[Suit Up]

HOME / PLAYBOOKS / Asia is winning the wholesale race; the US is winning the retail race
Playbook · Decision support

Asia is winning the wholesale race; the US is winning the retail race

The thesis

Most tokenisation analysis conflates two structurally different markets. Our read: Asia is winning the wholesale race, anchored by regulated bank rails, central-bank-supervised settlement layers, and GSIB-led pilots that have moved (or are visibly moving) from sandbox to production. The US is winning the retail race, anchored by fintech-issued payment stablecoins at scale, retail-distributable tokenised products under named SEC wrappers, and post-SAB-122 custody normalisation. They look like one story. They are tracked as one story. They produce bad strategy when conflated.

What conventional analysis gets wrong

Read almost any 2026 tokenisation deck and you will see the same numbers stacked side by side: Circle's USDC float, BlackRock's BUIDL AUM, JPMorgan's Kinexys cumulative notional, the total addressable market for tokenised real-world assets, the HKMA EnsembleTX phase launch, the Project Guardian participant roster. These get presented as if they were comparable answers to a single question, then averaged into a "tokenisation is X bn and growing" headline.

They are not comparable. They are answers to different questions in different markets with different success criteria, asked of different counterparties, supervised by different regulators, and measured against different KPIs. Treating them as one number produces strategy decks that recommend the wrong jurisdiction for the wrong product to the wrong customer base. We have watched too many of those briefings to keep quiet about it.

The argument

Different regulatory architectures

Asia and the US have built structurally different regulatory architectures for digital-asset finance, and the architectures shape what gets shipped first.

Asian jurisdictions converge on the integrated-supervisor sandbox pattern. Singapore has MAS sitting as central bank, banking supervisor, securities regulator, and payments licensor in one entity, running Project Guardian industry pilots and Project Orchid purpose-bound money work without inter-agency negotiation. Hong Kong runs a slightly tighter HKMA / SFC split, with EnsembleTX as the wholesale tokenisation rail and the Stablecoins Ordinance as the licensing perimeter. Japan's FSA runs the Payment Innovation Project sandbox that produced the April 2026 Nomura, Mizuho, and JSCC tokenised JGB collateral trial on Canton. The pattern is the same in shape: one supervisor (or two coordinating ones), one named programme per jurisdiction, named bank participants moving real money inside the perimeter.

The US has a structurally different topology. The federal stablecoin perimeter sits with GENIUS under OCC, Federal Reserve, and FDIC joint rule-making. Market structure splits between SEC and CFTC under CLARITY. State-level layers (NYDFS limited-purpose trust charters, Wyoming SPDIs) overlay the federal perimeter, with crypto-specific federal carve-outs (the OCC trust bank charter route) added on top. There is no equivalent of EnsembleTX, no Federal-Reserve-coordinated wholesale tokenisation pilot with named GSIBs settling tokenised MMFs against tokenised deposits inside a regulator-operated layer.

The two architectures produce different behaviours. Where the Asian regulator can co-design a wholesale pilot with the named bank participants and observe the operational shape directly, the US infrastructure is built bottom-up by individual institutions on their own ledgers without a coordinated wholesale layer above them. Wholesale fits the integrated-supervisor pattern; retail fits the multi-agency-with-state-overlay pattern.

Different counterparty profiles

Look at who actually shows up on the named press releases.

Asian wholesale programmes are GSIB-led and central-bank-coordinated. EnsembleTX names Standard Chartered, HSBC, and Bank of China as the issuance counterparties, with BlackRock and Franklin Templeton on the asset-manager side. Project Guardian's roster recurs across DBS, UOB, OCBC, Standard Chartered, HSBC, and the Japanese megabanks (MUFG, SMBC, Mizuho, Nomura). The asset-manager and fintech layer follows the bank-and-regulator layer.

US retail leadership runs the other way. Fintechs (Circle, Securitize, Ondo Finance) and asset managers (BlackRock, Franklin Templeton, Apollo) lead. The bank layer is reactive to the fintech-and-asset-manager layer. JPMorgan's Kinexys is the exception that proves the rule: it is the deepest single-bank deposit-token rail in the world by notional, and it is one bank on a permissioned ledger rather than a regulator-coordinated multi-bank layer in the EnsembleTX shape.

This counterparty-profile gap is not a coincidence and it is not transient. It reflects which institutions have the policy oxygen, the balance sheet, and the regulatory cover to lead in each jurisdiction.

Different success KPIs

Each market measures itself by the metrics that fit its leadership profile, and those metrics do not translate across the Pacific.

Asia tracks settlement volume on regulated rails. EnsembleTX progress is measured in real-value transactions executed inside the HKMA-coordinated wholesale layer, with tokenised MMFs settling against tokenised deposits and the cash leg routed through HKD RTGS pending tokenised CeBM upgrade. DBS Token Services reports intra-bank tokenised flows. Partior reports interbank settlements across its DBS / JPMorgan / Standard Chartered consortium. The Project Guardian Fixed Income workstream's ICMA addendum is measured by whether the operational guidance gets adopted by other tokenised-bond programmes. Settlement throughput on regulated rails is the leading indicator.

