The short version: Central banks and the largest custodians are already moving tokenized settlement from theory into live infrastructure. The Bank of Canada's June 2026 framework for tokenized financial systems — programmability, integration, openness — is useful. But it was already behind events when it was published.
This is inconvenient for frameworks, which generally prefer their examples to remain politely in the future.
The "permissioned networks won" consensus is broadly right. But it answers only the easy half of the security question. The connections between permissioned networks are where the money has actually been stolen, and there will be more of those connections, not fewer.
The case for tokenization is weakest exactly where instant payment rails already operate: fast domestic cash settlement. It is strongest where conventional rails cannot reach: bundling both legs of a transaction, cash and asset, into one atomic event.
That is the structural centre of the piece.
Norway is a useful place to watch this from: a central bank doing real liquidity modelling, a domestic instant-payment rail in NBO INST, a major bank, DNB, that has chosen not to join the European stablecoin consortium, and NBX, a listed crypto operator issuing a dollar-backed stablecoin on public blockchain infrastructure.
The signal is not that Norway is absent from tokenization.
It is that different Norwegian actors are placing bets on different layers of the same stack.
A Framework That Was Already Behind Events
On June 12, 2026, the Bank of Canada published a framework for evaluating tokenized financial systems along three dimensions: programmability, integration, and openness.
It cited BIS Project Agorá as an example of integration. But two weeks earlier, on May 27, the Bank of Canada had formally joined Project Agorá, and the project had already demonstrated tokenized commercial bank deposits and tokenized central bank reserves settling cross-border payments instantly across participating jurisdictions.
The framework was published as forward-looking analysis. The example it used had already happened by the time the framework went to print.
That is roughly the speed this space now moves at.
For a bank, the three dimensions translate into three plain questions.
Programmability: which manual processes could run automatically when conditions are met, and what happens to controls when they do?
Integration: if settlement becomes instant and continuous, what happens to overnight liquidity, intraday funding, and systems designed around batch processing?
Openness: which parts of the bank's deposit base and payment infrastructure can be exposed to wider networks, and which parts must stay inside a controlled perimeter?
These are not abstract blockchain questions.
They are deposit, liquidity and counterparty questions.
In other words, the core job of a bank's treasury desk.
A place where the word "innovation" tends to arrive only after someone has asked who is funding it.
Three Things That Get Called Tokenized, and Are Not the Same
Before going further, it is worth separating three things that are often bundled together as "tokenization" but matter to banks in very different ways.
Stablecoins are digital tokens designed to maintain a steady value, usually one dollar or one euro, backed by reserves held by the issuer. They function like digital cash. Under new rules in both the US and EU, the issuer generally cannot pay the holder interest on them, and regulation is keeping this function separate from ordinary bank deposits.
Tokenized money market or Treasury funds are investment products. They are conventional funds holding short-term government debt, but ownership is recorded on a blockchain. BlackRock's BUIDL, Circle's USYC and Franklin Templeton's BENJI are examples. These can pay yield because they are funds, not cash substitutes.
Tokenized securities and tokenized deposits are the category that changes bank infrastructure. This is not a new investment product. It is the same bond, security or client deposit a bank already holds or owes, represented on a shared ledger so it can settle, move or be pledged as collateral faster.
That distinction matters because the market-facing products get the headlines, but the operational decisions sit in the third category.
Stablecoins compete for payment flows.
Tokenized funds compete for institutional cash.
Tokenized deposits and securities change the settlement layer itself.
That is the layer banks cannot delegate to an innovation team and revisit next quarter, unless the plan is to discover the balance-sheet consequences later, with everyone present and nobody responsible.
Programmability: What Project Agorá Actually Changes
Project Agorá's headline is "instant cross-border settlement", but the detail that matters for banks is not only speed.
It is what happens to messaging, compliance checks and settlement.
Agorá is best understood as a technology upgrade to correspondent banking, not a fantasy in which correspondent banking politely disappears because someone added a token.
It still uses banks.
It still uses regulatory checks.
