Crypto payment processing fees just reset — and the percentage was never the rail's cost

What changed on July 31, 2026

Yesterday a promotional window closed. It closed quietly, without a press release, because closing was always the plan.

When PayPal announced Pay with Crypto on July 28, 2025, it gave US merchants the ability to accept more than 100 cryptocurrencies — BTC, ETH, USDT, USDC, XRP, BNB, SOL and roughly 94 others — from wallets including Coinbase, Binance, Kraken, MetaMask and Phantom, with near-instant conversion into fiat or PYUSD. The headline number was a 0.99% transaction rate, pitched against cross-border card acceptance that routinely runs 1.5%–3.5%. Coverage at the time framed it as cutting international acceptance costs by up to 90%.

Read the fine print of that same announcement and one line does the real work: 0.99% represents the Pay with Crypto transaction rate until July 31, 2026. Not "our rate." Not "our rate, subject to review." A date. Reporting across the following twelve months consistently described what came next as a step to 1.5%.

That is still competitive — 1.5% beats most cross-border card acceptance, and nobody should pretend otherwise. But the shape of the event matters more than the number. A merchant who built a checkout around 0.99% in August 2025 did not renegotiate anything, did not receive a rate review, and did not get a vote. Their cost per crypto sale rose by roughly half, on schedule, because the calendar said so.

This is the third time in five weeks that the same lesson has arrived in a different costume. On July 30, Revolut finished delisting USDT for European users — a balance customers thought was theirs, redenominated on a date set by someone else. On July 15, Stripe and Advent bid $53 billion for PayPal, a deal we covered as payment consolidation and what it does to merchant choice. Now the price of a crypto acceptance product resets on its own schedule. Different mechanisms; identical structure. Whoever holds the position in the middle sets the terms of the position in the middle.

What a 51-basis-point step-up actually costs

Half a percent sounds like rounding. It isn't, because it compounds against volume rather than against margin.

Here is the arithmetic on the difference between 0.99% and 1.5%, before anything else in the stack:

For a business running on a 12% net margin, the $1M row means about $42,500 of additional revenue needed just to stand still. For an importer or an exporter clearing $20M in cross-border settlements, it's a six-figure line item that appeared without a meeting.

The second cost is harder to put in a table: you now know your acceptance price is a variable someone else controls. That changes what you can plan. You cannot quote a distributor a landed cost eighteen months out on a fee that resets on a schedule you don't hold. You cannot build a thin-margin cross-border corridor on a rate whose next revision date is unpublished. Pricing certainty is itself a product feature, and a promotional rate is the explicit absence of it.

There's a related exposure sitting one layer down. The same product lets merchants hold proceeds as PYUSD and earn a stated 4% — a rate PayPal itself flags as subject to change. That number isn't just a business decision; it's currently the subject of an active legislative fight in Washington over whether platforms may pay rewards on stablecoin balances at all, which we broke down in the CLARITY Act and stablecoin yield. A merchant treating that 4% as part of their economics is holding a position exposed to a Senate calendar as well as a pricing committee.

Why a percentage fee has nothing to do with the blockchain

Card pricing is a percentage for a reason, and the reason is risk. An acquirer fronting a $5,000 transaction carries five thousand dollars of chargeback exposure for up to 120 days; one fronting $50 carries fifty. Interchange scales with ticket size because the liability scales with ticket size. Percentage pricing is an honest expression of card economics.

None of that is true on-chain, and this is the part the market has not repriced yet.

A USDC transfer on Base costs the same whether it moves $50 or $5,000,000. Confirmation takes roughly two seconds either way. On Solana it's around 0.4 seconds; on Polygon, about two. The transfer is final on confirmation — there is no reversal window, no representment, no 120-day tail, so there is no size-dependent liability for anyone to price. Flagship Advisory Partners put it plainly in their April 2026 study of stablecoin merchant acceptance: core stablecoin payment processing "can cost very little compared to traditional rails — a fraction of a penny versus a few cents for a bank transfer."

So when a crypto acceptance product quotes you 1.5% of the sale, essentially none of that is the rail. It's the intermediary.

