IBM argues that tokenised assets, stablecoins and central bank digital currencies will become table stakes by 2030. It describes tokenisation as a fundamental rewiring of how value moves and warns that the decisions banks make during the next two years could create an irreversible competitive advantage.
It is absolutely a strong report that delivers a wakeup call. It asks serious questions about market structure, liquidity, custody, infrastructure and the future role of banks. It is also refreshingly honest about how early we still are. While 26% of the executives surveyed say tokenisation is central to their strategic direction, only 9% are live or ready to deploy an initiative. More than half expect tokenised and conventional financial systems to coexist.
But as I was reading it, I kept coming back to a more basic question. And one you know I keep having if you are a frequent listener of our podcast Fintech Daydreaming.
What problem are we actually solving?
This is the point I feel that the conversations about tokenisation tend to become religious. The crypto believers prepare to explain why I do not understand the technology. The critics dismiss the entire subject as speculation. In Europe, someone will inevitably mention the digital euro, and the discussion heads off in another direction entirely.
That is not where I want to go today.
I am setting central bank digital currencies aside because they are fundamentally different in my view. A CBDC is a public-policy and monetary-infrastructure decision. Banks may be required to participate whether or not they believe there is an immediate commercial opportunity.
Stablecoins and most other forms of tokenisation must earn their place.
They need to solve something.
Moving money is only one part of a payment
Stablecoins are often presented as an answer to slow and expensive cross-border payments.
The argument is attractive. Stablecoins can be transferred across a blockchain almost instantly, at any hour of the day, without passing through a chain of correspondent banks. The transaction can be transparent, programmable and relatively inexpensive.
All of that may be true.
But a cross-border payment is not simply the movement of a digital object between two wallets.
Imagine someone in Helsinki sending €500 to a family member in another country. The recipient needs local currency to pay rent, buy food or withdraw cash.
The complete transaction might look something like this:
Euros are exchanged for a dollar-denominated stablecoin. The stablecoin is transferred across a blockchain. It is then exchanged into local currency and delivered into a bank account, mobile wallet or cash distribution network.
The blockchain transfer in the middle may take seconds.
But the sender still needs an on-ramp. Someone must perform the foreign-exchange conversion. Both parties may need to be identified and screened. Liquidity must be available in the destination market. The recipient still needs an off-ramp into money that can actually be used.
The middle of the transaction may have become more efficient. The payment has not necessarily become cheaper, easier or safer from the customer’s perspective.
The BIS made this point rather clearly in April 2026. A transfer between two blockchain addresses may be fast and cheap, but a proper cost comparison must include exchange fees and the on- and off-ramps. For smaller transactions, the total cost can be higher than a bank transfer.
That does not mean stablecoins can never improve remittances. It means we should stop comparing the cheapest part of a stablecoin transaction with the total price of a conventional one.
The stablecoin casino
I have written before that stablecoins remind me of casino chips.
Inside the casino, they work brilliantly.
Everyone recognises them. They move instantly. There is plenty of liquidity. Nobody needs to convert the chip into another form of money each time they move from the roulette table to the bar.
The problems begin at the door.
Outside the casino, the chip only has value because someone will exchange it for money that works elsewhere. Its usefulness depends on the issuer, the cashier, the redemption rules and the continued willingness of other people to accept it.
A credible stablecoin is obviously more sophisticated than a casino chip. It may be backed by high-quality liquid assets and issued under a serious regulatory regime. It can also be transferred between people who have never entered the same physical or digital venue.
But the underlying issue remains.
Stablecoins are most efficient when the sender and recipient are already inside the same ecosystem and are both willing to remain there.
This is why their strongest established use case has been crypto trading. Participants need something that behaves like cash without repeatedly leaving the crypto ecosystem. The BIS says crypto trading remains the dominant stablecoin use case, followed by access to offshore stores of value in countries with vulnerable currencies. Its assessment of cross-border payment performance is more mixed once spreads and conversion costs are included.