It still has routing, sanctions screening, fraud checks, AML controls, payee verification and jurisdiction-specific constraints.
How traditional.
How deeply unfashionable.
And also, how necessary.
The difference is that Agorá tries to collapse the messaging and settlement of a cross-border payment into one conditional, indivisible atomic event.
Today, a cross-border payment often passes through several intermediary banks. Each one runs its own sanctions, AML and compliance checks in sequence. That is one reason cross-border payments can be slow, expensive and operationally awkward.
A relay race where every runner insists on re-tying everyone's shoes before passing the baton.
Agorá's design tests whether those checks can be orchestrated differently. Some checks can still be performed bilaterally and off-ledger, where banks and regulators are used to doing them. But the results are fed back into the settlement logic, and the payment is only finalized when the required checks have passed.
In plain English: the message and the settlement stop being two separate events.
They become one conditional atomic event.
That is a much stronger argument than "blockchain makes payments faster", which is usually where the hopium pipe enters the room.
There is a broader point here that deserves its own follow-up piece.
Tokenization does not only make settlement faster. It allows more of the transaction logic to be embedded into the transaction itself: conditions, routing, conversion, settlement preferences, compliance triggers and tailored outcomes.
In the old model, much of this sits beside the payment, in messages, workflows, spreadsheets, reconciliations and systems that politely pretend to be integrated because they share a file format.
Very modern.
Very ISO.
Still a sidecar.
The more interesting model is transaction logic that travels with the transaction. Mure is one adjacent example of where this design direction is already visible in the market: one transaction, specified by outcome, with routing, conversion and execution handled beneath the surface.
That is not the core subject of this piece, and it should not become a product detour here. But it matters because it points to the next layer of the tokenization argument: not just faster settlement, and not just atomic settlement, but programmable, tailored financial transactions that reduce the amount of operational choreography banks have to perform after the fact.
The attractive part of programmability is therefore not automation for its own sake. It is the possibility of making one transaction satisfy multiple legal, operational and commercial conditions without reintroducing the delays the technology is meant to remove.
The less attractive part is what happens when the automated system is attacked.
The recent pattern in digital assets is instructive. Earlier losses were often smart-contract failures. By 2026, attackers had shifted toward infrastructure, credentials and cross-network messaging. The KelpDAO incident, where a forged cross-chain message was used to drain funds, is a useful example because it was not a failure inside one controlled system. It was a failure in the connection between systems.
That matters for banks because the institutional answer to tokenization is usually "permissioned networks".
Correct.
But incomplete.
A permissioned network can control who gets inside one system. It does not automatically solve how that system talks safely to the next one.
Permissioned is not a magic spell.
It is a guest list.
Integration: The Liquidity Question Nobody Can Skip
Today, banks settle many obligations in batches after netting out who owes whom. Netting is what lets a bank support large payment and trading volumes without holding the full gross amount in cash throughout the day.
Instant atomic settlement changes that.
The upside is obvious: payment and delivery happen together, or not at all.
The catch is equally obvious once stated: if everything settles instantly, everything has to be funded instantly.
This trade-off is sometimes called the atomic settlement paradox. The system gets operationally faster and financially less forgiving at the same time.
A rare innovation that improves speed and immediately sends treasury a bill.
Norges Bank has modelled exactly this. Its March 2026 staff paper asks what happens when tokenized settlement takes place either through existing infrastructure or natively on a new ledger, and whether the banking system is operating with abundant or scarce central bank reserves.
When reserves are abundant, the difference is manageable.
When reserves are scarce, a large instant settlement late in the day can leave a bank short before it can obtain new liquidity. That can push up overnight rates unpredictably and force the central bank to intervene more actively during the day.
This is why tokenized settlement is not simply an IT upgrade.
It changes liquidity timing.
Treasury will notice.
Treasury tends to notice when cash has gone somewhere.
The Eurosystem has effectively split its answer into two tracks. Pontes is the near-term bridge, allowing tokenized assets traded on private platforms to settle in central bank money through existing TARGET infrastructure. Appia is the longer-term build, aimed at a fuller European platform for tokenized financial markets.
