversify, each auction leans a little more on price-sensitive private demand. The effect is real but bounded — foreign holdings are at record nominal levels (~US$9.35tn custodial, Nov 2025). So the channel is a rising risk premium, emphatically not a buyers' strike.
Fiscal policy steepens the supply curve while de-dollarization flattens the most reliable part of the demand curve. Either alone is manageable. Arriving together — more bonds, fewer indifferent buyers — is what carries a term premium to a fifteen-year high without a single failed auction.
The structural reading (Gourinchas, building on Triffin and Rey) [DIR]: the US cannot indefinitely expand its debt to supply the world's safe asset without eventually eroding confidence in its own fiscal sustainability — the privilege contains the seed of its own limit. The counter-view (Econlib) [DIR] holds this is no iron law but a policy choice: self-inflicted profligacy, fixable by consolidation. The distinction decides whether fiscal reform can relieve the term premium, or whether only confidence can move it. We present it as live uncertainty because the evidence does not resolve it.
Two facts hold at once. Foreign Treasury holdings are at an all-time high in dollars (~US$9.35tn custodial Nov 2025; ~US$9tn = 32% of the marketable stock Q1 2025) — the demand is there. And the term premium is at a fifteen-year high. The correct frame is "the world still funds the US, but charges more for it," never "the world is walking away." The marginal buyer changed; the buyer base did not collapse.
Sources: Federal Reserve 2025 dollar report (foreign holdings ~US$9tn = 32% Q1 2025; 55% domestic-private / 13% Fed) · US Treasury TIC, Mar 2026 (the US$162.1bn private inflow vs US$11.4bn official outflow split) · NY Fed ACM (10-year term premium ~0.59–0.8%, +~70bps since Sep 2024, ~44bps post April-2025 — the level is firm, its fiscal-vs-de-dollarization decomposition is [DIR]) · Atlantic Council Feb 2026 (record ~US$9.35tn custodial, Nov 2025). The 2026 deficit ~6.6% of GDP is [DIR]; the New Triffin Dilemma (Gourinchas/Rey) and its fiscal-choice counter (Econlib) are both [DIR] — a genuine, unresolved dispute.
The cross-asset system
The four legs of the reserve story — gold, the dollar, the Treasury curve and stablecoins — are usually reported one at a time. Read together they reveal what no single number can: gold climbing while the dollar holds firm, and a private dollar bidding the front of the curve precisely as the official dollar drifts off the back. The signal is in the interaction, not the parts.
| Leg | The standout fact | What it tells you |
|---|---|---|
| Gold — the standout | 863.3t bought by central banks in 2025 vs a 473t 2010–2021 average; gold has passed the euro to the #2 reserve asset by value. | The diversification trade with a destination — but the share is part price (ECB: gold ~27% of reserves at end-2025 prices vs ~16% at end-2023). |
| The dollar — firm | The dollar index sits near 99.8 mid-June 2026 (a Hormuz/Iran safe-haven bid), after a weak H1 2025 (down ~10%, worst start since 1973). [DIR] | A "declining" reserve currency that is strong in the spot market — the paradox dissolves once you know most reserve-share decline was valuation, not selling. |
| Stablecoins — re-dollarization | ~99% of stablecoin market cap is dollar-linked; supply ~US$323bn (Jun 2026), about half held in T-bills; GENIUS Act signed 18 Jul 2025. | The most-cited "dollar alternative" is mechanically a dollar demand channel — a digital eurodollar that bids short-dated Treasuries. |
| The Treasury curve — split | Stablecoins ranked the 3rd-largest net buyer of US Treasuries across 2024–Q1 2025, even as official foreign demand drifted. | Two flows hit opposite ends: private bills demand supports the front; term-premium pressure lives at the long end. |
Here is the nuance aggregate "is foreign demand falling?" framings miss. De-dollarization and re-dollarization are not cancelling out — they land on different maturities. Stablecoin issuers back their tokens one-for-one, so every new token mechanically bids short-dated Treasury bills; stablecoins were already the 3rd-largest net buyer of USTs across 2024–Q1 2025. That demand sits at the front of the curve. The term-premium pressure from drifting official demand and heavy fiscal supply lives at the long end. The same curve is supported at the front and pressured at the back, by two different flows — and TBAC's sketch of stablecoins reaching ~US$2tn by 2028 (a projection, not a print [DIR]) would only deepen the front-end bid.
Gold and the dollar are historically meant to move apart — a stronger dollar usually weighs on gold. Yet in 2025–26 both are strong at once, and that co-strength is itself the signal. It tells you gold is bought for reasons orthogonal to the dollar's cycle: sanction-hedging after the 2022 freeze, and structural diversification into an asset no government can freeze, print or default on. If gold were simply an anti-dollar trade, a firm dollar would have capped it. That it has not is the cleanest evidence the central-bank bid is structural, not a bet against the exchange rate. The WGC also estimates unreported buying ran at ~57% of the 2025 total — so the true pace is probably faster than the 863.3t print suggests.
