How it works

How a program launches, step by step

Here's the payroll platform's path from idea to a cook using the card. The partners set the pace: bank due diligence, network approval and custom card production each take weeks to months, and finished code doesn't speed any of them up. Through an existing card-issuing platform, it could issue stock or virtual cards within days, because the platform did the bank and network work once and even its wallet approvals cascade down to its customers.

1. Model chosen: own BIN (you're the bank), a sponsor bank plus your own program management, or embedded issuing through a platform; and the card type, prepaid, debit, credit or commercial. I'd also decide here who holds the ledger, since it's the hardest thing to change later. What can go wrong: picking a bank only on its interchange, then finding it's under a regulator's order or close to $10 billion in assets.
2. Bank approved: the bank signs after due diligence and a program agreement. Expect to send two years of financials, insurance, your anti-money-laundering, identity, PCI and fraud policies, a business plan and a flow of funds. One processor's guide says sponsor banks look for $3-5 million of capital raised or a large user base. Visa requires the bank to review your senior officers and inspect you on site. Vendors put this step at one to six months; I haven't found a neutral source. What can go wrong: the bank declines because the program is outside its risk appetite (a prohibited business type, thin capital, a thin compliance team), or the term sheet carries minimums the program can't meet for a year.
3. BIN assigned: the bank registers the payroll platform with Visa as a Third Party Agent before it performs any service, assigns a BIN or, more often, an account range inside one of its own, files a Regulation II certification for debit and prepaid ranges, gets Visa's approval for a prepaid program and submits the card design. Visa approves designs for every form factor before production and again each time the design changes. Durations aren't published; in practice it's weeks inside the bank's timeline. What can go wrong: someone tweaks the card art after approval and the design goes back for review.
4. Cards ready: the processor certifies your integration while plastics and wallet tokens are prepared. Custom plastic takes 8-10 weeks with a 2,500-card minimum at Stripe and 90-120 days at Column. Apple grants the push-provisioning entitlement (adding the card to Apple Wallet from your app) per app, and it only works in distribution builds, so you test through TestFlight or the App Store. Google's review takes 24-48 hours, then comes an NDA, allowlisting, video test cases and field testing. What can go wrong: a chip shortage like 2021-23, which put up to a billion cards at risk; an "Add to Apple Wallet" button shown to cards that aren't eligible, which can get the app rejected.
5. Live: the earliest users come in as a pilot, then the program opens. I'd start small (for the payroll platform, its own staff and two restaurant groups) and widen once reconciliation has run clean for a few weeks. What can go wrong: the daily tie-out between the ledger and the bank's account fails in week one and nobody owns the gap.

Two exits sit off that path. Declined is the bank saying no at due diligence, usually because the program is outside its risk appetite. Migrated comes later, when a live program moves to a new bank or processor, covered below.

Virtual cards work the moment they're issued, so Stripe, among others, advises launching on stock or virtual cards while the custom plastic is made.

When a program moves

Programs move when the bank ends the relationship, when the bank gets into trouble with its regulator, or when the program outgrows its processor, and each move is a project of its own. Chime's filings show the shape of the contracts: 60 months with The Bancorp, running to July 2028, and 36 months with Stride Bank, renewing yearly, with 365 and 180 days' notice for non-renewal. On a termination without cause, Chime can move to a successor bank or wind down, "subject to regulatory approval".

A bank move is either a BIN migration (the same card numbers move to the new bank) or a reissue (new cards, new numbers). One processor notes that some BIN migrations may count as mergers under the Bank Merger Act, and that trailing settlement on the old BIN can run six months or more. Visa won't let the old bank release a BIN until it's been inactive for at least 120 days, and if the old processor won't hand over the data, the fallback is to reissue every card. A processor change at the same bank is lighter and may keep the BINs; one European processor quotes four to six months, longer for credit.

Mercury moved its customers from Evolve to Choice and Column after the Fed's 2024 order against Evolve; a new virtual card was available immediately and the physical one within one to two weeks. Chime moved all its processing and its ledger from Galileo to ChimeCore by November 2025. Block is moving some new Cash App business away from Marqeta, whose contract with it runs through June 2028 by the latest report, and Green Dot's partners will face a new counterparty once its split is approved.

Some fintechs are leaving the sponsor model altogether. Regulators cleared 29 bank charter applications in 2026 up to September 11, against 14 in all of 2025, including OCC conditional approvals for Nubank (January 2026), Mercury (April 28, 2026) and Revolut (September 2, 2026). Months of organizing and further approvals follow before a bank opens. Varo, the first consumer fintech with a national bank charter, still lost about $92 million in 2025, so a charter doesn't fix the economics on its own.

Program models and card types compared

Cash App's card is legally "prepaid debit", while Chime's is linked to a deposit account, and the two get different treatment under Reg E and Reg II. Credit adds underwriting and a balance sheet to absorb losses, which a later guide covers.