A third decision may matter just as much: from March 2026, the Eurosystem started accepting eligible tokenized bonds issued through existing central securities depositories as collateral for central bank lending, on the same terms as ordinary bonds.
That quietly removes one of the hidden questions banks had about tokenized assets.
If we hold it, can we still use it for liquidity?
In the euro area, for properly issued tokenized bonds, the answer is now yes.
A sentence short enough to hide how important it is.
Openness: The Question Everyone Ducks
The Bank of Canada framework presents openness as a spectrum: open public blockchains at one end, closed permissioned networks at the other.
In practice, the market is converging on a third model: take an existing, centrally governed system and add a blockchain record-keeping layer without changing who is legally in charge.
DTCC is the clearest example. It already holds custody for a huge share of US equities and government debt, around $114 trillion.
A number large enough that writing it out in full feels like showing off.
In December 2025, DTCC received approval to offer blockchain-based digital twins of securities it already custodies. The securities may be represented on-chain, but DTCC remains the legal record-keeper and the existing governance model stays in place.
For European and Nordic banks, this is likely the template: add a blockchain layer to existing custody, but do not abandon legal control.
Blockchain, but with adults still in the room.
As far as it goes, this is the right answer.
It is also where the easy part ends.
"Permissioned won" answers how a single network is controlled. It does not answer how two permissioned networks, run by different banks, central banks, custodians or consortia, coordinate one transaction across both systems.
That connection is where the risk sits.
Closed networks reduce risk inside the fence.
They do not remove the fences.
And fences need gates.
That is the genuinely unresolved design question. Not open versus closed. Interoperable under what governance, with what security model, and what happens when something goes wrong between networks?
A vendor or central bank project that says "we chose permissioned" has answered the easy 80 percent and left the painful 20 percent for later.
Banks should not let "later" become the architecture.
Later has a habit of arriving with invoices.
Why Build Any of This When Instant Payments Already Exist?
The second question is blunter.
Why build tokenized settlement when SEPA Instant already exists?
Since 2025, instant euro payments have become mandatory across the euro area. Norway is building its own kroner equivalent, NBO INST, because SEPA Instant covers euro, not NOK. Both move the cash leg faster, in central bank money, around the clock.
That is important.
It is also one leg.
The cash leg is becoming more ambitious too. The next frontier is not only domestic instant settlement, but cross-currency instant settlement in central bank money. The TIPS cross-currency pilot has shown that instant, atomic cross-border settlement in central bank money is no longer merely a design objective. Banking Circle's June 18 post is useful here because it shows the commercial layer moving in the same direction: domestic instant scheme participation, One-Leg Out Instant Credit Transfer participation, multi-currency FX and clearing on one platform.
Banking Circle is not the story.
The direction of travel is.
The cash rail is getting faster, domestically and cross-border.
That narrows the space where tokenization can claim "speed" as its own argument.
The real tokenization case is not faster cash. It is cash plus the thing the cash is paying for.
Any transaction beyond a simple payment has two legs: the cash and the asset — a bond, a security, collateral, or some other claim. Instant payment rails improve the cash leg. They do not move the asset leg.
In the US, most equities and corporate bonds have moved to T+1. Europe is moving to T+1 from October 2027. But the structural point remains: the cash and the asset still move through different systems, and someone has to reconcile afterwards that both sides happened correctly.
That reconciliation is not just an inefficiency waiting to be optimized.
It is the consequence of conventional infrastructure never having had a mechanism to make both legs one event.
The destination is T+0, or more precisely atomic settlement: cash and asset moving together, as one transaction, or not moving at all.
For that to happen, one settlement logic must be able to see and control both legs at the same moment. That is possible when both the cash and the asset exist as digital records on a shared ledger, or across ledgers coordinated by a robust synchronization layer.
Representing the asset that way is what tokenizing it means.
Atomic settlement is therefore not a feature added to tokenization.
It is the thing tokenization makes possible.
But this needs one important correction.