That same study is worth reading for scale, because it explains why pricing is still soft everywhere. Direct stablecoin acceptance sits at 4% of top-50 US merchants, 8% of large European merchants, and 12% of large Latin American merchants — early enough that most pricing in the category is customer-acquisition pricing rather than steady-state pricing. Flagship's observed band for stablecoin PSPs is 0.8%–1.4%, against 0.5%–2.5% for debit and 1.8%–4% for credit. A move to 1.5% doesn't just raise a merchant's cost; it steps outside the band the rest of the category is currently occupying.

Expect more of these. Introductory rates across crypto acceptance were set during a land-grab, and land-grab pricing has an expiry date by construction. The useful question for a merchant in August 2026 is not "which provider is cheapest today." It's "whose decision is my cost per sale?"

What that 1.5% is actually buying: custody and conversion

Follow the money through a custodial crypto checkout and the fee stops looking arbitrary.

Your customer sends USDC. It does not go to you. It goes to the provider's wallet. The provider then holds a crypto position, quotes an internal conversion, sells or nets it against inventory, absorbs the price movement between quote and execution, meets its own compliance and reserve obligations on the balance it's carrying, and finally credits you dollars or PYUSD out of its own balance sheet. Somewhere in that chain there's a bank relationship, a treasury desk and a licence, all of which cost money to keep.

That is a genuine service, and it is genuinely expensive to run. It is also, structurally, the only part of the flow that behaves like a percentage. You are not paying 1.5% to move value across a blockchain. You are paying 1.5% for someone to stand between you and the blockchain, take the asset, and give you a different one.

It follows that the fee has a second cost attached, and this one doesn't show up on an invoice: while the provider holds the position, the position is theirs, not yours. Which is precisely the exposure European Revolut customers discovered when their USDT balances were scheduled for automatic conversion at a rate and on a date chosen for them.

How Payzum's model changes the question

Payzum is a non-custodial crypto payment processor. The distinction is not a marketing adjective — it's a description of where the money physically goes.

When a customer pays through Payzum, the funds move from the customer's wallet to a wallet the merchant controls. Payzum never takes possession, never pools merchant funds, never carries a balance on the merchant's behalf. There is no "Payzum balance" to convert, freeze, redenominate or reprice, because the settlement is the payment. On-chain finality means the money that arrives is money that arrived — no chargebacks, no reversal window, no 120-day tail.

Because there's no custody, there's no conversion spread to fund, and the economics stop being a percentage of your revenue and start being a cost of running software. Network fees are cents on Base, Polygon, Solana, Arbitrum, Optimism, BNB Chain and Avalanche.

Volatility, which is the other honest reason merchants accept a conversion middleman, gets handled differently: Payzum offers optional auto-conversion to USDC or USDT. A customer can pay in whatever they hold; you can settle in stablecoins. The difference from the custodial model is who decides — the merchant chooses whether to convert and holds the resulting asset in their own wallet, rather than receiving whatever the intermediary hands over after taking its cut.

Now the honest limitation, because a fee comparison that hides it isn't worth reading. Payzum is crypto-only. It does not settle to a bank account. If your business needs dollars in a bank on Monday morning and has no appetite for holding a stablecoin balance, a custodial provider is doing something Payzum does not do, and 1.5% is what that service costs. The right comparison isn't "1.5% versus cents." It's "1.5% for acceptance plus conversion plus fiat settlement, versus cents for acceptance settled in stablecoins to your own wallet, with the fiat leg handled on your terms, by your own off-ramp, at a time you choose." For a growing number of businesses — cross-border sellers, exporters, agencies with overseas clients, anyone already holding working capital in USDT — that second shape is simply better. For others it isn't. Both statements can be true.

How it works, step by step

1. Connect the wallet you already control. You register the receiving address per chain. That address is yours — hardware wallet, multisig, treasury wallet, whatever your policy requires. Payzum has no key and no claim.
2. Pick how you want to get paid. Hosted checkout (redirect, modal or inline), no-code payment links and buttons, invoices with expiration and overpayment detection, recurring subscriptions, a donation/tip-jar page, or a POS flow with a fresh QR per sale, physical terminals and PIN-protected cashier accounts.
3. Set your conversion policy. Accept the coins your customers actually pay with, then decide whether to auto-convert to USDC/USDT. You choose, per your treasury policy, not per the provider's rate card.
4. Reconcile from signed webhooks. Every payment fires a signed webhook into your stack, with a REST API, API keys, an integration playground, 2FA, encrypted secrets and a full audit log behind it. Confirmation runs in seconds — Solana ~0.4s, Base ~2s, Polygon ~2s — and the money is already in your wallet when the notification lands.

