The battle for the future of money: stablecoins, tokenised deposits, CBDCs or something else?

The interesting story is no longer simply whether bitcoin goes to $50,000, $100,000 or $500,000, but whether the technologies created around crypto are becoming the foundations of a new financial system. If that is right, it raises a much bigger question: what is the money in that system?

For most of our lives, the answer has been fairly straightforward.

Central banks create central bank money, commercial banks create most of the money we actually use, card networks move instructions between buyers and sellers, clearing systems reconcile the obligations and settlement systems eventually move the money.

It is a complicated machine, but consumers don't see the complication.

I tap my card in London, Warsaw or New York and something happens behind the scenes that makes the merchant believe they have been paid. I don't particularly care whether the transaction involved an acquirer, issuer, processor, card network, correspondent bank, clearing house or settlement system. I just bought the coffee.

Now we are rebuilding that machine, and the difference is that the next financial architecture may involve money that is digital, tokenised, programmable, available 24 hours a day and capable of being controlled not just by humans but by software and artificial intelligence.

There is a battle developing over who gets to create that money, who gets to control the infrastructure on which it moves and, perhaps most importantly, who gets to decide which form of money is used when machines rather than people are making the decisions.

The stablecoin proposition

Stablecoins have an obvious advantage because they already exist at scale. The basic proposition is simple: take a dollar, euro or pound, represent it as a token on a blockchain and allow that token to move around digital networks.

That sounds almost trivial, but it changes quite a lot because a traditional bank payment moves through a network of institutions, whilst a stablecoin can potentially move directly between digital wallets. It can be transferred at any time, incorporated into a smart contract and used by software without requiring somebody to log into a bank account and press PAY.

This is why stablecoins have become so interesting to governments, banks, fintech firms, payment companies and technology companies, and it is also why the regulatory debate has moved so quickly.

Europe has MiCA, America has moved decisively towards a federal stablecoin framework through the GENIUS Act, Britain is building its own framework and other jurisdictions are doing the same. Governments have realised that stablecoins are not going away, but they have also realised something more important: stablecoins could become strategically significant to the future of their currencies.

If dollar stablecoins become the default money of the digital economy, then dollarisation spreads into digital networks. Someone in Argentina, Nigeria, Vietnam or virtually anywhere else could potentially hold and transact in digital dollars without maintaining a conventional American bank account. That could strengthen the international role of the dollar rather than weaken it, which explains why other governments and central banks are paying such close attention. Money is never just money. It represents sovereignty, economic influence and power, and putting money onto global digital networks does not change that.

Then the banks woke up

Commercial banks have another idea, which is essentially to ask why we need stablecoins at all.

Banks already create digital money.

When your banking app tells you that you have £10,000 in your current account, there isn't a pile of £20 notes sitting in a vault with your name written on them. What you have is a claim against your bank represented by entries in its ledger, and so the obvious question for banks is why not take that existing commercial bank money and make it programmable?

That is the proposition behind tokenised deposits.

A tokenised deposit is essentially conventional commercial bank money represented in a form that can operate on programmable digital infrastructure. That might sound like splitting hairs, but the distinction is fundamental because a stablecoin is generally a claim against the issuer and the reserves supporting the token, whereas a tokenised deposit remains a deposit with a regulated bank. That means it potentially retains many of the characteristics that make bank money useful, including its relationship with lending, liquidity management, regulation and, depending upon the jurisdiction and structure, deposit protection.

Banks therefore have a powerful argument. Why invent a parallel form of digital money when we can modernise the money that already exists?

The problem is that a tokenised deposit issued by Bank A is not necessarily the same thing as a tokenised deposit issued by Bank B. If every bank creates its own tokens on its own infrastructure, we risk recreating the fragmentation that distributed ledgers were supposed to remove. How does my tokenised HSBC pound move to your Barclays account? How does a JPMorgan tokenised dollar interact with a Deutsche Bank tokenised euro? What happens when the transaction crosses currencies, blockchains and regulatory jurisdictions? We have spent decades connecting banking systems together and may now have to do it all over again.

