How Stripe Became the Financial Infrastructure of the Internet

The Original Insight That Changed Payments

In 2010 Patrick and John Collison brothers aged 21 and 19 respectively identified a problem that seemed mundane but was actually huge: accepting online payments was absurdly complicated. Legacy payment processors required weeks of paperwork opaque approval processes technical integration that took months and fee structures that didn't make sense for startups. The dominant players PayPal Authorize.net traditional merchant acquirers had been built for a traditional world.and adapted for the Internet. None of them had been built natively for developers. The Collisons' idea was to make payment acceptance a seven-line snippet of code. Their original API was elegant in a way that payments had never been: you could copy the integration code directly from the documentation and have a functional payment flow in hours not weeks. By gaining developers Stripe integrated into products before those products had customers generating switching costs that only grew over time

A Worked Example: Where the 2.9 Percent Actually Goes

The main price is 2.9 percent plus 30 cents and almost everyone treats it as a Stripe income. Follow a single transaction through the system and a very different deal appears

Make a card payment of $100. Stripe charges the merchant $2.90 plus 30 cents or $3.20 in total

Now pay what Stripe owes. Exchange It goes to the bank that issued the customer's card and on a typical consumer credit card it's about 1.8 percent plus 10 cents or about $1.90. Network Assessment Fees go for Visa or Mastercard in the region of 0.14 percent or about 14 cents

So a company that advertises a 2.9 percent fee actually keeps about 1.16 percent and about two-thirds of what it collects is passed directly to parties it doesn't control. As for gross fees that's a margin of about 36 percent before a single engineer is paid

This single table explains more about the payments industry than any strategic discussion. Scale matters enormously because the retained margin is small. Large merchants negotiate the gross fee down and every basis point they earn comes entirely from the 1.16 instead of the transfer. And no processor can compete on prices below their own interchange cost which is set by card networks and issuing banks

Now put a price on the moat because that's the other half of the business. Consider a company that uses Stripe for payments invoicing fraud detection and marketplace payments. Migrating means rebuilding every integration retesting every webhook recertifying compliance and running both systems in parallel while money moves

Call it six engineers for nine months at $200,000 each fully equipped. That's 6 times $150,000 or about $900,000 and ignores the risk of reducing payments during the transition which for most companies is the real deterrent

By contrast suppose a competitor offers prices 20 basis points cheaper for an annual volume of $50 million. The savings is $100,000 per year. Divide the cost of the migration by the annual savings and the payback period is approximately nine years

That's the moat expressed as a number. It's not brand loyalty or developer affection but rather a nine-year payback on a project that puts the company's cash collection at risk. These are illustrative figures and the ratio varies wildly by size which is very important in the case study below

The Expansion Playbook

What makes Stripe's story analytically interesting is the discipline of its expansion. Most fintech companies expand by moving into adjacent consumer products. Stripe consistently remained focused on business customers and went deeper into the payments and financial services stack rather than deeper into the consumer. Stripe Atlas (global business formation) Stripe Treasury (banking as a service) Stripe Radar (fraud detection) Stripe Connect (marketplace payments) Stripe Billing(subscription management) Stripe Issuing (card issuance) each product expanded Stripe's relationship with existing customers while deepening switching costs. A company that uses Stripe for payments fraud detection banking and marketplace payments faces a huge migration challenge to abandon it. It's not one integration to replace it's dozens. The business model is elegantly simple: about 2.9% plus $0.30 per online transaction with volume discounts negotiated for largecustomers.Revenue grows automatically as businesses using Stripe grow without requiring additional sales expenses.This is a structurally attractive economic model

Read that list again with the table above in mind. Each of those products is an attempt at non-exchange revenue because exchange is the two-thirds that Stripe never keeps. Fraud detection billing and treasury all involve software margins rather than payments margins and each of them also lengthens the migration project. Product strategy and moat strategy are the same strategy

What Stripe Does Not Own

For a company commonly described as the financial infrastructure of the Internet it pays to be precise about which parts of that infrastructure it actually controls

Does not establish interchange. Those fees are set by Visa and Mastercard and paid to the issuing banks and are the largest single cost in the table above. It does not issue the cards own the majority of the underlying deposits or operate the settlement networks. In several jurisdictions it relies on sponsoring banks for regulated activities

What Stripe owns is the developer interface the risk and fraud models built from observing huge transaction volumes and the integration surface built up within its customers' code. Those are genuinely valuable and genuinely defensible. They are also a layer on rails that someone else operates which limits the margin to 1.16 percent on the table and means that a change in network rules or exchange regulation flows directly through the business without Stripe having a vote

Case Study: The Day Adyen Fell 39 Percent

The above argument about switching cost is real and the clearest evidence of its limits came to Stripe's closest competitor

Adyen is a Dutch payments company with a similar profile: technically excellent respected developer serving large merchant companies and has long been treated by investors as if it had an unassailable position. In the early 2020s it was trading at a substantial premium exactly according to the reasoning in this article that once a large merchant integrates a payments platform it does not abandon it