If a recipient wants to keep the stablecoin rather than convert it, the economics improve significantly.
But that tells us something important about the actual proposition. The product may not really be a better remittance. It may be easier access to US dollars.
That can be extremely valuable in a country with high inflation, weak banking infrastructure, capital restrictions or limited access to foreign currency. I do not dismiss that value.
I would simply describe it accurately.
It is digital dollarisation, rather than proof that tokenisation has fixed cross-border payments.
Are we solving a technology problem or a coordination problem?
There is no question that international payments still have problems.
They can be slow, opaque and expensive. Banks repeat compliance checks. Data is truncated or poorly structured. Payment systems operate at different times. Foreign-exchange liquidity is fragmented. Smaller institutions depend on correspondent banks that may have little economic incentive to serve certain corridors.
But many of these are not technology problems.
They are problems of regulation, standards, market access, liquidity, liability and coordination between institutions and countries.
Instant-payment systems already demonstrate that regulated account-to-account money can move in seconds. Within Europe, SEPA Instant significantly weakens the argument that a new form of private money is needed merely to make a payment immediate.
Connecting domestic instant-payment systems, improving data standards, extending operating hours and creating more efficient foreign-exchange mechanisms could address much of the remaining cross-border friction. The G20 and Financial Stability Board have been pursuing precisely this through the cross-border payments roadmap.
So why has progress been so slow?
Because getting hundreds of institutions, regulators and infrastructures to agree is hard.
That is the strongest counterargument to my scepticism.
It may be theoretically possible to repair existing rails, but that does not mean the industry will do it. Stablecoins can bypass some of the coordination challenge. An issuer can create a globally transferable asset without waiting for every domestic payment system to become interoperable.
That is real value.
But bypassing a problem is not the same as solving it.
Instead of depending on correspondent banks and payment systems, users depend on an issuer, a blockchain, wallets, exchanges, market makers and liquidity providers. Instead of fragmented domestic infrastructures, we may get fragmented stablecoins operating across multiple incompatible blockchains.
The BIS has warned that stablecoins on different chains are not inherently interoperable and that this fragmentation undermines the fungibility expected of money.
We may remove one chain of intermediaries and create another.
It might still be a better chain. But we should evaluate the complete arrangement rather than assume that a blockchain has somehow removed intermediation.
Tokenising an asset does not create a market
The same concern applies to the wider tokenisation discussion.
We are told that tokenisation will make assets more liquid, divisible, programmable and accessible.
Sometimes it might.
But fractional ownership is not new. Investment funds, listed shares, securitisation and collective investment structures have allowed people to own fractions of larger assets for decades.
Programmability is not new either. Financial institutions already use workflow engines, rules platforms, APIs and automated contract execution.
Nor does creating a token automatically create liquidity.
A tokenised building is still a building. A tokenised invoice still carries the credit risk of the company expected to pay it. A tokenised work of art does not suddenly acquire a continuous market of willing buyers and sellers.
Liquidity comes from market participants, credible valuations, legal certainty, market makers, distribution and confidence that the asset can be sold when needed.
A token may make ownership easier to record and transfer. That is useful. But a more transferable representation of an illiquid asset is not necessarily a liquid asset.
This is where I think parts of the IBM report move too quickly from technical capability to economic outcome. It states that tokenising invoices, inventory and real estate can draw in investors, reduce dependence on traditional lenders and improve access to capital.
Perhaps.
But the token itself does not improve the quality of the underlying credit or create investor demand. It may reduce the operational cost of distribution and servicing. It may make smaller investment sizes practical. Those are meaningful benefits, but they are not the same as creating liquidity from nothing.
Where tokenisation may genuinely change the model
There is one area where I see a relevant argument.
Tokenisation can allow money, assets and contractual conditions to exist on a shared infrastructure. That can enable atomic transactions, where the asset and payment move together, or not at all.
In securities, collateral and wholesale markets, this could reduce reconciliation, counterparty exposure and the number of separate records maintained by participants.
