Suit Up · Asia's institutional desk on tokenisation

The thesis

The institutional tokenisation conversation has been dominated since 2023 by settlement-efficiency narratives: atomic delivery-versus-payment (DvP), T+0 cycles, reduced counterparty risk in clearing, the disappearance of reconciliation breaks. Those wins are real, and the desks that built them have done good work. But they are a back-office optimisation story dressed as a market-structure story, and the press releases have flattered the impact. The actual capital-cost unlock, the move from "tokenisation as faster ops" to "tokenisation as balance-sheet expansion", arrives when tokenised assets start being recognised as collateral in repo, prime brokerage, margin, and central-counterparty (CCP) arrangements. The next 18 months will produce the first inning of that story. Anyone tracking the settlement metrics in isolation is looking at the wrong scoreboard.

The setup

The settlement-efficiency narrative was the easy story to tell, which is why it shipped first. Anyone who has run bank operations recognises the pain: T+2 settlement cycles that lock up funding, reconciliation breaks that consume back-office headcount, correspondent-banking chains that turn a notionally instantaneous payment into a multi-day exercise in operational risk. Tokenised settlement work attacks all of that legibly. Project Ensemble in Hong Kong, Project Guardian in Singapore, Partior across the DBS-JPM-SCB corridor, Kinexys inside JPMorgan, Fnality in the wholesale-CeBM space: all of them produce clean before-and-after metrics, all of them are championed by operations leaders inside the participating institutions, all of them are championed by regulators because they reduce systemic counterparty risk. The story practically writes itself.

The problem is that operational efficiency, on its own, is bounded. Cost of operations affects how cheaply you can run your business. Cost of capital affects how much business you can do. The first matters; the second matters more. Settlement efficiency does not change a bank's risk-weighted asset density, does not change a fund manager's leverage capacity, does not change the cost of repo funding. It is a feature improvement at the margin, not a category change in what an asset can do. Tokenisation as collateral, by contrast, is a category change. When a tokenised asset becomes recognised collateral inside a CCP, prime broker, or central-bank facility, the asset gains a new economic role and the holder gains new optionality. That is a structural shift, not a process improvement.

The argument

Three pieces, in order.

Why settlement was the easy story

The settlement narrative produces clean before-and-after numbers, and it ships fast because the binding constraint is engineering rather than rule-making. T+2 to T+0. Reconciliation breaks down by 90 percent. Intraday liquidity demand reduced by some published percentage. The numbers fit a slide. Bank-operations leaders championed it because it was self-evidently good and required no prudential rewrite. Regulators championed it because it reduces systemic counterparty risk in clearing without forcing them to recalibrate capital. The story shipped on the operational side because it could ship without anyone in the prudential function having to make a hard call.

But the cost-of-operations win has a ceiling. Even taken at face value (and the headline numbers are usually the best-case framing), settlement efficiency does not change the prudential treatment of the underlying asset, does not change what the bank can hold against the position, does not change what the asset class can fund. It is plumbing. Important plumbing, but plumbing. A bank that has gone from T+2 to T+0 on a particular asset class still owns the same balance sheet at the end of the day. The CIO at the asset-management counterparty still sees the same fund-level capacity. The prime-brokerage desk still extends the same leverage. None of those numbers move because the operational layer got cleaner.

Why collateral is the bigger, harder story

When a tokenised asset becomes recognised collateral in a CCP, repo desk, prime-brokerage agreement, or central-bank standing facility, two things change at once. The asset gains a new use, because it can now fund leverage. The holder gains new optionality, because the position can be pledged for liquidity without sale. Both changes are structural rather than incremental. The asset's economic role is different on the day after the eligibility decision than it was the day before, in a way that the operational story cannot match.

The early moves are visible and worth tracking carefully. The Centrifuge V3 + Aave Horizon architecture, with Janus Henderson's JAAA collateral-loan-obligation product crossing USD 1bn in tokenised assets under management (AUM) and the Sky / Grove allocator deploying USD 250m of JAAA on Avalanche, is the worked DeFi-native example. BUIDL's acceptance on Aave Horizon and as institutional collateral on Binance from November 2025 is the cleanest cross-perimeter example, carrying the regulated-MMF structure into venues that previously had no native bridge to one. The HKMA EnsembleTX repo workstream and the wholesale Orion + DIGIT cash-leg architecture both nest tokenised collateral inside the wholesale-bank stack rather than the DeFi stack. Each of those moves is operationally narrow today. Each is structurally precedential. The moment a major CCP accepts the first tokenised asset as eligible collateral with a published haircut schedule, the curve tilts because the calibration question becomes a public artefact rather than a bilateral negotiation, and competing venues have to respond.

The four collateral milestones to track

Each of these milestones is a specific named event whose first occurrence resets what is possible. None of them have happened at the time of writing. The candidate venues are listed because the universe of plausible firsts is small in each case; they are not predictions of which institution will move first.

First, a major CCP accepting a tokenised asset as eligible collateral with a published haircut schedule. The candidate venues are LCH (the dominant European clearing operator), ICE Clear, JSCC (the Japanese securities clearing corporation), and the HKEX clearing arms. Bilateral repo and prime-brokerage acceptance is one thing; CCP eligibility with a transparent schedule is qualitatively different because it sets the haircut benchmark that the rest of the institutional stack inherits. The first move here is the largest single inflection in this list.

