During a recent conversation with a payments executive at one of the countryâs largest banks, they told me about the moment tokenization stopped being just some theoretical thing for them. A corporate client had repatriated money back into the US in a single day. Theyâd run it through stablecoins with a couple of hops, and explained how they would never use the bankâs wires again. They didnât talk about fees, actually what the client was blown away by was the money came home the same day.
It does make you wonder if the industry is having the wrong argument about all this. The standard stablecoin pitch says look ma, cheaper rails. Ran Goldi who runs payments at Fireblocks, which moves as much stablecoin volume as anyone, and here is what he says about cross border, the biggest stablecoin use case there is
âitâs a great use case not because stablecoin transfers are cheaper. They are not cheaper.â A cross border move still needs an FX leg. âItâs not going to be cheaper than Swift. But it will be way faster. And if you can find economic value in that speed, then that is the real value.â
A few weeks ago Wenxin Du of Harvard Business School stood up at Jackson Hole and made the same point. Her worked example sent $100 from the US to Europe through USDC, funded the exchange by ACH, paid the gas, converted to euros at a 1% trading fee, paid the SEPA charge out, and you arrive at "at least 219 basis points on a $100 transfer, excluding the U.S. wire fees," above what Wise or Remitly charge all in.
So we have a man whose business depends on stablecoin adoption and an economist skeptical of it that have just agreed on the headline claim of a thousand pitch decks, and really they're both right, because the cheaper rails pitch has always had a bit of a problem as the rail was never really the expensive part.
The rail was never the expensive part
Swiftâs reputation for slowness is probably a good fifteen years out of date. Its 2025 Spotlight on Speed data shows 75% of payments reaching the beneficiary institution within 10 minutes, and less than 20% of elapsed time spent in flight between countries. More than 80% of the time a payment takes is spent in the last mile, inside the receiving bank. The message gets there fast. And the fees, while real, are small against the flows. JP Morgan and Oliver Wyman put global cross border transaction charges at roughly $120 billion a year, on an average fee of $27 per transaction, excluding FX. Thatâs a big number and itâs a pretty genuine annoyance, but it is not where the money is.
Duâs decomposition of a single Wise transfer makes the point. A pound to euro payment costs the customer ÂŁ0.24 all in, of which customer support is ÂŁ0.11, compliance ÂŁ0.06, other servicing ÂŁ0.05, and âthe actual costs of moving moneyâ ÂŁ0.02. McKinseyâs August report reaches the similar conclusion. $190 trillion of annual cross border volume and more than $290 billion of revenue, the transfer itself is sliding into what they call a commodity trap. It feels like anyone who has looked closely agrees that the rail is the cheap bit.
The money is in everything wrapped around the rail. The same JP Morgan report notes itâs
ânot uncommon for payments to take 2-3 days to reach end beneficiaryâ
Not because the message is slow necessarily but because the money underneath it moves through pre-funded nostro accounts, cutoff times, batch windows, and compliance queues. To make the system work at all, banks park cash in advance, and its pretty staggering amounts with top tier Fedwire participants holding an average of $630 billion a day in intraday liquidity buffers over the decade to 2018. Industry estimates put the cost of carrying that buffer at something like $600 million a year for a single top tier bank.
The corporate version of this is the operating balance that sits idle because moving it is slow, and the pre-funded account across the border because settlement canât be trusted to arrive on time. Rain, which issues stablecoin settled cards, did the arithmetic for a program spending $1 million a day, and because ACH and Fedwire donât run nights, weekends or holidays, it needs $4 million sitting idle by Friday night ahead of a long weekend, and a non US issuer serving five markets runs five separate pots of it, each with its own currency and holiday calendar. I did the arithmetic on one version of this in The Operating Account Was Never Free and a treasurer moving $1 billion a day through four hours of settlement delay is lending the system roughly $6.7 million a year, for nothing.
So actually what weâre looking at is the fee is $27, but when you look under the covers, the wait is billions, parked, everywhere, all the time. It seems to me selling stablecoins as a cheaper fee focuses on the smallest number on the pageâŚ
Pricing the wait
When Malcom McLean put freight in steel boxes in 1956, the obvious saving was loading cost, which collapsed from nearly $6 a ton to 16 cents. But the fortune wasnât made on cheaper stevedores. It was made on what speed did to inventory. Ships that turned around in hours instead of weeks, goods that stopped sitting in warehouses at both ends, and eventually the just in time supply chain, which is a working capital revolution. The freight was the fee, and the inventory was the dividend.
Now letâs run the same logic through corporate treasury, because this is where I think the demand is actually going to come from. The treasurerâs job, for a century, has been managing latency. The vagaries of the AP and AR cycles mean the operating account has a balance in it because payments are slow and lumpy and the account has to absorb the mismatch. Atomic settlement changes the job description. When the payment and the exchange happen in the same instant, the treasurer can hold the balance in a tokenized money market fund and discharge it into cash at the moment of settlement, intraday, around the clock. Eric Saraniecki at Digital Asset calls the end state cash as a financial capacitor, charged in yield bearing form, discharged at the instant of need, and predicts that once conversion is instant,
âuniversally you will see that people sit in very little cash, very, very little cash.â
This is why I think the demand signal confuses so many bankers. The largest corporate clients pushing Citi and JP Morgan into tokenized cash are asking for this kind of liquidity in the system, and the token is really just the first instrument that delivers it. Nobody on a treasury desk wants a blockchain. The technology is really incidental to the buyer, which I reckon is usually how you can tell a technology is going to win.
