Virtual Cards in B2B Payments: Interchange, Rebates and Suppliers

An accounts payable clerk at a hospital network in Ohio approves a $50,000 invoice from a linen supplier. Instead of printing a cheque, her payment platform mints a sixteen-digit card number that can be charged once, for exactly that amount, by exactly that supplier, before it expires in a month. The supplier's receivables team keys it into a terminal. The hospital earns a rebate. The linen company pays a fee it may never have agreed to in writing. That three-way bargain is the whole story of B2B virtual cards, and almost every argument about them comes down to who ends up holding the cost.

What a B2B virtual card actually is

A virtual card is a commercial card account with no plastic, issued programmatically by a bank to a business buyer. In accounts payable, the common form is the single-use account: a number generated for one payment, with a hard limit, a merchant lock and a short expiry. The buyer funds it from its credit line with the issuing bank, and pays the bank later, usually on a monthly cycle.

That last detail matters more than the technology. A buyer that pays a supplier by card on day one and settles with its bank on day 30 or later has gained working capital that an ACH transfer never gives it. The supplier, meanwhile, is paid in roughly two business days through the card network, which is often faster than the buyer's normal terms. Both sides gain on timing. The price of that gain is interchange, which the supplier pays through its acquiring bank.

Virtual cards also solve a fraud problem that has dogged paper payables for decades. A stolen cheque can be washed and re-presented. A single-use card number that has already been charged for its full limit is worthless to a thief. That control, rather than rebates, is often what gets treasury teams to sign off.

The market in numbers

Juniper Research, the analyst firm most often quoted on the category, put global virtual card transaction value at about $5.2 trillion in 2025 and projected $17.4 trillion by 2029, according to FinTech Global's summary of the study. An earlier Juniper forecast, reported by IBS Intelligence in 2021, made the telling point about shape: B2B would be about 71% of virtual card value in 2026 but under 1% of transaction count. Business card payments are rare and large; consumer ones are frequent and small.

Treat those forecasts as directional. Analyst houses define "virtual card" differently, and some include consumer tokenised cards stored in wallets. The harder data comes from the networks and from US payment surveys.

Visa reported $1.8 trillion of commercial payments volume in fiscal 2025, up 7% in constant dollars, with revenue from its commercial and money movement business up 14%, per MediaNama's breakdown of the results. That figure covers all commercial cards, plastic and virtual, but it gives a sense of scale against Visa's total payments volume of about $14 trillion.

The incumbent virtual cards compete with is still the paper cheque. The Association for Financial Professionals found that 26% of US B2B payments were made by cheque in 2025, down from 81% in 2004, according to Nacha's write-up of the AFP survey. The bigger winner was ACH: B2B ACH volume rose from 2.9 billion payments in 2015 to 7.4 billion in 2024, worth $58.2 trillion. Cards are fighting for the slice between those two rails.

Where the money comes from: interchange

Interchange is the fee the supplier's bank pays the buyer's bank on every card transaction. In the US, Visa and Mastercard publish the schedules, and commercial cards sit at the top of them because issuers fund rebates and credit lines from the proceeds. The European Union's interchange caps, by contrast, do not apply to most commercial cards, which is one reason the economics look similar on both sides of the Atlantic despite very different consumer rules.

Visa's US schedule effective April 18, 2026 shows the spread. A standard card-not-present purchase on a purchasing or corporate card costs 2.70% plus 10 cents. Send full line-item data and the transaction can qualify for Commercial Product 3 at 1.75% plus 10 cents. Large invoices paid card-not-present on purchasing cards drop sharply under Visa's Large Purchase Advantage tiers.

Interchange only. Suppliers also pay network assessments and their acquirer's markup, which vary by contract and are not shown.

The same invoice can therefore cost a supplier anywhere from about $350 to nearly $1,500, depending on card type, how it is keyed and what data travels with it. A supplier that does not know which bucket its payments land in is almost certainly paying for the top rows.

The data rules changed in 2025 and 2026

Visa rebuilt its enhanced-data regime over the past eighteen months. Stripe's guide to the Commercial Enhanced Data Program lays out the timeline: a new scheme fee on Level 2, Level 3 and large-ticket transactions from April 2025, Level 3 replaced by CEDP and Product 3 rates in October 2025, higher Product 3 and Level 2 rates for business credit cards from January 2026, and Level 2 retired entirely in April 2026. Visa now checks whether line-item data is meaningful, not merely present. Suppliers that stuff "goods" into every product field risk losing the lower rate after the fact.

For suppliers, the practical effect is that cheap card acceptance now depends on clean invoice data. For AP platforms, it means the integration work that pulls line items out of an ERP has a direct dollar value.

Rebates: the buyer's side of the ledger

Issuers share part of the interchange with buyers as a rebate, typically scaled to annual spend and how fast the buyer repays. Commercial rebate schedules are private, but one large program publishes its results. The US government's GSA SmartPay cards carried $39.4 billion of spend in fiscal 2025 across more than 82 million transactions and 4.2 million accounts, and agencies earned $471 million in refunds, according to GSA's April 2026 release. That works out to roughly 1.2% of spend returned to the buyer. GSA's own statistics page warns against using its figures to calculate basis points for any one agency, so read 1.2% as a programme-wide average for a very large, very creditworthy buyer, not a benchmark for a mid-sized company.

