AribaPay moved the money. Taulia tried to bring the card. SAP Pay now starts inside the ERP. The question is whether SAP can earn the economics of a payments company.
On October 6, SAP announced SAP Pay, an ERP-native payment service. It is the third chapter in SAP’s attempt to turn a software position into payment economics: AribaPay moved bank payments in 2013; Taulia pursued virtual cards after SAP bought it; SAP Pay now starts in the ERP.
These are different products, not three launches of the same one. The people and partners changed. The problem did not: owning the workflow does not automatically mean owning the payment economics.
I worked at Mastercard and spent eleven years building Finexio, embedding payments inside the AP and procurement platforms where the work was already happening. I also worked closely with executives who knew parts of SAP’s payments history firsthand. Here is the history as I understand it from that experience and the public record. The people who built these products know more, and I want to hear from them.
AribaPay: volume without much margin
SAP agreed to buy Ariba for $4.3 billion in 2012. At Ariba LIVE the next year, Ariba president Kevin Costello and Discover president Roger Hochschild unveiled AribaPay.
The design was sensible. A buyer approved an invoice in Ariba. Discover debited the buyer’s bank account and paid the supplier through bank rails, with remittance data attached to the transaction. Suppliers registered their account details once. No card acceptance was required.
There was a real customer problem in that design: the payment and the information about the payment often travel separately, leaving AP and AR teams to stitch them back together. But there was also an economic problem. Discover’s CFO Mark Graf described AribaPay to analysts in 2015 as “very low margin business.”
The volume was not imaginary. Discover said AribaPay generated $1.5 billion in the fourth quarter of 2014. In January 2016, SAP CEO Bill McDermott and First Data CEO Frank Bisignano announced an expansion into Europe and Latin America, and said more than $40 billion in payments had been made with AribaPay to date. These were serious people working on a real product.
Discover’s Network Partners volume, a broader reporting line that included AribaPay, reached $44.5 billion in 2022. Discover repeatedly named AribaPay as a driver of that line’s growth. Then it named lower AribaPay volume as a driver of the decline in 2023 and 2024. In the first quarter of 2025, Network Partners volume fell 73%, from $11.1 billion to $3.0 billion, after what Discover called “the anticipated exit of a partner.” Discover did not identify that partner. I cannot establish from the public record that it was AribaPay or that SAP formally shut AribaPay down.
What I can say is that AribaPay’s economics were thin from the start. AFP’s 2022 survey put the median cost of an ACH for the largest companies at 11 to 25 cents. There is not much to divide among a network, a processor and a software company when the underlying transaction sells for cents.
SAP announced further Ariba-related payment integrations. I have not found public customer counts or payment volumes for those arrangements. If they produced material scale, I would like to see it.
There was another seam inside the customer. Procurement bought Ariba. Treasury controlled the bank relationships, payment methods and cash. Hofler put it plainly in 2019: “Our sales DNA is procurement, so we’ve always had a little bit of a hard time really getting into that market.”
That is the kind of sentence only somebody doing the work says out loud.
Taulia: a card program with too many hands on the economics
SAP’s next move was its 2022 majority acquisition of Taulia, a working-capital business founded in 2009. Cédric Bru stayed on as CEO. Starting in 2023, Taulia announced virtual-card relationships with Mastercard, Visa and issuing and acceptance partners.
The names are impressive. The missing number is spend. I have not found a disclosed volume figure for Taulia Virtual Cards in those partner announcements. That does not prove the program has failed. It means outsiders cannot tell how much of the announced distribution became payments.
The structure matters more than the announcement count. Taulia described a “bring your own bank” model. The issuing bank supplies the credit. The network takes its share. A partner may help convert suppliers to card acceptance. The buyer expects a rebate. Taulia provides software and orchestration, but it does not capture the full economics of an issuer. The same split runs through the bring-your-own-card model now reaching enterprise software: the platform can win retention without becoming the issuer or capturing the interchange.
I do not think SAP could simply have become a bank. It did not need to. But there is a large difference between owning the customer experience and owning enough of the program, supplier enablement and ongoing service to make the economics work. A referral or thin integration does not get you there.