Second, a major prime broker programme for tokenised collateral with a standardised haircut framework across asset classes. The candidate sponsors are the JPMorgan, Goldman Sachs, and Morgan Stanley prime-brokerage desks. Bilateral acceptance of BUIDL as repo collateral by individual prime brokers is already happening under negotiated terms; the inflection comes when one of the bulge-bracket desks publishes a tokenised-collateral schedule across multiple asset classes (tokenised MMFs, tokenised sovereign bonds, tokenised credit) so the asset-management buyer base can price acceptance into product design rather than negotiating it bilaterally each time.

Third, a cross-border tokenised-collateral arrangement at material scale. The structurally interesting version is a tokenised US Treasury MMF accepted as collateral against an HKD-denominated or SGD-denominated repo, or an equivalent non-USD funding leg against a USD-asset collateral position. This is the version that breaks the parallel-rails pattern that characterises the current institutional tokenisation map (BUIDL on USD rails, Orion on GBP and HKD rails, Kinexys on USD-anchored rails, the Japanese consortium platforms on JPY rails) and tests whether the cross-currency collateral-acceptance plumbing can move at speed once one bilateral arrangement opens the path.

Fourth, a central-bank standing facility accepting tokenised collateral for intraday or end-of-day liquidity. The candidate central banks are HKMA, MAS, the Bank of England, and the Bank of Japan. This would be the most consequential single milestone on the list, because central-bank collateral acceptance is the strongest possible signal of prudential status: if a tokenised asset is good collateral against monetary-authority liquidity, every downstream venue has structural cover to accept it on similar terms. None of the named central banks has signalled imminent action on this, but the EnsembleTX progressive trajectory toward tokenised central-bank money (CeBM) is the closest jurisdictional signal that the architectural prerequisites are being put in place.

Implications for practitioners

Four reader-side reads.

For asset managers, design tokenised products with collateral acceptance as the north star, not settlement efficiency. The secondary-market depth and price-discovery infrastructure that collateral acceptance requires is not optional; a tokenised wrapper that distributes cleanly but cannot be liquidated under stress will not clear the lender-side evaluation in the collateral-evaluation playbook. The asset-class hierarchy from that playbook (tokenised MMFs strongest, tokenised sovereign bonds next, tokenised gold mature in narrow use, tokenised private credit and equities weak) sets the credible target window for the next 18 months.

For bank treasuries, stop benchmarking the tokenisation programme against settlement-efficiency KPIs. Benchmark it against capital-efficiency KPIs: risk-weighted asset density, repo financing cost, intraday liquidity demand, prime-brokerage funding spreads. The Basel SCO60 Group 1a route (see the capital-treatment playbook) is the gating prudential filter, and the targeted review under SCO60 is the principal Basel-side process to track for any horizon longer than a year. A tokenisation programme that lands a Group 1a classification on the bank's own holdings is the entry ticket; collateral eligibility on top of that is what changes the cost-of-capital arithmetic.

For prime-brokerage and CCP risk leadership, the calibration window is opening, and the first-mover programmes will set the haircut curve that competitors inherit. There is a structural premium to being the venue that publishes the first transparent tokenised-collateral schedule across multiple asset classes, because the asset-manager buyer base will route flow toward the venue with the cleanest published terms. The reverse is also true: a venue that lags will inherit the schedule that the first mover sets, with limited room to renegotiate haircuts down once the precedent is established.

For allocators, tokenised products that achieve major-CCP collateral eligibility will trade at a structural premium to those that do not, because the optionality of pledging without sale is a real funding-cost benefit that the secondary-market price will eventually reflect. Pricing that optionality into the allocation decision now, rather than after the eligibility decisions are public, is where the asymmetric reward sits.

Where this could be wrong

Two scenarios where the thesis breaks. First, the regulatory perimeter for tokenised collateral could turn out to be more conservative than we expect. The Basel SCO60 targeted review endorsed by GHOS in March 2026 could narrow the Group 1a route, tighten the technology-infrastructure conditions, or restrict the public-permissionless-ledger path in ways that slow CCP and prime-broker acceptance for a meaningful subset of asset classes. The substantive direction of the review is not yet public; our read assumes the calibration loosens or holds at current levels rather than tightening, and the tightening case would push the milestones into 2027-2028.

Second, the operational-integration cost for ingesting on-chain price-and-position data into existing risk systems could prove harder than the early pilots suggest. Most lender-side risk systems were not built for on-chain feeds, and the build cost per asset class is meaningful even where the prudential and legal pieces clear. If the integration cost compounds across multiple asset classes, the first published CCP haircut schedule could lag the legal and prudential clearance by 12-18 months, again pushing the milestones into 2027-2028 rather than 2026-2027. Our read is that the milestones land within an 18-24 month window from now, though the longer tail is plausible.

Neither of these scenarios changes the directional argument that collateral, not settlement, is the next major unlock. They affect the timing and the velocity, not the structure. The settlement-efficiency narrative will continue to ship cleanly through both scenarios; the collateral-acceptance story will ship more slowly under either of them but is the structurally larger prize regardless of pace.

[Suit Up]

The next 18 months are about collateral, not settlement