AI’s Choice of Money vs Banks’ Choice of Money: The Forkpoint of Digital Finance

※ This article is published in its current draft form and will be updated in 2 days to match the final Daily Crypto Times (DCT) format.

AI’s Money vs Banks’ Money: The Forkpoint of Digital Finance

This article is a direct follow-up to “The Future of Digital Money: Stablecoins vs. Tokenised Deposits” . If you first read that piece, which lays out the basic concepts and differences between stablecoins and tokenised deposits, the structure and arguments in this article will feel more natural and connected.

Major U.S. banks (JPMorgan, Citi, Bank of America and others) are building a shared tokenised deposits network targeting launch in the first half of 2027. This network will be operated through The Clearing House (TCH), the key U.S. interbank payments utility, and aims to provide blockchain-based digital deposits with 24/7 instant settlement fully inside the existing banking regulatory framework.

The crucial point is that this system operates in a way that is fundamentally different from public blockchains like Ethereum. Public blockchains are open networks where anyone can participate, validators are a diffuse set of actors, the network is sustained by token incentives, and regulation, audit and KYC are handled off-chain as separate layers.

In contrast, the tokenised deposits network is a permissioned blockchain restricted to banks, where all transactions are processed within banking regulatory and audit frameworks, and settlement is directly linked to bank ledgers in real time.

If public chains are built around “openness and decentralisation”, tokenised deposit networks are built around “regulatory compliance, stability and integration with financial infrastructure”. Understanding this distinction makes it easier to see why U.S. banks are choosing the tokenised deposits model rather than stablecoins as their core digital money infrastructure.

At the same time, digital money infrastructure is facing another major turning point: the rise of AI agents as primary participants in financial transactions. As portfolio rebalancing, liquidity provision and algorithmic trading increasingly move from human-driven to AI agent–driven automation, stablecoins and tokenised deposits reveal very different strengths and limitations.

Against this backdrop, this article explores four key themes:

• 1) Tokenised deposits — the digital money model chosen by banks
• 2) Stablecoins vs tokenised deposits — structural differences in issuance and blockchain design
• 3) Integration with capital markets tokenisation — decisive differences between the two models
• 4) Automatic trading between AI agents — the strengths of stablecoins and the limits of tokenised deposits

Let’s walk through each of these in turn.

1) Tokenised deposits — the digital money model chosen by banks

Tokenised deposits do not create a new form of money; they convert existing bank deposits into blockchain-based tokens.

• Deposits remain on the bank’s balance sheet,
• Tokens handle real-time payment and settlement on the blockchain,
• Deposit insurance, banking regulation and liquidity rules still apply.

In other words, tokenised deposits combine regulatory stability, blockchain speed and programmable payments.

For global banks like JPMorgan, HSBC and BNY Mellon, the appeal is clear: they can implement 24/7 payments, conditional settlement (DvP) and automated financial flows without dismantling their existing core infrastructure.

2) Stablecoins vs tokenised deposits — structural differences in issuance and blockchain design

Stablecoins and tokenised deposits both look like digital money, but they diverge completely in terms of issuer, liability structure, settlement model and blockchain architecture.

① Issuer

• Stablecoins: Issued by non-bank entities (e.g., Circle, Tether)
• Tokenised deposits: Issued by commercial banks

② Legal and risk profile

• Stablecoins: No deposit insurance; token risk is directly tied to issuer risk
• Tokenised deposits: Same legal protection as traditional bank deposits

③ Blockchain architecture

• Stablecoins: Circulate on public chains like Ethereum and Solana
• Tokenised deposits: Operate on permissioned or bank-specific blockchains

④ Payment and settlement model

• Stablecoins: Payments occur on-chain, but final settlement happens off-chain via issuer accounts and backing assets. This makes it structurally difficult for stablecoins to become the core “base settlement layer” of traditional financial infrastructure.
• Tokenised deposits: Payments and settlement are both directly connected to bank systems, with bank ledgers and blockchain ledgers synchronised for real-time settlement.

Because of these differences, institutions that prioritise regulatory compliance, stability and scalability naturally gravitate toward tokenised deposits rather than stablecoins for their core digital money infrastructure.

3) Integration with capital markets tokenisation — decisive differences between the two models

Since 2026, DTCC and other market utilities have been driving large-scale tokenisation of U.S. Treasuries, ETFs, equities and real-world assets (RWA). In this context, the nature of the settlement asset becomes critical. Capital markets tokenisation requires on-chain assets and on-chain money to be fully integrated within a single system.

