Card processing harder than charters
Aaron Huang, Head of Commercial at Productfy, on choosing the right fintech customers
The key point is that card processing is a deeper systems problem than banking charters for a scaled fintech. Buying or launching a bank gives a company legal control over deposits and lending, but becoming the processor means owning the real time ledger, authorization logic, network connections, dispute flows, fraud controls, and uptime for every card swipe. Square chose to own a bank while still relying on Marqeta for that infrastructure, which shows how hard processor replacement is in practice.
-
Marqeta sat underneath Cash App card issuing early on, and its value was not just printing cards. It handled card management and processing, the machinery that decides in milliseconds whether a transaction is approved, how it is funded, and how it lands in the ledger. That is specialized infrastructure with long implementation cycles and high switching costs.
-
Square went after banking control through Square Financial Services, an FDIC insured industrial bank that began operations in 2021. That move helped it own lending and banking products directly, but it did not mean rebuilding the processor layer. The fact pattern supports Aaron Huang's point that the charter path was more tractable than replacing processor infrastructure.
-
This also explains the BaaS 1.0 to 2.0 distinction. Point solutions like Marqeta won by doing one hard thing extremely well. Newer platforms try to bundle bank accounts, payments, cards, compliance, and multiple bank relationships, but the processor layer remains the hardest piece to recreate because one large customer can dominate the roadmap and demand enterprise grade reliability from day one.
Going forward, the winners in BaaS will be the companies that treat processing as the anchor layer and build bank flexibility on top of it. More fintechs will add bank control where it improves economics or product speed, but few will rip out a working processor unless they have enormous scale, years of engineering time, and a reason to own every part of the card stack.