The bill for velocity
Now theres a bill attached to all this because tokenized rails settle gross. The whole value moves with the payment. In the netted world, if I send you 100 bucks and you send me 95 bucks, its actually 5 bucks that moves at the end of the day. On an atomic rail, 100 moves and 95 moves, in full, in real time, which is how RTP already works and why RTP participants pre-fund. You can think of the netting window as a piece of financial engineering that manufactured liquidity out of patience, and now deleting the patience deletes the manufacturing. The dividend comes from money that never waits, and so the price is now liquidity that has to be ready all the time.
I think this an underpriced consequence and it cuts both ways. It creates a new cost center, the just in time liquidity function, which is why the DTCC is productizing collateral mobility and why every serious tokenization design ends up reinventing something that looks suspiciously like a clearing house. And it exposes a gap at the top of the system, because the commercial rails are going atomic while the Fed still runs on the Feds hours. Every private workaround for that gap, pre-funding, reserve pools, mutualized liquidity facilities, is basically somebodyâs new business and somebody elses new dependency. The plumbing bill doesnât cancel the dividend, the containerâs bill didnât either, ports had to be rebuilt around the box. But whoever ends up owning the liquidity function will have bought the best seat in the new system.
If cash becomes just in time, the pool of idle tokens stays small while the flow through them explodes, which is why Saraniecki doesnât buy the parabolic supply forecasts, and Iâm not sure I do either. Issuance becomes a throughput business, not an AUM business. On that measure the numbers may already be telling the story. Stablecoin transaction volumes ran to $33 trillion last year by Bloombergâs count, against roughly $300 billion outstanding. Maybe supply is the warehouse and velocity is the product.
New trips, not stolen passengers
So if the value is actually velocity, then the volume should show up where the wait is most expensive. If you think about Uber, it actually didnât just take rides away from taxi cabs. It introduced an entirely new category of rider. Stablecoins my well do the same with new corridors, new hours and new use cases. We may well see types of money movement that we canât imagine now. Unlocking thin corridors that correspondent banking barely serves. Or weekend settlement, which is why Citiâs clients move tokenized dollars for Saturday M&A closings. Or things like payments timed to a physical event, a container, a delivery scan, and a payroll moment, that batch systems structurally canât hit.
I expect the uncomfortable part for banks is that the repricing will arrive anyway. Iâm sure the fee pool survives this, banks will charge for everything, but Iâm not sure the float does, and the float has always been the big number. As a senior digital assets executive at one super regional bank mentioned to me that 80bps of their NIM is funded by non interest bearing deposits, and the bank now treats that as under threat. Their response wasnât a product launch, it was the first line of business the bank had ever funded without a business case, on the logic that a few million spent learning beats billions lost by waiting or doing nothing. Thatâs a very sensible posture, and Iâd note itâs the same one Jamie Dimon has held for a decade, skeptical in public, building in private. We all tell ourselves the community relationship is what makes deposits sticky, and thereâs truth in it, but what price a relationship, once getting yield on a balance stops requiring the customer to do anything at all?
For most banks the practical answer is think big, start small, and itâs the same one I gave for the long tail in Friction as the Franchise, that is you have to do enough to receive. So accept the inbound tokenized flows your customers platforms are already generating, keep the conversion and redemption revenue, and learn the plumbing on live money before deciding what to issue. The banks that treat velocity as a product to sell, pricing it against demurrage and pre-funded accounts and committed lines rather than against wire fees will, I expect, find corporates surprisingly willing to pay for time. The ones that keep defending the $27 may well will win the argument and lose the balance.
The shipping container didnât win because shipping got cheap. It won because a factory in Ohio could promise a customer in Rotterdam a delivery date and hold inventory measured in days, and once one competitor could do that, everyone had to. Tokenized money is the same offer made to cash, and I reckon the transfer fee will be the last thing anyone remembers about it.
References
The velocity argument
Ran Goldi (Fireblocks), Payments on Fire episode 282: âWhy You Need a Stablecoin Strategyâ
Eric Saraniecki (Digital Asset), Tokenized episode 89: cash as a financial capacitor
The cost of the wait, and the netting counter-argument
J.P. Morgan Ă Oliver Wyman, âUnlocking $120 Billion Value in Cross-Border Paymentsâ (2021) â $120B annual transaction charges excluding FX; $27 average fee; ânot uncommon for payments to take 2-3 days.â
BIS Working Paper 1089, âIntraday liquidity around the worldâ â the scale of intraday buffers; the $630B/day Fedwire average and ~$600M/yr per-participant cost estimates appear in DTCC and Finadiumâs 2026 collateral infrastructure work drawing on this data.
Marc Levinson, The Box (Princeton University Press) â loading costs from $5.86/ton to $0.16/ton; the inventory revolution.
The aggregates and the deposit franchise