Atomic settlement is not hypothetical. It already exists in controlled institutional settings. Quantoz operates EURQ and EURD within tightly controlled parameters. 21X, under the EU DLT Pilot Regime, integrates Quantoz tokens for atomic settlement of tokenized securities. Archax has tokenized interests in abrdn's Euro Money Market Fund on Algorand using Quantoz EURD as settlement currency, with transfers between whitelisted entities. Societe Generale-Forge restricts EURCV to onboarded counterparties under its AML/CFT framework. Circle's EURC operates through authorized CASP networks with screening infrastructure around the transfer flow.
The pattern is consistent.
Atomic settlement works today where the perimeter is controlled.
Everyone is onboarded.
The wallets are known.
The venue is licensed.
The token does not pretend to circulate freely through the entire internet while maintaining institutional AML visibility by magic.
A bold strategy, generally best left to pitch decks and people not personally responsible for the risk register.
That is why the distinction matters. The unresolved question is not whether atomic settlement can be built. It has been built.
The unresolved question is whether it can become broad market infrastructure across separate venues, networks, currencies and asset classes without losing the things regulated finance cannot lose: legal finality, liquidity access, AML controls, privacy, operational resilience, and a credible answer to what happens when the system needs to connect to the next system.
A conventional bond sitting in a traditional custody system cannot be one half of an atomic transaction, no matter how fast the payment rail is. It simply does not exist somewhere the cash leg's system can see or move it.
This reframes the case for DLT.
It is weakest where instant payment rails already operate.
It is strongest where the two-legs problem exists.
It also helps explain why regions with older infrastructure may suddenly move faster than expected. A market starting from paper checks has less legacy instant-payment infrastructure to defend. The next thing built can aim directly at the two-legs problem.
Add regulatory clarity arriving in roughly the same eighteen months — MiCA, the GENIUS Act, and active US policy support — and you get the Ketchup effect: years of nothing, then everything at once.
Any bank evaluating a tokenization proposal should apply the same distinction.
If the pitch is "faster domestic payments", the burden of proof is high.
If the pitch is "cash and asset settle as one legally final event, with no reconciliation afterwards", the next question is not whether that is possible.
It is.
The next question is: inside what perimeter?
The "Start Internally" Trap
This is where interoperability becomes immediate rather than theoretical.
A large bank is rarely one legal entity. It is a holding company with subsidiaries across countries, each with its own balance sheet, regulator and booking systems.
Moving money or securities between a bank's own Norwegian and Swedish subsidiaries is legally not so different from moving them between two unrelated banks. It still runs through processes that create delay, trapped liquidity and operational work.
That makes "tokenize our own intra-group flows first" an attractive starting point.
The membership list is short.
The counterparties are known.
The bank controls every node.
The risk committee can be shown a clean diagram.
Everyone enjoys the clean diagram.
It is also a trap if the architecture is designed only to talk to itself.
The moment the bank wants that internal network to settle with a correspondent, connect to Agorá or Pontes, or serve a corporate treasury client moving tokenized cash between its own group entities, the closed loop faces the same problem as everyone else.
Either bridge it outward, recreating the attack surface described above, or rebuild on whatever external standard eventually wins.
A network designed only to talk to itself does not become interoperable because someone later adds "external connectivity" to a roadmap slide.
At that point, the bank is not executing a strategy.
It is smoking hopium in a windowless meeting room.
This is the principle worth pushing, not any single vendor implementation.
The same point shows up in BCG's May 2026 digital assets report: do not issue tokenized deposits or digital RWAs without a network strategy. Without interoperability, they risk becoming costly features rather than strategic assets.
That is exactly the trap.
A tokenized deposit that only works inside one institution may be useful. A tokenized security that only settles inside one venue may be useful. But neither is strategic unless it has a credible answer to where it connects next.
Otherwise, the bank has not built infrastructure.
It has built a very expensive room.
Canton, Corda, Chainlink/Swift and Taurium all approach parts of this problem from different directions. Canton uses a network-of-networks model with private sub-ledgers coordinated by a shared synchronizer. Corda has long been used for permissioned institutional workflows with controlled data visibility. Chainlink's CCIP and Swift-related work approach interoperability through messaging and cross-network infrastructure. Taurium is earlier stage, but is aimed at deterministic settlement of tokenized financial instruments, stablecoins and digital securities under FMI-grade assumptions.
The vendor landscape is not settled.
It would be too early, and too convenient, to declare a winner.
The better question is which interoperability principle a bank is prepared to bet its architecture on.
Point-to-point bridges. Protocol-level native interoperability. A governed synchronization layer. Existing financial messaging extended into tokenized settlement. Some hybrid of the above.
The expensive mistake is not choosing the wrong vendor on day one. The expensive mistake is building a closed internal ledger on the assumption that interoperability can be solved later, and discovering that "later" is exactly where the attack surface, legal uncertainty and migration cost all arrive together.
That is a different board paper.
And a longer one.
Which is how you know it is serious.
Norway: A Different Layer of the Stack
Norway brings together all three dimensions: liquidity, settlement, and layer selection.
Norges Bank finished Phase 5 of its CBDC exploration in late 2025 and concluded that a central bank digital currency — retail or wholesale — is not currently warranted. But it also made clear that this could change, and that tokenization research will continue.
That research includes the liquidity modelling described above: exactly the scarce-reserves, instant-settlement scenario that tokenized settlement would create.
At the same time, Norges Bank is building NBO INST, its infrastructure for 24/7 instant settlement of Norwegian kroner in central bank money. This addresses the domestic NOK cash leg, just as SEPA Instant addresses the euro cash leg.
Necessary infrastructure.
Wrong leg to solve the whole problem.
A Norwegian bank's tokenization roadmap therefore has to account for a central bank that is doing two things at once: building the conventional answer to faster cash settlement, while researching the tokenized answer to the harder asset-plus-cash settlement question.
Norges Bank, in other words, is hedging.
That is generally considered sound practice for an institution whose job is managing risk for a living.
On the bank side, the broader European pattern is that large banks treat tokenized deposits — not stablecoins — as the serious institutional option for moving money between banks.
That is not a technology preference.
It is a balance-sheet decision.
Tokenized deposits are not just the "bank version" of stablecoins. They are a balance-sheet defence mechanism.
A tokenized deposit remains a bank deposit. It stays on the bank's balance sheet, preserves the funding relationship, and travels through the licence the bank already has. A stablecoin is different. It is a bearer claim on a separately capitalised issuer, backed by segregated reserves, and sits outside the ordinary deposit base.
Stablecoins ask banks to compete with an instrument that may pull liquidity out of the deposit base. Tokenized deposits let banks modernise the settlement rail without first donating the funding model to someone else.
Generous, perhaps.
Not obviously wise.
Strip out the monetary theory and the operating consequence is plain enough: one route lets a bank keep the deposit and modernise the rail. The other asks the bank to let the deposit leave, then build a new regulated structure to follow it.
This is why the DNB/Qivalis decision matters.
Through 2025 and into 2026, a consortium of European banks — including Danske Bank and SEB — built Qivalis to issue a euro-denominated stablecoin under Dutch supervision, aiming to create a European answer to the dollar-dominated stablecoin market.
DNB was in discussions with the group but chose not to join the initial round.
That is worth pausing on.
Usually, "strategic decision" is the phrase used when nobody wants to say the quiet part clearly.
Here, the quiet part is actually useful.
This was not Norway being absent from tokenization. It was Norway's largest bank declining ownership in the layer where deposits leave the balance sheet, while Norges Bank continues work on tokenized deposits and wholesale settlement — the layer where commercial bank money remains inside the regulated banking perimeter.
NBX complicates the picture in a useful way.
Norwegian Block Exchange, Norway's only publicly listed crypto operator, sits on the other side of the stablecoin-layer question as the sole European issuer of USDM, a USD-backed stablecoin issued on Cardano.
That does not make DNB wrong or NBX right.
It shows that the Norwegian market is not choosing one answer.