Read as a system, the four legs tell one coherent story. The official dollar drifts at the margin; gold is the genuine diversification winner, though part of its ascent is price; the private dollar expands through stablecoins; and the Treasury curve is supported and pressured at once, at opposite ends. The dollar's total footprint may even be widening as its sovereign-balance-sheet footprint narrows — which is why the honest verdict is erosion at the edge, not displacement at the core.
Sources: World Gold Council FY2025 (gold tonnage, 473t average, #2-asset, the ~57% unreported-buying gap) · ECB International Role of the Euro 2026 (gold price-vs-quantity decomposition) · Federal Reserve 2025 dollar report (~99% dollar-linked) · Richmond Fed Jun 2026 (the ~US$323bn supply print; site band ~US$280–320bn) · Brookings Oct 2025 (3rd-largest net UST buyer). On the dollar index: the ~99.8 reading is the ICE US Dollar Index (DXY), a proprietary six-currency gauge with no free feed — a different series from the Fed's broad trade-weighted dollar index (~120, base-period indexed); the two are not interchangeable. The DXY level and the maturity-split reading are [DIR].
Catalysts and scenarios
A reserve order this entrenched does not drift away on sentiment. It moves only when a shock changes one of four things a reserve manager actually prices: the legal safety of the asset, the liquidity behind it, the payment access it buys, or the balance-sheet cost of holding it. A catalyst that touches none of these is a headline, not a transmission — and most "the dollar is finished" stories fail that test.
Run the live catalysts through the filter and the result is sobering in both directions: the loudest shocks move the dollar least, and the quietest move it most. Below, each shock is mapped to the one constraint it actually changes; then six council-blended scenarios put probabilities, cross-asset effects and falsifiers on the table. The base case is glacial drift at about half the weight. The risk that matters lives in the tails.
Hormuz / petroyuan — payment access, signaling not flow. The loudest catalyst, the weakest transmission. March 2026 reports that Iran might gate the Strait on yuan settlement [DIR] were never officially confirmed, and the flow is tiny — Iran sells China ~US$8bn of oil a year [DIR]. The ECB/IMF invoicing work finds no robust evidence of effective oil de-dollarization. High narrative power, low confirmed magnitude.
Sanctions escalation — legal safety, the real engine. The strongest real de-dollarization force, irreversible at the margin: each package pushes more of the sanctioned bloc permanently onto non-dollar rails, and plumbing once built stays built. The twist: the same sanctions that expel an adversary tighten allied cohesion. It de-dollarizes the periphery and re-anchors the core in one move.
US fiscal scare — balance-sheet cost, the live tail. The only catalyst that could move the dollar fast — a disorderly term-premium spike could trigger flight from it. But it is treacherous: acute risk-off usually triggers flight to dollars. April 2025, when equities, Treasuries and the dollar fell together, was the rare episode where the reflex inverted. Which way the next scare fires is the hinge.
Risk-off shock, non-US origin — liquidity, and it strengthens the dollar. A global recession or non-US crisis sends capital scrambling for the deepest, most sellable asset, overwhelming any political discomfort. The dollar's worst enemy and best friend are the same emotion — fear — aimed at a different target. This is why de-dollarization is never a one-way bet.
Payment-rail shock — payment access, slow burn. CIPS, mBridge and bilateral ledgers transmit not through one event but corridor by corridor, making a non-dollar lane cheaper or safer than correspondent banking one trade at a time. The scale check is unforgiving: mBridge's ~US$55.49bn lifetime volume is dwarfed by one day of the dollar's ~US$9.6tn FX market. Real, directional, measured in years.
Worth stating plainly, because the petroyuan narrative does more analytical damage than any other: as of mid-2026 there is no confirmed official Iran directive, the confirmed Iran-to-China oil flow is ~US$8bn a year [DIR], and the ECB/IMF invoicing dataset shows no robust evidence of effective oil de-dollarization. Treat petroyuan as a signaling event — a marker of intent and a precedent risk — not an established corridor break. Intent runs years ahead of flow, exactly as in the reserve data.
Six scenarios, council-blended. The probabilities are deliberately round and explicitly subjective — they convey relative weighting, not precision, and every one is directional. Read down the probability column for the shape of the risk; read across each row for how that world shows up in Treasuries, gold and the dollar, and what would prove it wrong.