And then there are the central banks

Central banks have been watching all of this with considerable interest because the ultimate settlement asset in most financial systems is central bank money, which brings us to central bank digital currencies, or CBDCs. The idea has been discussed so much that CBDCs sometimes feel simultaneously inevitable and permanently five years away, but the underlying logic remains compelling. If commercial bank money becomes programmable and tokenised, perhaps central bank money needs to become programmable and tokenised as well.

There is a crucial distinction here between retail and wholesale CBDCs. Retail CBDCs attract most of the public attention because they raise questions about privacy, government surveillance, financial inclusion and whether citizens really want wallets ultimately connected to their central bank. Wholesale CBDCs are less exciting over dinner but potentially far more important to the financial system because, if banks are going to exchange tokenised deposits, securities and other digital assets, they need something trustworthy with which to settle those transactions. Central bank money is the obvious candidate.

This means CBDCs may not replace the money in your wallet at all. Their more important role could be underneath the financial system, providing a trusted digital settlement asset connecting banks and other regulated institutions. The consumer may never see it, just as most consumers today have little idea how central bank reserves, clearing houses and settlement systems work, but the infrastructure could nevertheless become fundamental to the way digital money moves.

Maybe there isn't going to be one winner

This is where I think much of the debate goes wrong because we keep asking whether stablecoins will beat CBDCs, whether CBDCs will destroy stablecoins or whether tokenised deposits will make both irrelevant.

Why should there be one winner?

The existing financial system doesn't have one form of money.

We have cash, commercial bank deposits, central bank reserves, electronic money and various financial instruments that behave like money in particular circumstances. They coexist because they perform different jobs, and the digital financial system will probably develop in exactly the same way.

Stablecoins may become particularly useful for internet-native commerce, international payments and digital marketplaces, whilst tokenised deposits may dominate regulated banking and corporate finance. Wholesale CBDCs could provide the settlement layer connecting regulated financial institutions, whilst retail CBDCs may emerge strongly in some countries and struggle in others. Meanwhile traditional bank accounts, cards and cash will continue operating alongside all of them for far longer than the technology evangelists expect. The interesting question therefore isn't which form of digital money wins, but how all of these forms of money connect.

Then AI changes the question

This is where artificial intelligence changes everything because humans care about brands and interfaces in ways that machines do not. We know whether we bank with HSBC, Barclays, Santander, Chase, Revolut or someone else. We choose Visa or Mastercard cards, open PayPal and deliberately buy USDC or another stablecoin.

An AI agent doesn't need to think like that.

Give an AI agent authority to manage money and its job is to achieve an outcome within the rules you have given it, which means the underlying form of money can become a routing decision rather than a conscious consumer choice.

Tell an AI agent to pay a supplier, book a flight, move spare corporate cash somewhere safe overnight, buy electricity when the price falls below a certain level, convert euros into dollars when the exchange rate reaches a particular point or pay a software provider every time its API is called, and the agent can potentially decide how to execute the transaction. Which rail is cheapest? Which is fastest? Which has sufficient liquidity? Which provides the strongest protection? Which is legally permitted? Perhaps it chooses a stablecoin, perhaps a tokenised bank deposit, perhaps an instant-payment network or perhaps a card. The human may neither know nor care.

That changes the nature of financial competition. For decades, banks and payment companies have fought to own the customer relationship, but in an agentic world the customer may increasingly delegate financial decisions to software. The battleground therefore shifts from getting the customer to choose your bank, wallet or card towards getting the customer's AI agent to choose your money and your payment rail. Banks have spent decades designing products for people. They may soon have to start designing financial services that appeal to algorithms.

Money becomes machine-readable

This is the part that fascinates me most because today's money is digital, but it is not truly software-native. Yes, computers move it around, but enormous parts of finance still rely upon messages between databases. A payment instruction says move money from here to there, systems exchange messages, ledgers are updated, reconciliations take place and eventually everyone agrees about what happened. It works remarkably well, but it is essentially an architecture developed by connecting old financial structures to newer communications technologies.

