Tokenization and AI are building a whole new world of money

I just had an interesting podcast conversation about stablecoins, Central Bank Digital Currencies (CBDCs) and tokenised deposits. Most people in finance think they understand the differences, but it’s worth pausing for a moment, because the distinctions really matter.

At a very simple level:

• stablecoins are privately issued digital tokens (think USDC or Tether) backed by reserves;
• CBDCs are digital forms of central bank money issued by the state; and
• tokenised deposits are commercial bank money represented on modern rails, often using distributed ledger technology.

Same idea – digital money – but very different issuers, risks and roles in the system.

For example, stablecoins today process tens of trillions of dollars annually, but the majority of that volume is still within internal financial and crypto markets rather than everyday payments. Meanwhile, over 130 central banks are now actively exploring CBDCs, with countries like China already piloting large-scale deployments. And on the commercial side, banks like JPMorgan have been running tokenised deposit platforms (such as JPM Coin) for years to enable real-time institutional settlement.

The bigger point for me, however, is that tokenisation itself is nothing new. As I wrote here, it goes back around 10,000 years to ancient Mesopotamia, where farmers represented ownership of assets like cattle, grain or land using clay tokens stored in sealed pots. Those tokens were essentially the first abstraction of value – a way to represent ownership without moving the underlying asset.

As economies evolved from owning assets to trading them, and then to trading across borders, we needed shared systems of trust. That led to the development of common standards, institutions and infrastructures for commerce. Fast forward a few thousand years and we have global networks like SWIFT, and schemes like Visa and Mastercard, all acting as trusted intermediaries for moving value. These systems handle billions of transactions every day, but fundamentally they are messaging layers – they tell banks what to do, rather than actually moving the asset itself in real time.

But here’s the shift: those systems were built around messaging and authorisation. They move instructions about money. The actual settlement still sits somewhere else, often delayed, reconciled, and dependent on multiple parties. Cross-border payments, for example, can still take days because of correspondent banking chains, liquidity constraints and time zone mismatches.

Tokenisation changes that.

With tokenised money – whether stablecoins, CBDCs or tokenised deposits – the asset and the instruction can become one and the same.

The token is the value.

That means settlement can be instant and programmable. You’re not just sending a message about money; you’re moving the money itself. This is why central banks and institutions are experimenting with “atomic settlement” and “delivery-versus-payment” models on distributed ledgers, where cash and assets exchange simultaneously with no settlement risk.

Even more interesting is what happens when you combine tokenisation with automation and AI.

We’re moving from a world where humans authorise every transaction to one where we delegate authority to systems. These systems act on our behalf, within rules we define. In other words, money starts to become agentic.

Money can now be moved, authorised and settled without a human touch. It is all machine-driven. Machines can exchange tokens in a wholly delegated manner that is proactive, not reactive.

This is the critical difference.

Think about supply chains where machines automatically reorder inventory and settle invoices, or IoT devices that pay for energy, data or services in real time. In these environments, payments are embedded into the process itself.

We’ve already seen a glimpse of this in financial markets.

Algorithmic and high-frequency trading systems have been making decisions in microseconds for years, often with no human in the loop. Markets move, positions shift, billions change hands – all driven by code. In fact, the majority of equity trading volume in major markets today is executed algorithmically.

Now extend that idea beyond trading desks and imagine your personal financial life working the same way.

Your “financial agent” monitors your income, spending, subscriptions, investments and obligations. It optimises your cash flow, switches providers, refinances your mortgage, pays bills at the optimal time, hedges currency exposure and invests surplus funds – all automatically.

We’re already seeing many versions of this with robo-advisors, smart budgeting tools and open banking APIs – but tokenisation takes it to another level by embedding execution directly into the asset.

That’s the real implication of tokenisation.

It’s not just about digitising assets. It’s about creating a system where value can move, adapt and act autonomously. It also raises new questions: who is accountable when an AI agent moves your money? How do you set limits, permissions and identity controls? This is where digital identity, authentication and concepts like zero-knowledge proofs become critical foundations for the system.

And that’s why the distinction between stablecoins, CBDCs and tokenised deposits matters. They are not just different forms of money; they are competing building blocks for the future financial system.

Stablecoins may dominate crypto-native and cross-border liquidity use cases; CBDCs may underpin sovereign monetary systems and wholesale settlement; tokenised deposits may become the bridge that allows banks to stay relevant in a programmable world. Most likely, they will co-exist in a layered “digital money stack” but, whichever mix wins, the direction is clear: we are moving from static money to programmable money, and from programmable money to autonomous, agentic finance.

It’s a game changer. Are you keeping up?