| Scenario | Prob. | Cross-asset effect | Lead indicator → falsifier |
|---|---|---|---|
| (a) Glacial drift — BASE | ~50% | USTs: term premium grinds higher on fiscal, auctions clear. Gold: structural bid (~600–900t/yr). Dollar: range-bound (~95–105). | Lead: COFER USD falls ~0.5–1pt/yr; constant-FX share holds ~55–58%. Falsifier: constant-FX COFER rises above 58–60%, or RMB SWIFT tops 6%. |
| (b) Sanctions bloc-split | ~17% | USTs: bloc exits, allied demand holds. Gold: bloc accumulation accelerates (BRICS+ above ~7,000t [DIR]). Dollar: firm in the core. | Lead: mega-sanctions on a tier-1 Chinese bank; mBridge/Unit scale 5–10×. Falsifier: the Unit stalls; the Gulf reaffirms its peg; mBridge stays sub-US$100bn. |
| (c) US fiscal scare — flight FROM | ~12% | USTs: disorderly term-premium spike, 10y above ~5.5%. Gold: sharp spike. Dollar: falls hard (below ~90). | Lead: failed auctions; a Fed-independence breach; deficit above ~8% GDP. Falsifier: orderly consolidation; risk-off that rallies USTs and the dollar together. |
| (d) Stablecoin re-dollarization | ~10% | USTs: higher front-end T-bill demand; long-end pressure persists. Gold: neutral-to-soft. Dollar: firm/stronger. | Lead: stablecoin cap above ~US$1tn (TBAC sees ~US$2tn by 2028 [DIR]); ~99% dollar-link holds. Falsifier: regulation crushes growth; a major stablecoin run forces a T-bill fire sale. |
| (e) Dollar reasserts (non-US shock) | ~8% | USTs: flight to, yields fall. Gold: mixed. Dollar: spikes (above ~108–115). | Lead: global recession or EM crisis; Fed swap-line usage; broad-dollar funding stress. Falsifier: US-specific stress that breaks the safe-haven reflex — USD and USTs sell off together. |
| (f) Rapid commodity-rail shift | ~3% | USTs: higher risk premium if exporters recycle less. Gold: very strong if settlement links to hard assets. Dollar: loses commodity corridors. | Lead: oil exporters accept yuan/e-CNY at scale; mBridge commodity volume rises multiples. Falsifier: oil-invoicing data stay dollar-dominant. |
Collapse the six into the three families they form, and the structure becomes legible: one world where nothing breaks, one where the dollar's own house cracks, and one where the dollar wins — and the last two are driven by the very same stress.
Glacial drift dominates: the official share grinds lower while FX, payments, banking and safe-asset depth stay dollar-anchored, gold is bid structurally, and stablecoins quietly widen private dollar reach. The most likely future is the one already underway — erosion at the edge, not displacement at the core.
Combine the sanctions bloc-split (~17%), a US fiscal scare (~12%) and the rapid commodity-rail shift (~3%) and you have the discontinuity cluster — the worlds where the slow drift turns nonlinear. None is the base case; together they carry about a third of the weight. This is where the real risk lives, and it is fiscal and geopolitical, not monetary.
The mirror cluster the bears forget: stablecoin re-dollarization (~10%) and a non-US risk-off shock (~8%) both strengthen the dollar. Private digital demand and a global scramble for liquidity each deepen, not erode, the dollar's footprint. A complete read of the tails holds both sides.
This is the most important thing to hold about the tails. A US fiscal scare and a non-US risk-off shock are both fiscal-stress events — but one sends capital fleeing from the dollar and the other to it, and which fires depends entirely on whether the stress is US-specific or globally diffuse. The same emotion — fear — is the dollar's deadliest enemy when the crisis has a US return address and its most loyal friend when the crisis is everyone's. De-dollarization can never be modelled as a one-way slope.
A ~50% base case is the modal future, not the consequential one. Glacial drift is the world that changes nothing you need to act on — the share grinds, auctions clear, the dollar stays firm. Everything that would force a reserve manager or a portfolio to move sits in the tails. Planning to the base case is planning for the year nothing happens; the discipline is to weight the modal outcome honestly and still budget for the tails.
The verdict holds in one breath: the dollar's reserve crown is slipping, its working grip is not — a strong core with a fraying edge, durable in the base case and decisive only in the tails.
Sources: Scenario probabilities are council-blended and subjective/illustrative [DIR] — relative weighting, not precision (base-case estimates ranged 45–65%; the adopted ~50% sits at the center). Four-constraint framework: Federal Reserve 2025 dollar report; US Treasury TBAC 2025. Petroyuan/Hormuz: Foreign Affairs Forum and MEXC News, Mar 2026 (no confirmed directive); ECB/IMF invoicing, Jun 2025 (no robust evidence of oil de-dollarization); Iran-China oil ~US$8bn/yr [DIR]. mBridge ~US$55.49bn and ~US$9.6tn/day FX: Atlantic Council; BIS Triennial 2025. Stablecoin ~US$2tn-by-2028 is a TBAC projection [DIR]. Cross-asset effects, lead indicators and falsifiers are the council's diagnostic triggers, not forecasts.
What we can show live — and what we can't, yet
Feed-state is the status of a tile's number, never of a section. Every section ships its full frame; only the live figure waits on a wired feed. We say which is which.
No invented data. Directional figures from non-audited or single-source claims (Russia-China settlement, BRICS+ gold, Iran oil) are labelled [DIR] wherever they appear.