Supplier acceptance is hard, too. At Finexio, I saw how much work sat behind a card payment: a dedicated team had to speak with suppliers, resolve objections and keep them accepting after the first transaction. That was an operating function, not a button in the software. Taulia’s 2025 partnership with Boost was explicitly about automating that work. A virtual-card number in the product is not the same thing as a supplier who will take it, a bank that will fund it, and a finance team that can reconcile it without another spreadsheet.
SAP Pay: the workflow is closer to the money
The third chapter is SAP Pay, powered by Tereina, which SAP calls an SAP company. Bru leads Tereina while remaining Taulia’s CEO. SAP lists ACH, wires and checks among its payment methods, and announced cross-border settlement work with a Circle stablecoin partnership announced the following day.
This launch has two meaningful differences. First, execution begins inside the ERP, where the approved payment run already lives. The opportunity is not just moving money. It is carrying the invoice, approval, remittance and reconciliation data through to the end without making finance staff download a bank file, chase an exception and manually mark the invoice paid. Second, SAP identifies Tereina as the regulated payment-service provider behind SAP Pay. Its licenses and partner roles still need to be understood market by market.
SAP has a genuine advantage here. It sees the context around the payment before a bank sees the instruction. The question is whether it can turn that context into a service customers value enough to pay for.
Tereina says it aims to reduce payment total cost of ownership by 25%. That is a potentially meaningful claim. It is also impossible to evaluate without the denominator. If the comparison is a large company’s domestic ACH fee, the fee may run 11 to 25 cents, and its bank’s earnings credits may offset some or all of the cash charge. A quarter off that number will not change the CFO’s year.
If the denominator includes bank-file handling, payment exceptions, supplier inquiries, reconciliation labor and FX, that is a different case. SAP should show the before-and-after cost of a real payment file, including its own fees, implementation work, bank credits, card rebates and the people required to handle exceptions. That would be a useful number.
The product list still leaves me with a question: where is card? It is not in SAP Pay’s announced payment methods. SAP has a card effort in Taulia, but the new ERP-native payment product does not appear to join it to the rest of the payment run. Card is not right for every supplier, and it is not the only way to earn margin. But card and FX are two of the places where a payments business can create value beyond access to a low-cost domestic rail. Without them, the commercial case leans heavily on workflow savings.
Stablecoins may eventually help in specific cross-border corridors. The Circle announcement describes a path for eligible customers and further proof-of-value work, not established production scale. Before I would put that near the center of an enterprise payments strategy, I would want to know who handles conversion, compliance, liquidity, failed payouts and the supplier who just wants funds in its bank account.
The test for SAP
I would not write off SAP Pay because AribaPay’s margins were thin or because Taulia has not disclosed card volume. Starting inside the ERP is a better position than starting in a procurement network. But a better starting point is not a business model.
Three things would change my mind.
First, show live payment volume by method, not partnership announcements. Tell us how many customers are paying suppliers, what share of eligible spend converts to card or higher-value cross-border services, and what happens to the rest.
Second, show the net economics for all parties. What does the buyer save after bank credits and rebates? What does the supplier receive, when, and at what cost? What does SAP earn after paying the bank, network, processor and acceptance partners?
Third, show the operating service. Who fixes a failed payment? Who updates a bad supplier account? Who resolves a remittance mismatch? If the answer is still somebody in treasury moving a file between portals, the payment button has moved but the work has not. A bad supplier account is also where payment fraud often gets in.
The people who built AribaPay and Taulia’s card program were solving hard problems. Bob Solomon, who ran Ariba’s supplier network and financial services, called the AribaPay team “some really bright people.” I agree. This is not a story about bad people or a company that never saw the opportunity. It is a story about how difficult it is to turn software distribution into a payments business.
If you worked at SAP Ariba, Discover, First Data, Taulia or one of the banks, I would like your account. Who conceived AribaPay, and what happened after its early volume? Which bank partnerships went live at scale? How much spend is actually moving on Taulia Virtual Cards? What does SAP Pay do that the earlier products could not?
Correct my names, dates or conclusions where the evidence warrants it. I will update the piece.