Where the difference shows up

Fee structure is abstract until it hits a real business. Three that feel it immediately:

• A cross-border B2B supplier invoicing $80,000 a month. The step from 0.99% to 1.5% is $408 a month — but the sharper issue is the invoice itself. Quoting a distributor a landed cost for the next four quarters requires a fee you can commit to. Settled non-custodially, the buyer pays USDT against an invoice with expiration and overpayment detection, and the funds are in the supplier's wallet in seconds. The cost per invoice doesn't move when a promo calendar turns over.
• A digital agency with clients in five countries. Retainers are large, monthly and predictable — exactly the shape where a percentage fee hurts most, because a $12,000 retainer costs 240× what a $50 one does to process, for identical work on-chain. Recurring subscriptions settled to the agency's own wallet turn a revenue-linked fee into a flat operational cost. See accepting USDC payments online for the mechanics.
• A retailer taking crypto at the counter. In-person, a percentage of the sale is the exact thing the merchant was trying to escape by leaving cards. A POS flow that generates a new QR per sale, runs on any phone, gives each cashier a PIN, and settles straight to the owner's wallet with no chargebacks makes the acceptance cost independent of basket size — which is what the rail was always capable of.
• An API provider selling to AI agents. This is the reductio of percentage pricing: when the average agentic payment is measured in cents, a percentage-of-sale model collapses. Payzum's x402 middleware sits in front of an existing endpoint — you configure the URL, your API key and a price, and Payzum publishes an x402 URL, returns the 402, settles through an external facilitator (currently Coinbase's) and proxies the paid call through. Agents pay USDC on Base, per call, straight to your wallet. No code, no protocol to implement.

Custodial crypto checkout vs non-custodial settlement

Objections worth taking seriously

"1.5% is still cheaper than my card acceptance. Why would I change anything?"

You might not, and if crypto is 2% of your volume it's probably not worth a project. The argument isn't that 1.5% is expensive in absolute terms — it's that it's a price you don't set, on a rail whose actual cost doesn't scale with your ticket size, attached to a product that holds your money until payout. If crypto is becoming a meaningful share of your revenue, the question shifts from "is this rate good" to "do I want this to be a line item someone else revises." Our breakdown of card fees versus stablecoin acceptance covers the card side of the comparison.

"I don't want to hold crypto. Isn't that the whole point of a conversion provider?"

It's a fair reason to use one, and we said so above. But notice the two things being bundled: not holding volatile assets and not holding your own assets. Only the first is a real requirement. Auto-conversion to USDC or USDT solves volatility — a dollar-denominated stablecoin in your own wallet isn't a crypto position in the sense you're worried about. What the custodial model adds on top is that the balance sits somewhere else until someone else releases it, and that part isn't buying you anything.

"Managing wallets sounds like operational risk I don't have staff for."

It's the honest objection, and it's a real trade: you're exchanging counterparty risk for key-management responsibility. For most businesses that means one hardware wallet or a multisig with existing signing policy, plus signed webhooks feeding the same reconciliation process finance already runs. It's a different discipline, not a heavier one — and it's the discipline that makes "nobody can reprice or freeze this" true rather than promised.

Sources: PayPal newsroom, "PayPal Drives Crypto Payments into the Mainstream" (July 28, 2025) for the Pay with Crypto product terms and the 0.99% rate through July 31, 2026; Flagship Advisory Partners, "6% and Counting: The State of Stablecoin Merchant Acceptance" (April 20, 2026) for acceptance rates and PSP pricing bands; CBS News for launch coverage. The post-promotional 1.5% figure reflects consistent reporting through 2026; merchants should confirm current pricing with their provider directly. This article is analysis, not legal, tax or financial advice.