A major bank has declined ownership in a euro stablecoin consortium. A listed crypto operator is issuing a dollar-backed stablecoin on public blockchain infrastructure. Norges Bank is studying tokenized deposits and wholesale settlement inside the regulated banking perimeter.
Same market.
Three layers.
That is the more useful reading of Norway's position.
Not absence.
Not delay.
Layer selection.
The regulatory context matters too. Norway's MiCA transitional period is ending with the EU-wide deadline, after which unlicensed crypto firms face enforcement. Separately, the EU's DLT Pilot Regime authorizes trading and settlement venues themselves, while MiCA authorizes the firms operating in the crypto-asset market.
For a Norwegian bank, these are two different questions.
Is the venue authorized?
Is the firm authorized?
One does not imply the other.
Regulation, in a rare moment of comic commitment, has created two boxes instead of one.
Six Questions Worth Asking Before Committing
A bank looking at tokenization does not need a philosophical position on blockchains.
It needs answers to six practical questions.
- Does this settle in central bank money, commercial bank money, or a private token? The answer determines the settlement risk.
- Who controls the network, and can that change without our agreement? Governance is infrastructure. If it sits off-ledger and can shift over time, it is not a footnote.
- How does this connect to other networks? This is where the risk has actually appeared. A project that has answered permissioning but not interoperability has solved the easier half of the problem and placed the harder half in a drawer labelled "phase two".
- What does this do that SEPA Instant, NBO INST, TIPS-linked cross-currency settlement, OCT Inst or equivalent domestic instant rails do not already do? The cash leg is moving fast. The legitimate case for tokenization rests on what those rails still do not solve by themselves: bundling asset delivery, cash settlement, currency conversion, transaction logic and cross-network coordination into one legally final event.
- Inside what perimeter does atomic settlement work? A licensed venue, a whitelisted institutional network, a single issuer environment and genuinely interoperable market infrastructure are not the same thing. Similar diagrams. Very different consequences.
- Is this a payments project, or a balance-sheet decision? A tokenized deposit stays on the bank's balance sheet. A stablecoin does not. If treasury is not in the room, the meeting is probably about the wrong thing.
A few weeks of legal and treasury review is cheap.
Building on infrastructure the rest of the sector has quietly decided not to use, and discovering that later, is not.
That is not innovation.
That is archaeology with a budget.
References
- Boston Consulting Group, "The Future of Digital Assets," May 2026. Link
- Bank of Canada, "Tokenization: What it is and how to think about it," June 12, 2026. Link
- Bank of Canada, "Bank of Canada joins BIS Project Agorá," May 27, 2026. Link
- Reuters, "Top central banks forge ahead with 'always-on' cross-border payments testing," May 27, 2026. Link
- Banking Circle, LinkedIn post on TIPS cross-currency pilot, June 18, 2026. Link
- Mure, "One transaction. Any outcome." Link
- IMF, "Tokenized Finance and Money," Tobias Adrian, May 2026. Link
- ECB, "ECB commits to distributed ledger technology settlement plans," July 1, 2025. Link
- ECB, "Building the rails for Europe's tokenised financial markets," Piero Cipollone, March 23, 2026. Link
- ECB, "Pontes." Link
- DTCC, "DTCC Authorized to Offer New Tokenization Service," December 11, 2025. Link
- Norges Bank, "Norges Bank does not currently recommend the introduction of a central bank digital currency," December 10, 2025. Link
- Norges Bank, "Tokenised bank deposits, wholesale CBDC and the central bank's liquidity management," Staff Memo 1/2026. Link
- ESMA, "Report on the functioning and review of the DLT Pilot Regime," June 25, 2025. Link
- Kaupr, "SEB, Danske Bank and seven other European banks join forces to issue Euro stablecoin." Link
- Norwegian Block Exchange, USDM. Link
- Quantoz Payments. Link
- 21X. Link
- Archax. Link
- Regulation (EU) 2022/858 on a pilot regime for market infrastructures based on distributed ledger technology. Link
- Markets in Crypto-Assets Regulation, Regulation (EU) 2023/1114. Link