US retail tracks adoption and AUM. USDC market cap and USDT circulation are the headline metrics for stablecoin progress. BUIDL distribution-chain count (now nine ledgers) is the marker for tokenised-MMF reach. Retail-eligible product launches are measured by AUM gathered and platforms that distribute them. Product reach across distribution channels is the leading indicator.

Both sets of metrics are real. They measure different things. Pulling a wholesale settlement-rail KPI into a retail-distribution dashboard, or vice versa, produces nonsense. An EnsembleTX participant that benchmarks against Circle's float is going to conclude EnsembleTX is failing. A US fintech that benchmarks against EnsembleTX settlement volume is going to conclude the US is losing. Neither conclusion is right.

Different exit positions and brand stakes

The fourth observation is the one with the longest implications, and our read here is the most contested. Who wins what tells you who can extend in adjacent layers.

Asian wholesale leadership extends naturally into wholesale-CBDC plumbing and cross-border settlement. mBridge under post-graduation governance, Project Agorá, and Project Ensemble cross-border are the obvious adjacencies. The named bank counterparty set already in EnsembleTX is the same set that would settle a cross-border tokenised flow between HKD and SGD or HKD and JPY. The integrated-supervisor pattern that works domestically extends to bilateral regulator coordination because the supervisors already know how to talk to each other.

US retail leadership extends naturally into agentic commerce. USDC plus the x402 protocol, plus Visa Intelligent Commerce, plus Mastercard Agent Pay live in the US-retail ecosystem first because the retail-stablecoin layer is where AI agents can credibly hold and move value at small ticket sizes. The wholesale GSIB rails are not where the first agentic-commerce flows will route, because GSIB tokenised-deposit issuance assumes KYC-bound institutional holders, not delegated AI authority over a bearer-style instrument.

Read this charitably: the question is not who wins everything; it is which adjacencies each leadership position is best positioned to extend into. Asia is positioned for wholesale-CBDC and cross-border-settlement adjacencies. The US is positioned for agentic-commerce and retail-payment adjacencies. The two adjacency stacks are different categories of bet.

Implications for practitioners

For a strategy lead at an APAC GSIB: do not import US retail-tokenised-product KPIs as success metrics for your wholesale programme. If your CIO benchmarks your EnsembleTX or DBS-Kinexys-style tokenised-deposit programme against Circle's USDC float or against BUIDL's nine-chain distribution count, you will measure the wrong things, miss the actual wins (regulator-recognised real-value transactions, bilateral interbank settlement throughput, named CCP-grade integration), and conclude you are losing a race you are not actually running. Ask instead what the equivalent Singapore / Hong Kong / Tokyo programme has shipped this quarter and benchmark there.

For a US fintech building a tokenised product: do not anchor your launch on Asian wholesale partnerships. The customer-acquisition curve is structurally different. An EnsembleTX-style integration takes 12-18 months of regulator and bank-counterparty engagement, and the eventual customer is an institutional treasurer or asset manager rather than a retail wallet. If your product is fintech-shaped (retail wallet distribution, fast iteration, low-ticket payment flow), launching against the US retail stack and routing into Asia later through bank distribution partners is operationally cleaner than building Asia-first. The cross-jurisdictional rollout playbook sequencing supports this for tokenised funds; it applies equally to payment stablecoins.

For an allocator tracking where institutional capital is actually going: read each market's metrics in its own context. The April 2025 BUIDL nine-chain expansion is a US retail-leadership signal that says distribution reach is widening across the MMF asset class. The November 2025 EnsembleTX phase launch is an Asian wholesale-leadership signal that says regulated bank rails are settling real-value transactions. The two signals are not in competition with each other. Both can be true simultaneously, both can compound year-on-year, and an allocator who treats them as substitutes will misprice both.

For an editorial reader: when you see "tokenisation is X bn" headlines, ask which market is being measured. Stablecoin float plus tokenised-MMF AUM plus deposit-token notional plus bond-issuance volume is a meaningless aggregate, because the four measure different products in different markets with different success criteria. The numbers that matter are the ones inside each market's own framing.

Where this could be wrong

The thesis is directional, not deterministic. Two scenarios would force a meaningful update.

The first is a US wholesale programme of EnsembleTX scale shipping and getting traction. Fnality is the closest live candidate today: a regulated wholesale tokenised-cash settlement layer with named bank participants, structurally similar to what EnsembleTX is building. It has shipped slower than EnsembleTX and the US deployment has been delayed relative to the GBP launch, but it is real and it has named GSIB participation. If a Fnality-USD layer goes live in 2026-2027 with multi-bank settlement throughput comparable to EnsembleTX in HKD, the US wholesale gap closes meaningfully and the thesis weakens on the wholesale side.

The second is an Asian retail tokenised product hitting BUIDL-scale AUM. This looks less plausible in 2026 but is possible by 2028 if Hong Kong's April 2026 SFC tokenised-authorised-funds secondary-trading framework pulls in retail-distributable tokenised MMF AUM at scale, or if Singapore opens a retail-eligibility pathway under Project Guardian. Neither is operating yet at retail scale. We would update directionally if either ships in the next 24-36 months.

Related

Weekly briefing

Sunday evening Singapore time. Importance-3 items, one deep dive, what's worth watching next.