Alibaba Built the Perfect Flywheel. Then It Tore Out the Best Part.

In the early 2000s, China had a trust problem dressed up as a shopping problem. A buyer on Taobao would not wire money to a stranger who might ship nothing; a seller would not ship to a stranger who might never pay. So Alibaba slid a referee between them: pay into escrow, the seller ships, the money releases when the box arrives. That referee got its own legal life in December 2004 and a name — Alipay.³ It looked like plumbing. It was the most valuable thing Alibaba would ever build.

The story everyone tells is a flywheel: more trust drives more transactions, more transactions feed Alipay, more payment volume yields more data and more trust, and the whole thing spins faster every quarter until it becomes the largest U.S. listing in history in 2014.¹ Most of that is true. What the story leaves out is that by 2014, the best part of the flywheel no longer belonged to the company telling it. Alibaba had already torn out its own engine — and not by accident.

How escrow became a compounding loop

The mechanism is worth pausing on, because it is genuinely elegant. A marketplace's worst enemy is not competition — it is fraud, because fraud kills the willingness to transact at all. Alipay solved that by holding the money, which meant every additional buyer who trusted the system made the system more attractive to the next seller, and every seller made it more attractive to the next buyer. Payment volume was not a byproduct of commerce; it was the thing that created more commerce. And because Alipay sat on the money and watched every transaction, it accumulated something a pure marketplace can't: a real-time ledger of who pays, who delivers, who can be trusted. That data is the flywheel's flywheel — it priced credit, underwrote loans, and seeded a financial empire that would eventually try to go public at a $313 billion valuation.⁸ The marketplace fed payments; payments fed trust and data; trust and data fed the marketplace. Same loop, three turns of the wheel.

The night the engine left the building

Here is the part the legend skips. In August 2010, the equity in Alipay was transferred out of Alibaba Group; in the first quarter of 2011, Alipay was deconsolidated entirely. Alibaba's two largest shareholders — Yahoo and SoftBank — found out on March 31, 2011, that both moves had already happened. Per Yahoo's own filing with U.S. regulators, the transfers occurred 'without the knowledge or approval of the Alibaba Group board of directors or shareholders.'⁴ The flywheel's most valuable node now sat inside entities controlled by Jack Ma and his co-founders.

The official justification was regulatory: China's central bank was signaling that a payment license might be denied to companies tangled in the foreign-investor structures Alibaba used, so the VIE arrangement had to be unwound and Alipay made domestically clean. A Harvard analysis of the episode lays out exactly that chain of events — and then notes the cost. The eventual settlement, it concluded, reduced the value of Yahoo's stake in Alipay by more than 60%.⁶ The license may have been a genuine constraint. It was also a remarkably convenient one for the people who ended up owning the asset.

The July 2011 settlement set a price on the divorce: Alibaba would receive no less than $2 billion and no more than $6 billion when Alipay eventually had an IPO or liquidity event, and Alipay would keep processing Alibaba's payments on preferential terms.⁵ Read that carefully. Alibaba kept the service — the pipes still ran through Alipay — but it gave up the ownership of the compounding part, the credit, the data, the financial empire. It rented back the engine it used to own. A capped few billion dollars in exchange for a node that would later be valued north of $300 billion is not a flywheel turning. It is a flywheel being uncoupled.

Then regulators cut the other wire too

For a while it looked like a paper separation — Alipay was outside the group on the org chart but still the beating financial heart of the ecosystem, and Ant Group was barreling toward a listing that planned to raise $34.5 billion at a $313 billion valuation, one of the largest offerings ever attempted.⁸ Then, in late 2020, Chinese regulators suspended the IPO days before it priced, and over the following years forced Ant to restructure into something far more like a regulated financial holding company than an integrated super-app. In July 2023 the company was fined roughly $985 million for compliance failures spanning payments, financial services, governance, and anti-money laundering.⁸ The state did to the second half of the flywheel what Ma had done to the first: it pried payments and finance apart from commerce, and put walls between them.

So what is the flywheel today? Look at where the money comes from. Alibaba's China commerce retail revenue in fiscal 2024 was RMB 414.4 billion — about $57.4 billion — and the engine inside it is customer management revenue: fees merchants pay for marketing and placement, monetizing gross merchandise volume.⁷ That is a toll road. It is a very good toll road. But it earns by taxing transactions that pass through, not by owning the trust-and-credit loop that made the transactions multiply. The compounding part lives next door now, behind a regulatory wall, with its own license and its own owners.

But didn't the flywheel still work?

The honest objection is that the numbers don't look broken. Alibaba still hosts one of the planet's largest commerce platforms, still printed the largest listing in U.S. history, and still grows.¹⁷ If the flywheel were truly severed, where's the wreckage? The answer is that the wreckage is in the counterfactual, not the income statement. A loop that owns payments AND commerce compounds upside at both ends — credit, insurance, wealth management, the entire financial layer that Ant tried to take public for hundreds of billions. Alibaba's marketplace kept growing in the low single digits in fiscal 2024;⁷ the part that was capped at a few billion dollars was the part that was supposed to grow exponentially. The flywheel didn't stop spinning. It just lost the gear that turned commerce into finance — and that gear was always where the legend's real money lived.

Alibaba built one of the cleanest flywheels in commercial history — escrow that manufactured trust, trust that manufactured volume, volume that manufactured a financial empire. And then, twice, the most valuable half was carried out the side door: once by its own founders in 2010 and 2011, before the board even knew, and once by the state a decade later. What's left is real and profitable and large. It is also no longer the machine in the legend. The flywheel still turns. It just no longer turns money into the thing that turns money — and that, not the formula, was always the genius worth owning.