Square's Most Famous Win Was a $71 Million Loss. The Real Machine Was Built Elsewhere.

In 2009, a glass artist named Jim McKelvey lost a sale worth a few thousand dollars — a glass faucet — because he couldn't accept an American Express card.⁸ His co-founder told the story years later in a federal filing, with the flat formality such documents demand: 'We started Square because our co-founder, Jim McKelvey, couldn't accept a credit card for his art.'¹ That is the founding myth, and it is a good one: a tiny seller, locked out of the payment system, builds the key. But the more instructive story isn't the lost sale. It's the famous win that everyone celebrates — and that quietly cost Square more than $71 million.

The official story is that Starbucks validated Square — that when a 7,000-store giant signed on as a customer, the world finally believed the little white reader was real. The financial story, buried in Square's own S-1, is almost the opposite: Square lost money on every Starbucks swipe, the deal was being unwound before the IPO, and the company that 'won' the marquee logo nearly let it define the worst part of its prospectus.

The trophy customer that bled cash on every cup

In August 2012, Starbucks invested $25 million in Square's Series D, made Square the exclusive processor for all 7,000-plus U.S. stores, and put CEO Howard Schultz on Square's board.⁵ On paper, the dream endorsement. In practice, a structural trap: Square's contracted share of interchange came in below its cost to process each transaction, so the bigger Starbucks grew, the more Square lost.⁴ By the time the S-1 was filed on October 14, 2015, cumulative losses on the deal exceeded $71 million, and Square had already begun unwinding it.⁴² The lesson hides in plain sight: a megabrand has the leverage to negotiate terms that turn a 'customer' into a subsidy. The logo that was supposed to prove the business model was busy contradicting it.

Read against that origin, the contrast sharpens. Square was built to serve the seller too small to matter to the banks — the McKelveys of the world. The Starbucks deal pulled it toward the opposite pole: a single account so large it could dictate price. The real Square machine was never going to be powered by giants. It was going to be powered by the millions of small sellers no giant would ever bother to underprice.

The IPO that 'popped' and fell at the same time

When Square went public in November 2015, it priced at $9.00 and the stock opened at $11.20 — above the offer price, the kind of detail that gets reported as a win.³⁹ But $9.00 was set below the preliminary $11–$13 range in the S-1/A, and a full 42% below the $15.50 that private investors had paid in the last round.²³ The 'pop' was real and the down round was real, and they describe the same event. Public-market buyers got a discount; the last private backers took a haircut. The market wasn't disputing that Square had volume. It was pricing the unresolved question underneath the volume: how do you make a thin payment fee turn a profit when your most famous customer proves the fee can go negative?

The real machine: own the swipe, then sell everything around it

Here is the thesis, plainly. Square isn't a payments company that wandered into software. It's a software-and-lending company that uses the payment as the doorway — and the doorway, taken alone, barely pays. The little reader gets a seller in the door at near-cost. The margin comes from what attaches once they're inside: payroll, point-of-sale software, marketing tools, and above all credit, because a processor that watches every dollar a seller takes in knows exactly how much that seller can safely borrow. Payments are the sensor. Financial services are the business.

The honest objection: the flywheel still isn't finished

The fair counter is that this is too tidy. Block now runs two ecosystems — Square for sellers and Cash App for consumers (it launched as 'Square Cash' for peer-to-peer transfers in October 2013 and was later rebranded Cash App)¹⁰¹¹ — and the bull case is that they fuse into a single closed loop where a buyer pays a seller and both stay inside Block's rails. That loop is the prize. It is also not yet built. The two halves still operate largely apart, and the company's long-term edge rests on financial-services attach rates it has not proven at scale. The 2024 operating profit is genuine and hard-won, but it arrives a decade after the IPO and is thin against $24.1 billion of revenue.⁶ So the position isn't a finished flywheel; it's a half-built one, spinning on the strength of four million sellers it actually can up-sell — and waiting on the attach math to close the loop. The honest read is that Square survived its way to a strategy, layering product on product, and the elegant version was written backward from the result.

McKelvey lost a single glass-faucet sale over a card he couldn't accept, and out of it came a company that, in a single year, carried $228 billion across its rails.⁷ But the path there ran through a $71 million 'validation' that taught the most expensive lesson in Square's history: the swipe is never the prize. The prize is what you can sell to the person standing at the register — and the one customer Square couldn't sell anything to was the one it put on the cover. The machine works only as long as the door it opens leads somewhere worth more than the door.