Tokenisation changes that model because money itself can become an object that software can interact with. Add smart contracts and the money can carry conditions. Add digital identity and the system can know who or what is authorised to transact. Add tokenised assets and money and assets can exist on compatible infrastructure. Add artificial intelligence and software can decide when transactions should occur. Put those pieces together and finance becomes programmable in a way that goes far beyond simply making today's payments faster.

A company could issue an invoice that automatically triggers payment when delivery is cryptographically verified. A machine could purchase electricity from another machine. A supply chain could release working capital automatically as goods move through verified stages. An investment portfolio could rebalance continuously and corporate treasury could become an increasingly autonomous operation where AI agents manage liquidity across currencies, banks and tokenised markets around the clock. None of those agents particularly cares what we call the money because they care whether it works, whether they are authorised to use it and whether it achieves the required outcome.

Which brings us back to banks

This creates an uncomfortable question for banking: what is a bank when money becomes programmable? Banks will obviously still perform enormously important functions because they create credit, manage risk, provide regulated custody, finance companies, manage liquidity and connect households and businesses with the financial system. What may change dramatically are some of the activities wrapped around those core functions as payments become embedded into software, foreign exchange becomes automatically optimised, treasury becomes increasingly autonomous, settlement accelerates and reconciliation reduces where assets and money can share compatible ledgers.

That doesn't eliminate banks, but it changes where the value sits. The winners may be the institutions that provide the trusted balance sheets, liquidity, identity, compliance, risk management and infrastructure upon which autonomous financial services operate. That sounds less glamorous than launching another shiny banking app, but it may be far more valuable because, if software increasingly controls the interface, the financial institution underneath becomes less about owning the screen and more about providing the trust.

The internet of value finally arrives?

We have been talking about an internet of value for decades, and the phrase has always sounded wonderful – years ago, I called it the Valueweb – but there has been a fairly obvious problem: the internet never had native money. It had information, communications, websites, applications, social networks, cloud computing and eventually artificial intelligence, but payments had to be bolted onto the side. Cards were designed decades before the web, bank accounts were designed centuries before it and correspondent banking evolved long before anyone imagined instantaneous global digital communications.

We have therefore spent thirty years connecting twentieth-century financial infrastructure to twenty-first-century digital networks. Stablecoins, tokenised deposits and digital central bank money suggest that we may finally be doing something different by creating money for the network itself, and AI makes that far more important because the next generation of internet users may not all be human. Billions of software agents could eventually transact with companies, people and other agents, making tiny payments, negotiating contracts, purchasing computing resources, managing investments and allocating capital continuously.

Those agents need money, but more importantly they need money they can understand and use. That means digital identity, permissions, programmability, instant settlement and interoperability become as important as the currency symbol attached to the token. Once that happens, the financial system stops being something humans access through websites and apps and increasingly becomes a layer of intelligence and value embedded directly into the network.

So, who wins?

Probably everyone and no one.

Stablecoins have momentum, global reach and the enormous advantage of being internet-native. Banks have trust, customers, balance sheets, regulatory licences and the extraordinary power to create commercial bank money through lending. Central banks have the ultimate settlement asset and the authority of the state. Technology companies have platforms, software, data and increasingly AI agents, whilst card networks have global acceptance and decades of experience connecting billions of endpoints. None of these advantages disappears. Instead, they collide and combine.

The result is unlikely to be a single new financial system replacing the old one overnight. It will be a hybrid architecture in which stablecoins, tokenised bank deposits, central bank money and traditional payment rails increasingly interact, and eventually consumers may have no idea which one they are using. You will tell your AI agent to organise a holiday, pay the electricity bill, move some savings, send money to your daughter or purchase something from a company on the other side of the world, and the agent will work out how. Behind the scenes it might use a bank deposit, stablecoin, CBDC, card, instant-payment rail or something that hasn't been invented yet.

That is why I think the really interesting change is not simply that money is becoming digital, because money has been digital for decades. The change is that money is becoming programmable, interoperable and machine-readable, and the decision about how it moves may increasingly be made by software rather than people. We are moving from a world where people choose how to pay towards one where machines choose how money moves, and that may turn out to be far more important than the argument about whether stablecoins, tokenised deposits or CBDCs ultimately win.

The future of money isn't just digital. It is autonomous and programmable.