Bank money and bearer money

Tokenized deposits and stablecoins are doing different jobs. Banks that pick one are solving half the problem.

May 24, 2026

The lazy version of the current bank digital asset debate is that you have to pick. Tokenized deposits or stablecoins. Bank issued money or bearer instrument. Closed loop or open network. You may hear the framing show up in panel discussions, maybe in internal board memos, or in the consulting decks being pitched to regional and community banks right now.

It really is the wrong question. The US monetary system has been two tier for more than a century. Central bank money sits at the top, held only by banks as reserves at the Fed. Commercial bank money sits below, held by households and businesses as deposits. The two tiers settle against each other through the Fed’s payment infrastructure, and the interoperability between them is what makes the whole system work. Tokenization does not collapse that structure. It reproduces it onchain. The question is not which form of money wins. The question is how the two tier logic gets rebuilt with programmable instruments, and who owns the settlement fabric between them.

That is a harder, more interesting problem than the binary. And it is the one the banks that are actually moving have already started solving.

The two tier model, briefly

Allow me to briefly remind you why this structure exists. Reserves at the central bank are the ultimate settlement asset in a national monetary system. They are a direct liability of the Fed, carry no credit risk against any private counterparty, and are the instrument that interbank obligations clear in. Commercial bank deposits are a different instrument. They are a liability of the issuing bank, carry the credit risk of that bank, and function as the day to day medium of exchange for the real economy. The two tiers are connected by the fact that every commercial bank deposit is ultimately backed by that bank’s claim on reserves, and that interbank transfers of deposits settle in reserves at the central bank.

This is not a theoretical design choice. It is actually the structure that allows the system to scale. The Fed does not want to process every consumer payment. Commercial banks do not want to hold all settlement risk directly on a central bank balance sheet. The two tier model distributes credit creation, customer relationship, and liquidity transformation to commercial banks, while keeping the ultimate settlement asset at the central bank. Every functional monetary system in the developed world has some version of it.

When people talk about tokenizing money, what is actually happening is that each of those tiers is getting a new digital instrument. A wholesale central bank digital currency, or a tokenized reserve equivalent, serves the settlement asset tier. Tokenized deposits serve the commercial bank money tier. Stablecoins, depending on how they are structured, sit somewhere on the spectrum between the two, and that spectrum is where most of the analytical confusion lives.

What stablecoins actually are

A dollar pegged stablecoin is a bearer instrument denominated in dollars, fully reserved 1:1 against some combination of short dated Treasuries, reverse repo, and bank deposits. It is not a deposit liability of the issuer, carries no FDIC insurance, and sits in a segregated reserve pool the issuer cannot lend against. Under the GENIUS Act, payment stablecoins can be issued by three categories of entity: insured depository institution subsidiaries supervised by their primary federal banking regulator, federally licensed nonbank issuers supervised by the OCC, and qualifying state-licensed issuers below the size threshold. The largest issuers today — Tether and Circle — are nonbanks, but the framework is deliberately bank-inclusive. A nine-bank G7 consortium that includes Goldman Sachs, Deutsche Bank, BNP Paribas, Citi, and Bank of America has announced a jointly backed stablecoin, and several large banks have signaled they intend to issue under the new framework. USDC and USDT together account for over four fifths of the roughly $316 billion stablecoin market as of Q1 2026, with USDT at roughly $184 billion and USDC at $78 billion. USDC crossed a meaningful threshold recently, capturing 64 percent of adjusted stablecoin transaction volume for the first time since 2019, driven by institutional preference for the regulated instrument under the GENIUS Act framework. The overall market crossed $320 billion in the second quarter.

