Alibaba's Flywheel Spun on Coercion. Then the Regulators Drained It.

A merchant in Hangzhou wants to sell on Taobao and JD.com both. That used to be a problem. To stay in good standing on Alibaba's platforms, the merchant learned, you picked one - and if you didn't pick Alibaba, your products quietly fell to the bottom of search, your promotions stopped getting approved, your shopfront simply stopped being found.⁵ That practice had a name inside the industry: 'choose one of two.' It was not a side hustle. It was a load-bearing gear in the most admired flywheel in Chinese commerce - and in 2021 a regulator declared it illegal and pulled it out.⁴

The official story is that Alibaba's ecosystem is a perpetual-motion machine: merchants draw shoppers, shoppers draw merchants, data makes both sides smarter, and the wheel spins faster forever on its own gravity. Most of that is true. But the part everyone leaves out is that one of the gears was turned by force, not by gravity - and the moment a regulator could tell the difference, the wheel slowed.

Here's the thesis a smart friend can repeat at dinner: Alibaba's flywheel is real, but it was partly coercion-powered - and a flywheel spun by coercion is a moat a regulator can drain.

How the wheel was supposed to spin

The flywheel idea isn't Alibaba's invention - Jim Collins coined it in 'Good to Great,' and Jeff Bezos drew the famous platform version on a napkin in 2001.⁸ Alibaba applied the same logic at enormous scale. More merchants on Taobao and Tmall means more selection; more selection pulls in more shoppers; more shoppers mean more transaction data; that data sharpens search, recommendations, and logistics routing; a better experience pulls in more shoppers still - and the whole loop becomes more attractive to the next merchant deciding where to sell. The numbers it throws off are not small. In fiscal year 2024, Alibaba's total group revenue was RMB 941.17 billion, up 8% year-over-year, with China commerce retail revenue of RMB 414.4 billion.¹ The premium membership program, 88VIP, reached 46 million members by the September 2024 quarter - the loyal core the wheel is built to retain.²

And there's an outer ring most flywheels don't have: cloud. Having built the infrastructure to run the largest commerce engine in the country, Alibaba sells that same capacity outward. Cloud Intelligence revenue was RMB 29.6 billion in the September 2024 quarter, with AI-related product revenue growing at triple-digit rates year-over-year for a fifth straight quarter.² The commerce wheel funds and feeds the cloud wheel, and the cloud wheel sells back the picks and shovels. On paper, it's beautiful.

The gear that wasn't gravity

Genuine network effects don't need a contract. If your platform is the best place to sell, merchants stay because leaving costs them sales - that's gravity. Alibaba's flywheel had gravity. But it also had a lock. The 'choose one of two' practice required merchants to sell exclusively on Alibaba's platforms or face retaliation: products buried in search results, promotions blocked, visibility throttled. That deliberately starved rivals JD.com and Pinduoduo of the very merchants that would have given them selection.⁵ In flywheel terms, Alibaba wasn't only making its own wheel spin faster - it was reaching over and jamming a stick into everyone else's. The two are easy to confuse from the outside, because both produce the same scoreboard: high share, deep merchant base, the look of inevitability.

China's market regulator, SAMR, opened its investigation on December 24, 2020, and on April 10, 2021 it ruled. The finding was not 'