🏦 Friction as the franchise

No customer is going to ask for tokenized deposits, but the demand is arriving anyway, through a door most banks aren't watching.

Jul 19, 2026

In the last month the Association for Financial Professionals, the body that certifies corporate treasurers, launched a professional certificate in stablecoins and onchain liquidity. It was built with Kyriba, the treasury management platform, and it counts toward CTP recertification. The institution that teaches treasurers how to be treasurers has decided that moving money in tokenized dollars is now part of the job.

Its worth pausing here, because no bank customer asked for this. And it’s the thread I want to pull on, because customers never ask for rails. They didn’t ask for ACH in 1972 and they didn’t ask for SWIFT in 1977. What customers have always asked for is a small set of jobs that haven’t really changed in a century. They want to pay at the moment the work is verifiably done, keep cash earning until the moment it’s needed, know what landed on the other side of a border and when, move money between their own entities at any hour, and if you could, please make reconciliation do itself.

The thing to remember is that legacy rails have made those jobs expensive to do well, and while we’re at it, let’s just say it for those at the back, banks earned the spread on that expense. The float on idle cash, the FX margin on cross border payments, the intraday credit on trapped liquidity. All that friction is the franchise.

The way I think about it, that single fact explains almost everything about the order in which this market is moving. A bank that solves these jobs is in fact cannibalizing its own spread. A non bank that solves them takes the relationship. So the solving is being done by non banks first, the largest banks second, and everyone else, well it looks like they’re waiting for a customer request that isn’t going to arrive in a form they recognize.

Start with the jobs, not the technology

I believe tokenization is really interesting for one particular structural reason and that is it puts value with settlement finality and data in the same object. That sounds abstract until you map it against why B2B payments have resisted thirty years of point solutions. Erin McCune, who has spent a career in the trenches of B2B payments, argues the field doesn’t have a problem, it has twenty three, clustering into six main categories. Interoperability, trust, data, cost, working capital, and manual process. I really like her heuristic that a vendor that fixes one category leaves the other five untouched, and the buyer has no reason to switch, which is why the most impactful interventions address several categories at once and also why they’re so rare. A primitive that carries finality and remittance data in the same token cuts across four of the six at once, which is why I think this wave behaves differently from the e-invoicing and supply chain finance waves that came before it.

Run the old jobs through that primitive and each one becomes a product somebody can ship. Conditional payment, where funds release on verified delivery, is being piloted in freight right now, TCS and PayPal announced in March that PYUSD will settle trucking invoice flows they expect to exceed $1 billion this year. Continuous treasury, where balances sweep automatically between transaction accounts and tokenized money market funds, ends the idle balance as a concept. And with net interest income making up roughly 60 percent of retail bank revenue globally, that is an income statement event, not a feature. Cross border settlement with a stable value leg removes pre-funding. The self reconciling payment already exists, SAP’s Digital Currency Hub pays supplier invoices in USDC or PYUSD straight from the ERP payment run and hands back a camt.053 statement, so the payment arrives looking like a bank statement and books itself. PayPal paid an EY invoice this way back in September 2024!

None of this creates new customer needs. Think of it more as removing the excuse for not meeting the old ones.

The demand arrives through vendors, not through customers

I’ve talked to many bankers who say their customers aren’t asking for this. Honestly, I think bankers who report no client demand are watching the wrong door. A corporate treasurer is not going to walk into a branch and ask about tokenized deposits. That treasurer is going to adopt these capabilities inside the tools they already use, often without registering that anything monetary has changed.

In April, Kyriba and Circle embedded USDC execution directly into the treasury management system, agent orchestrated, settling cross border and intercompany payments around the clock, routed through the customer’s existing approval workflows and audit controls. It ships to customers this month, the same month as that AFP certificate. Ripple paid $1 billion for GTreasury in October, a treasury platform that processed roughly $12.5 trillion in payment volume in 2025, and has already added digital asset accounts to it. SAP is doing the same thing at the ERP layer.

