A workable tokenization strategy starts with one balance-sheet problem, one settlement asset and one accountable owner, and reaches production inside licences the institution already holds. The ground is firming: the Basel Committee set 1 January 2026 as the implementation date for its revised cryptoasset standard, and in May 2026 seven central banks and more than 40 private institutions reported results from the Project Agorá prototype. What still fails most often is execution.
Who decides this
The board approves the mandate and the risk appetite; the head of digital assets owns delivery. The board is accountable for three things: which business line the programme serves, how much capital and operational risk it may carry, and the date by which a pilot reaches production or stops. The head of digital assets is accountable for everything below that line: use case, rails, custody, vendors and the regulatory file.
If one committee holds both roles, nobody is accountable for saying no. Keep them separate, and give the head of digital assets authority over the budget lines that the plan depends on.
Which problem should tokenization solve first?
Start where the institution already holds the asset, the client and the licence; for most banks and asset managers that means fund distribution, collateral mobility or settlement, not a new asset class. Tokenizing something you do not already custody adds a legal question to every technical one. In our experience, enterprise tokenization programmes that begin with a new asset class tend to spend their first year on structuring, not on clients.
The Bank for International Settlements framed the test well in its 2025 Annual Economic Report (opens in new tab): the gain from tokenization is that messaging, reconciliation and settlement can collapse into one process. Apply that test to every candidate. If a use case removes no reconciliation step you pay for today, treat it as a communications project and fund it accordingly.
The four starting points most institutions weigh differ on what goes on-chain, which licence question opens, and what the programme depends on.
| Starting point | What the institution puts on-chain | Licence question it opens | Hardest dependency |
|---|---|---|---|
| Tokenized deposits | Its own deposit liability | Usually within the existing banking licence; confirm with the supervisor | Counterparties on the same ledger; interbank models are still at prototype stage |
| Payment stablecoin | A new liability, often in a new entity | A dedicated issuer regime in the US, EU and Hong Kong; in Singapore, a 2023 framework with draft legislation pending | Distribution and redemption liquidity |
| Tokenized security entitlements | A record of holdings, not the security itself | Depends on the market infrastructure provider’s permissions | The scope of that provider’s pilot |
| Tokenized fund units or other real-world assets | A distribution wrapper around an existing product | Securities and fund rules in each market where units are sold | Key custody and transfer controls on each holder |
Our view: pick one row, not two. RWA tokenization of a fund you already run is the shortest path to a production result for an asset manager. For a bank, tokenized deposits or collateral are usually the better first move, because the liability is already yours.
What will the tokens settle against?
Decide the cash leg before the asset leg; a tokenized asset that settles against a conventional payment has kept the old settlement risk. The choice is between tokenized deposits, a regulated stablecoin, or conventional cash with a delay.
The evidence on tokenized deposits is growing but not finished. Project Agorá tested whether wholesale cross-border payments could settle atomically, with tokenized central bank reserves on one side and tokenized commercial bank deposits on the other, on a two-layer design that leaves each central bank with autonomy over its currency. Its published next steps include real-value transactions in certain currencies with certain participants. That means production use is not yet available to most institutions.
Stablecoin rules are close to live in the largest market. The US GENIUS Act was signed on 18 July 2025, and the OCC issued its final implementing rule (opens in new tab) on 21 October 2026, effective 18 January 2027. The BIS remains sceptical: it argues stablecoins fall short on singleness, elasticity and integrity, the three tests it applies to sound money.
What we would do: design the wholesale programme around tokenized deposits and keep stablecoin settlement as a configurable option. We are inferring, not reporting, that regulated stablecoins will become acceptable to most bank risk committees once the US and EU regimes have a year of supervision behind them.
Permissioned ledger or public chain?
Choose the ledger your counterparties and supervisors can already see; for most institutions in 2026 that means controlled access first, with public-chain distribution only where every wallet is screened. The design choice is about who can hold the asset, not about the technology.
The DTC tokenization pilot shows the shape a regulator will accept today. Under an SEC staff no-action letter of December 2025, DTC participants may record security entitlements as tokens on approved blockchains and transfer them between registered wallets that DTC screens for sanctions before registration. The letter runs three years from launch. During the pilot those tokens carry no collateral or settlement value at DTC.
Capital treatment may still move. The Basel Committee’s revised cryptoasset standard (opens in new tab) carried an implementation date of 1 January 2026, but in February 2026 the Committee said it had accelerated a targeted review (opens in new tab) and would give an update later in the year. Build the programme so a change in capital treatment changes the ledger choice, not the whole plan.
In the EU, the settlement token decides which intermediaries can touch the product. In an October 2026 opinion (opens in new tab), ESMA told national authorities that crypto-asset service providers should not provide services for asset-referenced or e-money tokens that do not comply with MiCA. Our inference: a public-chain product that settles in a non-compliant stablecoin should expect to lose EU service providers as distributors.
Who holds the keys, and on whose balance sheet?
Custody is a design decision, not a procurement step: who holds the keys determines bankruptcy remoteness, the operational risk charge and who answers to the supervisor. Decide it before choosing a platform vendor, because many platforms bundle a custody model you would then inherit.
The accounting obstacle has eased for US reporters. SEC Staff Accounting Bulletin 122 (opens in new tab), effective 30 January 2025, rescinded the earlier staff guidance on safeguarding crypto-assets for platform users. Safeguarding obligations are now assessed under the contingent-liability guidance in ASC 450-20 or IAS 37. That eases the accounting treatment, but not the work of segregating assets and controlling keys.
The three options are self-custody under an existing charter, a regulated sub-custodian, or a split model where the institution operates and a third party holds. Each moves a different risk to a different party. Write down which risk you are keeping before you sign.
What should the board approve, and in what order?
The board should approve five things in sequence: the single use case, the risk appetite, the operating model with a named owner per workstream, a production gate with a date, and a kill criterion. Each later decision depends on the earlier one, so approving them as a bundle hides the dependencies.
The operating model is where most plans are thin. Name the person accountable for legal opinion, custody, treasury operations, reconciliation, technology and the regulatory file. If any line reads “working group,” it has no owner.
What goes wrong: the strategy nobody could execute
The common failure is a well-argued strategy with no path to the second week. The board receives a target state, three use cases and a vendor longlist. Legal waits for the technology choice, technology waits for the custody model, and custody waits for legal.
Months later the pilot runs in a sandbox with test balances. It never touches a client, never reconciles against the core ledger, and never meets the supervisor. The budget cycle closes and the programme is renamed.
The warning signs are visible early: no production date, no kill criterion, no named person for reconciliation, and regulatory engagement scheduled after the build. Fix the sequence: one use case, one owner per workstream, and a conversation with the supervisor before vendor selection.
How Alcyone approaches it
Alcyone starts from the execution end: what has to be true on launch day, who owns each part, and which decisions are blocking the others. In one engagement its partners delivered a compliant, live sustainability NFT initiative for a national banking association in a regulated environment. A fixed-scope assessment lands in week two, and the plan is built over three to six weeks with a named owner for every part; see Alcyone’s strategy and launch-readiness advisory (opens in new tab). There is no obligation to use TrustSwap products.
Where the programme needs segregated custody of reserves or of assets backing an issued token, issuer custody through Panthea (opens in new tab) pairs reserve tooling operated by Panthea with assets held at Blockfort, a Swiss-regulated financial intermediary and VQF member.