Every rebate dollar a buyer books is a fee some supplier paid, usually without negotiating it.

This is why virtual card programmes are sold to CFOs as revenue lines. A company moving $100 million of supplier spend to cards at a 1% rebate earns $1 million a year for changing a payment method. It is also why suppliers push back.

Worked example: one invoice, three ways

Take the hospital's $50,000 linen invoice, paid on 30-day terms, and assume the buyer's bank shares a rebate of 1.0% of spend.

1. ACH on day 30. The supplier waits a month and pays a bank transfer fee measured in cents. The buyer earns nothing and keeps its cash for 30 days.
2. Virtual card on day 1, keyed without data. Under the commercial card-not-present rate the supplier's interchange is $1,350.10, so it nets about $48,650 before acquirer fees, but gets paid roughly four weeks sooner. The buyer books a $500 rebate and does not pay its card bill until the cycle closes, gaining another month of float.
3. Virtual card on day 1, qualifying for Large Purchase Advantage. Interchange falls to $352.50. The supplier keeps about $998 more than in case 2. The issuer has far less to share, so a buyer's rebate on large-ticket spend is usually lower.

If the supplier's own cost of short-term borrowing is around 8% a year, getting $50,000 a month early is worth roughly $330. Against a $1,350 fee the card loses; against $352.50 it roughly breaks even. That arithmetic, not marketing, is what decides whether a supplier says yes.

Supplier acceptance is the bottleneck

Buyers cannot force cards on suppliers. AvidXchange, one of the largest US AP networks, says in its 2024 annual report that suppliers choose card, ACH or cheque according to their own rules and perceived value, that it does not mandate a method, and that preferences can change at any time. The same filing shows the scale of the network: more than 8,500 buyer businesses and over 1.35 million suppliers paid in five years, with about $258.9 million of 2024 revenue coming from its card and direct ACH products combined.

The friction points are consistent across the industry:

• Manual keying. Many suppliers receive the card number by email and type it into a virtual terminal, which costs staff time and lands in higher interchange tiers.
• Thin-margin sectors. Distributors, construction subcontractors and freight firms working on 3% to 5% margins see a 2.7% fee as half their profit.
• Surcharging. Some suppliers add a card surcharge where law and network rules permit, which erodes the buyer's rebate and sours the relationship.
• Reconciliation. A card payment arrives in a merchant settlement batch, not tagged to an invoice, unless the AP platform sends remittance data alongside it.

The vendors that win suppliers over tend to offer "straight-through" processing, where the platform charges the card on the supplier's behalf and posts remittance detail automatically, and large-ticket rates that make the fee comparable to the value of early cash.

How virtual cards compare with the other B2B rails

Cards are not the cheapest way to move a business payment and were never meant to be. Their case rests on control, float and data. ACH and same-day ACH are cheaper for the supplier and dominate recurring, high-value domestic flows. Real-time rails such as FedNow and RTP win on speed and certainty but carry no credit. Cross-border, the calculus changes again, because a card settles in the supplier's currency at network FX rates, and the alternatives include correspondent wires and, increasingly, stablecoin settlement. Our pieces on cross-border B2B payments and stablecoin payments for business cover those trade-offs, and this primer on payment rails sets out the basics.

For buyers, a sensible programme segments suppliers: cards for the long tail of small, frequent invoices and suppliers already set up for card acceptance; ACH for strategic suppliers with thin margins; real-time rails where speed is worth paying for. Programmes that try to push every supplier onto cards tend to stall.

What to watch next

• Visa CEDP enforcement through 2026. With Level 2 gone since April 18, 2026, watch whether Visa resumes verifying new merchants for the programme, which it paused in November 2025, and how many suppliers lose Product 3 rates on data-quality grounds.
• GSA SmartPay 4. GSA's request for information closed on June 19, 2026. The next-generation contract will show how the largest public card buyer in the world thinks about virtual accounts, data and refunds.
• US interchange litigation and legislation. Any move on the Credit Card Competition Act or merchant settlements could reach commercial cards, even though the debate centres on consumer cards.
• Stablecoin-linked commercial cards. Visa reported 130 stablecoin-linked card programmes across 40 countries in fiscal 2025. If card networks settle in stablecoins with issuers, as we cover in this explainer, cross-border virtual card economics could shift.

Virtual cards will keep growing because they give buyers something real: fraud control, working capital and a revenue line. The category's ceiling is set by suppliers, and the suppliers that matter most are the ones who have done the maths in the worked example above. Buyers and suppliers weighing a programme should model their own interchange tier and financing cost, and take advice from a qualified adviser before changing terms.

About this researchMuffinPay InsightsIndependent analysis of stablecoins, CBDCs, cross-border payments and digital public infrastructure. Every figure is linked to its source. Read our editorial standards →