Look at what this means from the corporate customer’s seat. Instant supplier settlement shows up as a feature of the ERP. The 24/7 sweep belongs to the treasury system, the instant payout to the payroll platform. Nothing in that experience says tokenized deposit, nothing says crypto, and nothing involves asking the bank for permission. I reckon every one of those features will move a balance or a fee from a bank. By the time the change is apparent to the average banker, those flows will already be well on the way to moving.

Charters second

The second group through the door is buying its way into the regulated perimeter, and the pace has changed character. The OCC received 18 de novo and conversion applications in 2025, nearly as many as the previous four years combined. On a single day in December, the OCC granted conditional national trust charters to Circle, Ripple, Paxos, BitGo, and Fidelity Digital Assets. Anchorage, which has held a national charter since 2021, now issues Tether’s US regulated stablecoin. Kraken got a limited purpose Fed master account in March, the first crypto firm with direct access to Fed payment rails. Erebor opened in February as the first de novo national bank of this cycle with roughly $625 million of capital. And in May, SoFi launched SoFiUSD, the first stablecoin issued by a US national bank inside a consumer app, in front of about 15 million members.

The May executive order asking the Fed to decide completed account applications within 90 days, followed a day later by the Fed’s own proposal for a limited purpose payment account tier, turned what was a queue into a conveyor. Five years ago Anchorage was the only firm that had assembled charter, regulatory status, and a path to settlement access from outside the banking system. Now it’s a pattern with a playbook.

Banks last, and only the largest

The largest banks are not ignoring this. They’re rebuilding the rails privately, for themselves. JPMorgan’s Kinexys has settled more than $1.5 trillion cumulatively and its deposit token now lives on Base, a public chain, restricted to institutional clients. Citi runs Token Services. BNY launched tokenized deposits in January with ICE, Citadel Securities, and Circle among the first users. Goldman is spinning its digital asset platform out into industry ownership. And in the clearest signal yet, JPMorgan, Bank of America, Citi, and Wells Fargo, along with a dozen others, are building a shared tokenized deposit network operated by The Clearing House, targeting the first half of 2027.

Two details in that announcement deserve more attention than they got. The first is that Bank of America’s head of payments conceded that clients are “not beating down the door” for tokenized deposits. The network is being built anyway, ahead of demand, I suspect because the people building it understand the vendor dynamic described above. The second is scale. Tokenized deposit flows at the largest banks are already estimated above $4 trillion a year, against roughly $300 billion of stablecoins outstanding. The volume story is bank money moving onto new rails, concentrated at the top, not crypto displacing anything.

It feels to me that there’s an unsolved seam running through all of it. A JPMorgan token and a Citi token are claims on different balance sheets. The network design does not yet specify how one bank’s token becomes another bank’s token at par, and nobody owns the liquidity that has to stand behind that exchange. The two tier monetary system solved this a century ago with reserves at the central bank. The tokenized version hasn’t solved it yet, and whoever does will own the most valuable piece of the new plumbing. Hold that thought.

The cascade to the long tail

Now lets follow the thread to its end. There are about 4,300 banks in the United States and roughly 3,900 of them are community banks. Their deposit economics have been moving before any of this shipped. Non interest bearing deposits at US banks fell from $5.5 trillion at the March 2022 peak to $3.85 trillion three years later, a 30 percent decline that happened before stablecoins paid anyone anything. The friction was already eroding and now I think tokenization will just remove what’s left of it.

In May the FDIC published a forensic study of the 2023 runs built from the actual core deposit and wire systems the agency seized at SVB, Signature, and First Republic. The headline speed is bad enough, two banks losing roughly half their deposits in three business days. The finding that should worry every asset liability committee sits lower down, the depositors who ran emptied their business operations accounts completely, the category every ALM model treats as sticky, leaving little to nothing behind. At Signature, escrow style balances that beneficial owners could withdraw on demand fell 83 percent in two business days.

The stickiness was never really about loyalty. It was actually friction, and the FDIC has now documented what happens when frictionless redemption meets a reason to move.