What makes a stablecoin structurally different from a tokenized deposit is not the peg or the issuer, under GENIUS a bank can issue either, it is the legal characterization of the instrument and the settlement model that follows from it. A stablecoin is a bearer token, fully reserved against HQLA, segregated from the issuer’s balance sheet, not subject to fractional-reserve treatment. A tokenized deposit is a deposit liability of the issuing bank, backed by that bank’s balance sheet, FDIC insured to the limit, fractional reserve, and subject to the bank’s capital and liquidity rules. The settlement model follows from that distinction. A stablecoin transaction is the bearer transfer of a token from one wallet to another. It settles onchain, finally, in minutes. No bank sits between the two parties. Neither party needs to have a relationship with the other’s bank, or indeed with any bank, to receive value. The instrument is designed for use cases where you want dollar exposure without bank credit exposure, or where the two parties to a transaction do not share a common banking infrastructure. That is a real set of use cases. Cross border commerce between parties in jurisdictions where correspondent banking is slow, expensive, or politically contested. Crypto native trading and custody. Onchain settlement of tokenized assets where the payment leg has to match the asset leg atomically. 24/7 programmable payments between entities that are not commercial bank customers of any single bank.

The BIS has been blunt about what stablecoins are not. In its 2025 annual report, it argued that stablecoins do not deliver “singleness of money, elasticity, and integrity,” and should play at most a subsidiary role in the financial system if adequately regulated. That position is defensible as a statement about what you would want the backbone of a monetary system to look like. It is not a statement about whether stablecoins solve any real problem. They solve an obvious problem. They move bearer dollar value across the internet at near-zero cost, instantly, between parties who do not know each other. No existing instrument does that, and that is why the market cap is over $300 billion.

What tokenized deposits actually are

A tokenized deposit is commercial bank money issued onchain. It remains a liability of the issuing bank, carries the credit risk of that bank, remains subject to the same capital, liquidity, BSA, and AML frameworks the bank’s other deposits are subject to, and is fully redeemable against that bank’s balance sheet. What is different is the representation. Instead of a ledger entry in the bank’s core deposit system, it is a token on a permissioned or hybrid blockchain. The token can be programmed. It can move between wallets on the same network in seconds. It can settle against tokenized assets atomically. It can carry metadata that conditions its own movement.

This is the instrument that the global systemically important banks have chosen. JPMorgan’s Kinexys platform (formerly Onyx, launched in 2020) is processing over $5 billion in daily transactions and has moved more than $3 trillion in cumulative volume since inception, much of it in tokenized deposits. JPMorgan’s tokenized USD deposit, now branded JPMD, went live on Base in late 2025, with Polygon, Arbitrum, and Ethereum mainnet on the announced multi-chain roadmap, extending what was an internal settlement asset out into public infrastructure under a permissioned model. HSBC launched tokenized deposit services in Hong Kong and Singapore in 2025, extended to the UK and Luxembourg, and announced expansion to the US (live April 2026) and the UAE in 2026. Citi, DBS, Standard Chartered, and Deutsche Bank all offer or settle on tokenized deposit infrastructure for institutional clients — DBS and Standard Chartered as founding banks of Partior, Deutsche Bank as a euro and dollar settlement bank on Partior since May 2025, and Citi through Citi Token Services. The Cari Network, announced in Q1 2026, is a five bank consortium of regional lenders — Huntington, First Horizon, M&T, KeyCorp, and Old National — building on ZKsync’s Prividium for shared-ledger tokenized deposits across US regional banks.

What tokenized deposits are good for is the universe of use cases where the client has a banking relationship and wants the movement logic of bank money to become programmable. Intraday liquidity across a multinational’s entities. Cross-border intra-bank transfers that currently sit inside the SWIFT messaging model. Delivery versus payment settlement against tokenized securities. 24/7 movement between corporate accounts at the same bank or between banks on a shared network. These are flows where the user wants the credit quality and legal framework of bank money, and wants to keep that relationship with the bank. They want what the bank already provides, with the additional property of being programmable and real-time.

The two instruments are not substitutes

This is the bit that gets lost in the binary framing. Stablecoins and tokenized deposits are not competing products aimed at the same customer. They are different instruments aimed at different flows. A corporate treasurer moving liquidity between the firm’s Luxembourg and New York entities wants tokenized deposits. A crypto native trading firm settling a large OTC trade with a counterparty on a Saturday wants a stablecoin. A US importer paying a Vietnamese supplier may want one, the other, or a hybrid, depending on whether the supplier has a bank relationship, which jurisdictions the flow is crossing, and whether the payment needs to settle against a bill of lading token.