That is potentially important.
Today, two institutions may agree a transaction, exchange messages, update separate systems, move cash through one infrastructure and move the asset through another. Each stage needs controls and reconciliation because no participant fully trusts the records maintained by the others.
A legally recognised shared state could remove some of this duplication.
But the value comes from the shared state, common rules and legal recognition. It does not necessarily come from public blockchain technology or a private stablecoin.
A permissioned industry ledger, tokenised bank deposits or improved financial-market infrastructure could deliver many of the same outcomes.
And the moment the cash, asset, identity or legal record sits outside the shared environment, interfaces and reconciliation return.
The casino gets another door.
Does agentic AI really require tokenised money?
IBM also links tokenisation closely with agentic AI. The report argues that autonomous financial agents will demand programmable, tokenised rails to operate at scale.
I am not convinced.
An AI agent needs a reliable way to identify itself, prove that it is acting with authority, understand limits, initiate a transaction and deal with exceptions. It needs controls around consent, fraud, liability and recourse.
None of these inherently requires a token.
Agents can already call APIs, use conventional bank accounts, initiate account-to-account payments and operate through card networks. Existing money can be made programmable through rules and software without changing its legal form.
Tokenised assets may be useful when an agent is managing collateral, executing complex conditional trades or operating inside an on-chain market. But ordinary agentic commerce does not require every euro or dollar to become a token.
The danger is that we confuse machine-readable instructions with tokenised money.
Software has been following payment instructions for quite a long time.
The question banks should ask
I am not arguing that banks should ignore tokenisation.
That would be as lazy as assuming that every tokenisation initiative will transform finance.
Banks should build the architectural capabilities that give them options: reliable APIs, event-driven integration, clear data ownership, modern ledgers, real-time liquidity information, strong digital identity and the ability to operate continuously.
Those capabilities support stablecoins, tokenised deposits, central bank digital currencies and improved conventional payment systems.
But tokenisation should not become the north star for every modernisation programme.
Progressive transformation should begin with the business outcome and the constraint preventing it. It should then consider the deliberate build, buy and consume choices available, including whether an existing rail can solve the problem more simply.
Before approving a tokenisation initiative, I would ask some fairly blunt questions.
What customer or market problem cannot be solved using existing regulated mechanisms?
Is the claimed saving measured across the complete transaction, including foreign exchange, compliance, liquidity and on- and off-ramps?
Does the recipient actually want the token, or do they immediately need to convert it?
Which intermediaries and dependencies disappear, and which new ones replace them?
What happens when markets are stressed, liquidity disappears or everyone tries to redeem at the same time?
If the answers are convincing, there may be a real use case.
If the answer is mainly that the token moves quickly, we probably have a technology demonstration looking for an economic problem.
I may still be wrong
Stablecoin advocates may ultimately be proven right.
Wallets could become ubiquitous. Merchants, employers and governments might accept stablecoins directly. People may increasingly choose to hold and spend digital dollars without converting back into domestic bank money.
If that happens, the on- and off-ramp problem becomes less important because the tokenised economy is no longer a separate casino. It becomes part of the high street.
But that is a network-adoption thesis.
It is not proven merely because a blockchain transaction is fast.
The IBM report is right to force banks to think seriously about how money, assets and market infrastructure may evolve. Dismissing tokenisation completely would be unwise.
Equally, banks should not mistake momentum, executive interest and impressive technical demonstrations for evidence of sustainable customer value.
Tokenisation can make moving the chips around the casino remarkably efficient.
The question is whether customers want better chips, or whether they would rather fix the payment system outside.
Join us for the Helsinki launch of Rip Out the Core: A DIY Guide to Platform-Enabled Banking Transformation:
When: Thursday 13 August 2026, 16:30 to 19:00 Where: Maria 01, Helsinki Hosted by Digital Banking Academy & Fintech Farm. Spaces are limited, so sign up early via this link.