Where do those deposits go when they leave? The New York Fed’s February staff report on stablecoin disintermediation traces it, deposits that migrate to stablecoins re-pool at a small number of partner banks that hold them as reserves rather than lending them. A separate Fed Board note from December was even more blunt, deposit reliant community banks facing this substitution may be forced to contract lending more sharply than their larger peers. The credit contraction lands in the communities the long tail funds.

And the long tail’s tokenization roadmap is whatever its core vendor ships. Here the news is a bit grim for duller reasons. The combined market value of Fiserv, FIS, and Jack Henry has fallen from roughly $140 billion two years ago to about $60 billion today. Fiserv’s FIUSD stablecoin, announced a year ago, is now slated for July. FIS has made some moves with a digital currency platform that has completed seven proofs of concept. Jack Henry is routing the problem to a third party integration. None of it really feels like its production infrastructure a community bank can deploy today, and the vendors selling the roadmap are fighting for their own lives.

So yes, the signal arrives in a form legacy thinking struggles to parse. The community bank CEO is waiting for a customer to walk in and ask about tokenized deposits. The actual request arrives as a commercial customer’s ERP offering instant supplier settlement, a treasury platform sweeping the operating balance at 4:59pm on a Friday, a payroll provider offering instant tips. Each one looks like a vendor feature. None of this looks like crypto. Every one moves a balance or a fee, and by the time that demand is understood, the deposit will have moved and the relationship has a new front end.

Receive before issue

The market structure now forming does have a familiar shape. The Clearing House network gives the largest banks a shared rail they own. The regional consortia, like the Cari Network with its c$800 billion of combined assets, give the upper middle a hedge. The default outcome for everyone else is access to rails owned by their largest competitors, on their competitors’ terms. We have run this experiment before. RTP and Zelle are owned by the big banks, and the community banks answered by going to FedNow, which now has over 1,700 participating institutions, rather than clear over a rival’s infrastructure. I expect the same politics to play out here, and I think the question of whether the long tail gets a neutral settlement asset, rather than settling over a competitor’s balance sheet, will become one of the defining fights of the next three years.

But we all know politics is slow and the flows are not, so the practical question is what a bank below $10 billion in assets does now, before any of that resolves. I think the answer is unglamorous but within reach, you have to receive before you issue. Stand up a wallet. Accept the inbound stablecoin and token flows that are already arriving from your customers’ platforms, earn the conversion and FX revenue at the point of receipt, and keep the redemption function, because converting tokenized value back into spendable bank money is a banking function whoever moves the asset. A sidecar ledger next to the existing core makes this a bounded project rather than a core replacement, which matters if your core vendor’s roadmap is a press release. Issuance, if it ever makes sense for a $2 billion bank, comes after acceptance, and probably only ever through something shared.

So if you run a bank and you’re still waiting for the customer to walk in and ask for this, know that waiting is itself a decision. Everyone upstream of you is moving, the vendor are shipping features, there are new charters, the largest banks are building a network they own between them, and each of them, as it moves, will take balances that you used to earn on. The only choice is whether what replaces the friction runs over your rails or over somebody else’s, and I believe that choice is probably on a clock.

I do think the decision gets easier to make once we’re honest about what the spread always was. The spread was never a service the bank sold, it was the price of doing an old job the slow way, and the customer only paid it because there was no faster road. There is a faster road emerging now.

I really doubt the friction is coming back once it’s gone. The jobs were never really the bank’s to keep, only to do.

References

B2B payments framing

Erin McCune, Forte Fintech Field Notes: B2B Payments Don’t Have a Problem. They Have Twenty-Three. (April 2026)

Vendor and treasury layer

Charters and the regulated perimeter

Bank rails

Deposit economics and the long tail

Bill McCahey

Jul 20

This contains a structural rhyme worth noting: a tokenized deposit is only money on the balance sheet that issued it, and becomes a foreign currency once it leaves. That’s the absence of a portable obligation — and if obligations can’t travel, par can’t hold. It's the same limitation that creates liquidity islands everywhere else.

Atomic Settlement

Discussion about this post

Ready for more?