Limits of stablecoins

• Stablecoins can support on-chain payments, but final settlement still depends on issuer accounts off-chain.
• This makes full integration with core capital markets infrastructure (DvP, securities settlement systems) difficult.

Strengths of tokenised deposits

• Payments and settlement are both directly linked to bank systems.
• Bank ledgers and blockchain ledgers can be synchronised in real time.
• On-chain securities and on-chain money can be combined in a true DvP (Delivery versus Payment) structure.
• Tokenised deposits are well suited to serve as the settlement layer for large-scale capital markets infrastructure.

As capital markets tokenisation expands, it becomes increasingly likely that tokenised deposits, not stablecoins, will form the core settlement layer for institutional financial infrastructure.

4) Automatic trading between AI agents — strengths of stablecoins and limits of tokenised deposits

Looking ahead, the primary users of digital money infrastructure may not be humans at all, but AI agents. Portfolio rebalancing, liquidity provision, hedging strategies and cross-chain arbitrage are already moving toward environments where trades are executed automatically between AI agents with minimal human intervention. In this world, stablecoins and tokenised deposits reveal very different peaks and limits.

Stablecoins in AI-driven automatic trading

• Openness of public chains
  Stablecoins operate on open networks like Ethereum and Solana, allowing AI agents to freely access, create wallets and execute transactions.
• 24/7 global liquidity
  AI agents must be able to trade without time constraints. Stablecoins already provide continuous, real-time liquidity across global exchanges, DeFi protocols and on-chain liquidity pools.
• Maturity of programmable money
  Combined with smart contracts, stablecoins enable AI agents to fully automate conditional payments, auto-liquidations and algorithmic strategies on-chain.
• Potential to become the base unit of AI-native economies
  Because AI agents operate beyond borders, banking hours and local regulations, stablecoins are well positioned to become the de facto base currency of AI-driven digital economies.

Limits of tokenised deposits in AI-driven automatic trading

• Closed nature of permissioned networks
  Tokenised deposits run on permissioned chains restricted to banks, making it practically impossible for AI agents to freely join, create wallets or transact without institutional onboarding.
• Regulation- and KYC-centric design
  Large-scale AI-driven trading requires automated KYC, AML and risk management, but banking systems are fundamentally designed around human-centric regulatory processes.
• Lack of on-chain liquidity
  Tokenised deposits are not natively integrated with public-chain liquidity pools or global crypto exchanges, limiting the instant liquidity available to AI agents.
• Weak connectivity to AI-native economies
  AI agents operate in borderless digital environments, whereas tokenised deposits are tied to specific jurisdictions and banking regulations, making it difficult for them to become the base settlement layer of AI-native economic systems.

In summary, in a future where AI agents dominate automatic trading, stablecoins have clear advantages in terms of openness, liquidity and programmable settlement, while tokenised deposits face structural limits due to their regulation-centric and closed design when it comes to serving as the core currency of AI-native economies.

Conclusion: The future of digital money is a forkpoint, not a winner-takes-all game

Stablecoins and tokenised deposits are both critical pillars of digital money, but their roles and trajectories are diverging more clearly over time. The rise of AI agents as key participants in financial markets only sharpens the contrast between their respective strengths and limitations.

Stablecoins, with their public-chain openness, global liquidity and smart contract–based automation, are naturally aligned with AI agents operating in on-chain, borderless environments. For these agents, stablecoins are poised to become the default unit of account and settlement.

Tokenised deposits, meanwhile, excel in regulatory stability and deep integration with existing financial infrastructure, but their permissioned and restricted nature limits their scalability in open, AI-native digital economies. Instead, they are likely to remain central in interbank payments, capital markets tokenisation and regulated institutional settlement.

The key question is not whether one model will eliminate the other, but where their roles diverge and in which domains each is optimised.

• Stablecoins: Global payments, on-chain liquidity, AI agent–driven automatic trading, crypto/DeFi-native environments
• Tokenised deposits: Interbank settlement, capital markets infrastructure, regulated digital money, real-time institutional settlement

The 2027 launch of U.S. banks’ tokenised deposits network signals that digital money will not converge into a single model, but rather split into two optimised axes serving different layers of the financial system. Standing at this forkpoint, we are left with a deeper question: “In the age of AI, who will design the core infrastructure of digital money, and on which rails will it run?”

Younchan Jung
Researcher exploring structural shifts in AI, blockchain, and the on‑chain economy.

If you would like to read this article in Korean, please click the button below.

Read This Article in Korean →