The actual problem is the settlement fabric

Here is where I think the analysis gets structurally interesting and where most of the commentary falls a bit short. If you accept that both instruments will exist, and that they will serve overlapping but non identical use cases, then shouldn’t the question be what does the settlement fabric between them look like? When a stablecoin has to convert into a tokenized deposit, who runs that conversion, at what timing, against what collateral, with what legal finality. When a tokenized deposit at Bank A has to settle against a tokenized deposit at Bank B, and the two banks are on different networks, what does the bridge look like. When a tokenized Treasury is purchased with a stablecoin and needs to settle delivery versus payment, which ledger is the ledger of record and who reconciles if the two legs fail.

This is what Project Agorá at the BIS Innovation Hub has been working on, and it is what the Regulated Settlement Network and Partior and SWIFT’s blockchain trials are all aimed at. It is also what the Fed’s continued interest in a wholesale tokenized settlement asset is trying to answer from the public side. The work is not done. The standards are not set. The legal and operational finality of cross network settlement is still being built. The banks that understand this problem and have a position on how to solve it are the banks that are going to own the settlement layer of the next generation of money. The banks that treat this as a product-selection question and pick tokenized deposits or stablecoins as though the choice is terminal are one layer too shallow in the analysis.

The historical analogy here is the money market fund era of the late 1970s and 1980s. Regulation Q capped the interest rate banks could pay on deposits. Money market funds offered a substitute that paid market rates. Banks lost a huge amount of deposit share to MMFs before the regulatory response caught up. Critically, the structure that eventually stabilized was not banks beating MMFs or MMFs replacing banks. It was coexistence, with a new settlement and redemption fabric that allowed household and institutional money to flow between bank deposits and MMF shares continuously, depending on the relative economics. Banks that fought the existence of MMFs lost ground. Banks that built products and services that interoperated with the new instrument class kept the customer relationship. The MMF episode produced decades of eurodollar, sweep account, and money market deposit account innovation that became core bank products. The technology changed but the two tier logic did not.

It feels like the same pattern now. Stablecoins are the money market fund of this cycle. Tokenized deposits are the deposit side response. The eventual steady state is coexistence with a functioning settlement fabric between them, and the economics will flow where the economics flow based on use case. The banks that fight this by trying to kill stablecoins or refusing to build tokenized deposits are gonna repeat the 1980s, slowly. The banks that build a position on both, with a clear point of view on the settlement architecture between them, are running the 2026 version of what Citi and Chase were running in 1985.

Ok, so what should banks actually do?

The practical implication is pretty unglamorous. Every bank with a corporate franchise needs a tokenized deposit program, because that is what defends the client deposit relationship when the client starts programming their treasury. Every bank that services onchain counterparties, or that wants to service them, needs a stablecoin capability, either directly by issuing one, by partnering with an issuer, or by providing the bank rails that support a stablecoin issuer. Those are not competing programs. They are two capabilities in the same payments stack.

Above both, the bank needs a settlement layer that can move a client between the two instruments, reconcile across networks, and do it with the same legal and operational finality it can deliver today on a wire. That is the orchestration work. It is also where most bank programs are still, at best, a couple years behind.

I really think the binary framing is a distraction. It lets people think the strategic question is which instrument to back. I believe the actual strategic question is whether you have a position on the settlement fabric, and whether you are helping to build it or letting someone else build it on top of your balance sheet?

The two tier system is not going away. It is gonna get rebuilt in tokens. Feels like the banks that understand that are hiring for it. The banks that are still debating about which one to pick are the banks that are going to have to settle for running one rail inside somebody else’s orchestration layer.

References

The two tier monetary system and BIS framing

Stablecoin market data

GENIUS Act and US stablecoin regulation

JPMorgan Kinexys and JPMD

HSBC tokenized deposit service

Citi, DBS, Standard Chartered, Deutsche Bank, and Partior

Cari Network (US regional bank tokenized deposit consortium)

Project Agorá and the settlement fabric

The historical analogy of regulation Q and the money market fund era

Atomic Settlement

Discussion about this post

Ready for more?