Tokenizing a bond is straightforward. Settling it with money that actually behaves like money on-chain is not. The asset leg of tokenization has been solved in pilot after pilot. The cash leg still routes through the same real-time gross settlement systems built for a pre-tokenized world, and that gap is where the current wave of real-world asset projects breaks down.
The cash leg is the constraint, not the asset
The bottleneck in tokenized finance is not issuing the asset. It is settling the trade with on-chain money that carries the legal weight of a bank liability or central bank reserve. Without that, every tokenized transaction resolves through the same RTGS infrastructure that existed before tokenization. The asset leg has been solved. The cash leg has not.
The BIS has built an entire flagship project, Project Agorá, around this exact observation: seven central banks and over 40 private financial institutions are collaborating specifically on the settlement asset, not the securities. Most coverage of real-world asset tokenization treats the cash leg as an implementation detail. It is the entire problem. Tokenizing a treasury, a bond, or a fund share is a solved engineering exercise. Agreeing on what money looks like when it reaches the other side of the transaction is not. The BIS treats central bank money as the ultimate safe medium for settling transactions because it eliminates counterparty credit risk. Without a tokenized version of that settlement asset, every on-chain bond trade depends on an off-chain settlement instruction.
This is why tokenization pilots proliferate while tokenized markets do not. The pilot proves the asset can be represented. The production deployment requires a cash leg that does not yet exist at institutional scale.
What most "tokenized" assets actually are
Most tokenized RWAs in 2025 and 2026 are tokenized claims against off-chain settlement, not tokenized assets in any useful sense. The token sits on a blockchain. The settlement, custody, and legal finality sit in the same CSDs, RTGS systems, and custodian ledgers they always did. The token is a pointer. The asset has not moved.
The BIS makes this point directly in its 2025 Annual Economic Report: "Leveraging the full benefits of tokenisation requires including 'native' assets that are both issued and reside on the platform." Over $4 billion in tokenized sovereign, supranational, and agency bonds has been issued across roughly 20 transactions, and the BIS characterizes this volume as an "experimental phase." The reason is straightforward. These bonds are tokenized wrappers around positions whose legal ownership still clears through conventional infrastructure.
Fnality frames the same problem in operational terms. In its March 2026 note on governance, the firm observes: "If an asset is locked on one chain and reissued on another, a synthetic exposure is created. The legal and operational assumptions behind that exposure may be thinner than they appear." That is the condition of most tokenized RWA supply today. A token on a public chain referencing an asset whose true settlement happens somewhere else, under different rules, on a different timetable.
The uncomfortable implication is that a large share of what the market calls RWA tokenization does not tokenize the RWA. It tokenizes a claim against someone who holds the RWA through the existing plumbing. The token is faster. The settlement underneath is not.
The live evidence: even the best tokenized deposit systems route cash through RTGS
The most advanced live tokenized deposit environment in the world still routes interbank settlement through legacy RTGS infrastructure. Even Hong Kong's EnsembleTX, which moved from sandbox to live transactions in November 2025, routes interbank cash through the HKD RTGS system. The asset leg is programmable. The cash leg is not.
Hong Kong's EnsembleTX launched in November 2025 as the HKMA's production upgrade from its Project Ensemble sandbox. Eddie Yue described it as "a pivotal moment in our journey, upgrading from proof-of-concept to a real-value setting." The cash leg, however, was explicit in the launch release: "Interbank settlement of tokenised deposit transactions will initially be facilitated via the HKD Real Time Gross Settlement (RTGS) system."
That architecture is not a weakness of EnsembleTX. It is the honest state of the art. HSBC completed its first live cross-bank tokenized deposit transaction on EnsembleTX for client Ant International the same month, and the cash moved between banks through the existing RTGS rail. The asset leg ran on a programmable ledger. The cash leg did not.
The same pattern shows up in JPMorgan's May 2025 cross-chain delivery-versus-payment test with Ondo and Chainlink, which settled Ondo's tokenized U.S. Treasuries against "USD deposits at JPMorgan" via the Kinexys platform. Kinexys has processed $1.5 trillion notional to date and averages $2 billion daily. The cash leg for the most sophisticated tokenized DvP of 2025 was a tokenized bank deposit inside a walled garden, not a stablecoin and not a wire.
Across both examples, the asset leg is live and the cash leg is partial. Either it runs on legacy RTGS, or it runs inside a single bank's internal ledger. Neither is a market-wide cash leg.
Why stablecoins are not the answer
Stablecoins look like the cash leg. They are not, and the central banks have been explicit about why. Three properties define sound settlement money: universal acceptance at par, the ability to discharge obligations on demand, and institutional-grade safeguards against financial crime. Stablecoins fail all three at wholesale scale.
The BIS's 2025 Annual Economic Report identifies those three tests for money as singleness, elasticity, and integrity, and concludes stablecoins do not deliver on any of them at the scale required for institutional settlement.
The IMF's April 2026 note on tokenized finance reaches the same conclusion through a different frame. Tobias Adrian and co-authors describe stablecoins as resembling "money market funds more than actual money" and flag the risk of "confidence-driven runs as tokenized finance scales." The IMF's position is that "the settlement asset is the cornerstone of any financial system," and that the anchor must be central bank money, not a private claim on a corporate reserve.
There is a subtler balance-sheet problem. NY Fed staff report SR 1179 frames the choice between stablecoins and tokenized bank deposits as a question about whether society wants money and lending fused together or pulled apart. Stablecoins make payments safer but pull deposits out of banks, reducing the lending capacity that commercial banks provide to the real economy. Tokenized deposits keep the deposit-and-lending model intact, and the authors conclude that which instrument dominates will be settled primarily by regulatory design rather than by technology or consumer preference.
