Marqeta — What I Think
Marqeta’s insight is quietly inverted, which is why I love it: the card was never the product. The moment of authorization is the product.
Old card issuing worked like this. A bank pre-loads a balance, a processor maintains it, you swipe, the system checks a number that’s already sitting there. The decision was made in advance. Marqeta flipped the control flow. At the instant of the swipe, send a real-time call to you, the program owner, and let your code decide. Approve, decline, how much, against which funding source, under which rule, right now. They called it Just-in-Time funding. To an engineer it’s obvious the second you hear it: they turned authorization into a webhook. They made the card a programmable event.
That single inversion is why a whole generation of fintech exists. Every on-demand spend card, every expense tool issuing virtual cards on the fly, every BNPL checkout, every gig app paying a driver’s exact fuel cost at the pump and nothing more. They all need a card to ask permission at the moment of spend, not before. DoorDash paying for the exact order, Uber funding the exact trip, a SaaS tool issuing a single-use card scoped to one vendor. None of it works if the money has to be sitting on the card in advance.
What Marqeta got right is that issuing should be infrastructure other people build on, not a product a bank sells. They took the most ossified, paperwork-bound corner of payments, getting a card program stood up, and made it a thing a developer could call. The Stripe move, applied to the issuing side of the rails. Correct, and early.
Where the tension is real: an issuer-processor is deeply entangled with sponsor banks and the card networks underneath it. Marqeta enabled a thousand programs, but several are large enough to wonder whether they still need a middle layer, and the regulatory weather around bank-sponsorship has gotten heavy. The platform that made everyone else programmable has to keep proving it’s more than a pass-through. The deeper question: Marqeta made the card a programmable event, but it’s still a card, still riding the existing networks. The next inversion is when the programmable spend-decision happens on rails that don’t need a card at all. Marqeta taught the industry to think in that shape. Whether it gets to own the version without the card is the open question. Or something like that.
Favorite & worst CEO
This is essentially a founder-CEO story, so:
On its leadership. Jason Gardner founded Marqeta on a genuinely non-obvious bet, that real-time developer-controlled card authorization was a primitive the whole fintech wave would need, and he was right years before the market caught up. That’s the part I respect most. Conviction in an unsexy piece of plumbing that turned out to be load-bearing for an entire ecosystem. The harder part is the transition every founder eventually faces, from “we invented the category” to “we have to defend it against concentration risk and a shifting regulatory model,” which is a different game than the one that made the company. But the founding insight was sharp and early, and that’s the thing hardest to do and easiest to undervalue in hindsight.
Part of “What I Think About the Top 50 Fintech Companies of All Time.” I’m Prajjwal Chittori. prajjwalchittori.com.