In August 2023 Adyen reported that first-half results showed slower-than-expected growth particularly in North America. Management explained that large merchants were prioritizing costs and that competitors had been gaining volume based on price. The stock fell approximately 39 percent in a single session one of the largest one-day drops ever recorded by a major European technology company and continued to fall over the following months

Nothing had broken. The technology worked customers hadn't left and the company remained profitable. What changed was the market's belief about a single variable: whether large merchants are truly trapped

Go back to the payback calculation above and the answer becomes obvious. A company that processes $50 million a year faces a nine-year payback from migration so it stays put. A company that processes $5 billion a year saves $10 million annually for the same 20 basis point difference versus a migration project that might cost a few million. Your payback is a matter of months and you can afford to have a dedicated team to execute the project correctly

The moat is not owned by the product. It is a relationship between the cost and volume of the migration and it reverses as the customer grows. Which means that the customers a payments company wants the most the largest are precisely the customers it maintains with the least security and the strength of the moat is greatest exactly where revenue is lowest

The Valuation History and the IPO Question

Stripe's private valuation has been volatile in ways that say something about both the company and the venture capital market in general. The company earned a valuation of $95 billion in 2021 fell to $50 billion in 2023 a 47% downgrade reflecting rising fees and declining growth multiples and recovered to roughly $65-$70 billion in 2025 as itThe IPO question is one the company has studiously avoided answering. Patrick Collison has consistently said Stripe will go public when it makes sense for the company.company.The company has the financial profile to maintain its private status indefinitely having approached EBITDA profitability in 2024. The question is not if but when and that timeline is completely controlled by the Collisons

Where the Stripe Story Is Oversold

I think Stripe is a really great company and the standard telling of their story leaves out four things

The intake rate is successfully compressed. The 1.16 percent in the table is about what a small merchant makes based on the list price. Every customer that grows into a large one negotiates that reduction so the company is structurally exposed to the success of its own customers. A payments company that celebrates that its merchants are scaling is celebrating margin compression

The valuation round trip tells you what type of asset this is. Ninety-five billion dollars in 2021 fifty billion in 2023 and back to sixty in 2025 in a business whose operations were improving all the time. That is not the price behavior of a fortress. It is the price behavior of a long-duration growth asset that trades at interest rates which is perfectly respectable and not how the company is usually described

Maintaining privacy avoids argument rather than winning it. No audited financial statements no disclosed take rate no churn or cohort data no segment breakdown. Everything written about Stripe's economics including this article is based on estimates and inferences. That's a choice the company has the right to make and it means no one outside can really control the moat

Developer experience can and has been copied. Seven-line integration was revolutionary in 2010 and is now in play. All serious competitors offer good documentation and a clean API because Stripe showed that it matters. The first-mover advantage in the developer experience is a real historical advantage and is not permanent

My view is that the lasting moat is the depth of integration built into many products rather than the elegance of the API and that Adyen showed exactly where that moat ends

How I Would Analyse a Payments Company

Payments is one of the easiest industries to understand and one of the most difficult to evaluate mainly because the headline numbers are misleading in a consistent direction

I would start by separating gross income from net income exactly as in the table above because the difference is about two-thirds and companies vary in which of them leads. Total payment volume multiplied by a headline rate says almost nothing

Second I would look at the mix between payments revenue and software revenue. Fraud billing and treasury products have very different processing margins and a company that shifts its mix toward them is doing something structurally different from growing volume

Third I would calculate the migration payback period for the customer segment in question because that single ratio determines whether the customer base is genuinely fixed or simply currently satisfied. The answer differs by an order of magnitude between small merchants and large enterprises

Fourthly I would like to ask what happens to the business if the exchange is downregulated since this is a policy issue in several jurisdictions and flows through the economics of a processor without the processor having a say

Fifth for any private company I would treat every reported metric as marketing until an S-1 exists. That's not cynicism specifically about Stripe it's what the absence of an audit means

This is how I would frame the work. It is a description of the method rather than investment advice

The Bottom Line

Stripe turned payment acceptance from a months-long integration project into a seven-line snippet of code then spent fifteen years making sure no one could leave and Atlas Treasury Radar Connect Billing and Issuing each added another integration to the migration project

The economics are weaker than the headline suggests. On a $100 transaction 2.9 percent plus 30 cents yields $3.20 of which about $1.90 goes to the issuing bank as an exchange and 14 cents to the card networks leaving about $1.16. That's a net take rate of 1.16 percent on the rails that Stripe doesn't own. The moat is real and so isMeasurable: The nine-year payback from migration for a medium-sized trader is the reason almost no one moves

Adyen is the reminder that the ratio is reversed. It fell about 39 percent in a single session in August 2023 when large merchants turned out to be changing volume based on price because on an enterprise scale the recovery from migration is months instead of years. Stripe's valuation went from 95 billion to 50 billion and back to 60 without the business changing much indicating that it is valued as a long-term growth asset inrather than a fortress. The question is not if it will be made public but when and that timeline is completely controlled by the Collisons