Stablecoins can move value. They cannot carry the legal, monetary, and credit-intermediation functions institutions require for wholesale settlement.
The counter-risk: machine-speed settlement without an anchor
There is a counter-argument that deserves direct engagement. Programmable settlement at machine speed, without a proper anchor, can amplify crises rather than contain them. The IMF raises this directly: atomic settlement removes the end-of-day windows that give regulators time to intervene. The argument for tokenized cash is stronger for acknowledging this tension honestly.
Tobias Adrian and his co-authors warn that tokenization moves settlement to machine speed, outpacing the regulatory tools built for batch processing and end-of-day cycles. Traditional systems rely on settlement windows that give regulators time to intervene before a problem spreads. Atomic on-chain settlement removes those windows entirely.
This is the honest tension in the argument for tokenized cash. A programmable cash leg is both the precondition for tokenized finance to scale and a new vector for systemic fragility if built on the wrong base. Fnality's safe settlement anchor note makes the same point from the private-sector side: "When settlement happens on different rails with different standards, legal assumptions and credit exposures, speed can become a source of fragility rather than resilience." Speed without an anchor is not progress. It is a run waiting to happen.
The answer the IMF, BIS, and Fnality all converge on is the same. Anchor tokenized settlement in central bank money, or in bank liabilities that inherit the legal and regulatory treatment of central bank-anchored deposits. A tokenized cash leg built on a protocol-enforced claim against a regulated bank balance sheet is not the same instrument as a stablecoin, and the settlement properties that matter at crisis moments diverge.
Tokenized deposits as the specific cash-leg primitive
Tokenized deposits are the specific cash-leg primitive that clears the institutional bar. They inherit the legal status of bank deposits, carry deposit insurance, and settle with the same finality as a claim against the banking system. The EBA confirmed this in December 2024. No new legal infrastructure is required in Europe.
The EBA's December 2024 report on tokenised deposits settled the legal question: "The tokenisation of a deposit, in the narrow sense of recording the deposit claim of a depositor against the credit institution on the distributed ledger technology (DLT) instead of a traditional ledger, does not per se alter the fundamental nature of the claim and thus its regulatory qualification as a deposit." A tokenized deposit is a deposit. It carries deposit insurance and inherits the existing legal and regulatory treatment.
The American Bankers Association stated the industry position in March 2026: "As digital asset activity expands, particularly in regulated, non-crypto, blockchain-based markets, the need for tokenized money to serve as the cash leg in digital asset trading is becoming both clear and immediate." The same note identifies that tokenized deposits give banks "the benefits of programmability without the limitations of payment stablecoins."
Building tokenized deposits well is a protocol problem, not a trust problem. Every unit of tokenized deposit supply must correspond to a locked bank liability, and the equivalence must be enforced at the protocol level rather than attested through quarterly reserve reports. FractiFi's tokenized cash rails enforce that 1:1 reserve invariant at the protocol level rather than through quarterly attestations, which is the same architectural point we made in why you cannot build programmable finance on non-programmable money. The money layer sets the ceiling for everything built on top.
The settlement gap is now a named regulatory concern
The cash-leg gap is no longer an industry observation. It is a named finding in official reports. IOSCO's November 2025 final report on tokenisation identifies "lack of high-quality settlement assets" as one of two structural challenges preventing tokenization from scaling. The report flags settlement finality uncertainty as a separate concern, and notes that most participants continue to favor traditional systems because of operational familiarity and pre-funding requirements, not because those systems are superior.
This matters because IOSCO is the global coordinator for securities regulators. When IOSCO names a problem, the constituent regulators inherit the agenda. The cash leg has moved from conference-panel conversation to formal regulatory workstream within a year.
Market infrastructure investors have priced the gap directly. Fnality raised $136 million in a Series C led by WisdomTree, Bank of America, Citi, KBC Group, Temasek, and Tradeweb in September 2025, specifically to build settlement rails that hold funds "in a central bank account, providing 1-to-1 funds with the credit risk characteristics of central bank money." Banks do not co-fund infrastructure they already have. They fund infrastructure the market is missing.
What this means for institutions building tokenized products
Treat the cash leg as the architecture decision, not the last mile. The asset side of a tokenized product is the simpler half of the problem. The side that determines whether the product is a production system or a marketing exercise is the settlement asset and the rails it runs on. Institutions evaluating tokenization projects should interrogate the cash leg first: Is the cash a bank liability or a corporate claim? Does it inherit deposit insurance? Does settlement actually happen on-chain, or does the on-chain record trigger an off-chain wire?
Those questions separate tokenized markets from tokenized wrappers. The answers also determine which regulatory regime applies, which balance sheet carries credit risk, and whether the product can integrate with existing treasury mandates without bespoke legal analysis. An asset manager using stablecoins for settlement has a different counterparty profile than one using tokenized deposits, and the treasury committee will notice.
The next two years will expose which tokenization projects were actually tokenized. The ones with a proper cash leg will scale. The ones that pushed the cash leg onto legacy RTGS or a stablecoin issuer will hit a ceiling, either at the regulator, at the counterparty's risk committee, or at the first stress event that tests whether settlement was truly final.
Get in touch
The cash-leg gap sits between the banks building closed systems for themselves and the stablecoin issuers building for crypto markets. Mid-tier banks, asset managers, private equity firms, and family offices moving from tokenization pilots to production need tokenized cash rails that inherit the legal and regulatory treatment of bank deposits and settle at the protocol level. FractiFi builds that base layer: a protocol-enforced 1:1 reserve invariant, compliance embedded at the settlement layer, and integration paths for banks at three different points of digital readiness. If you are evaluating the cash leg for a tokenized product, we would like to talk.
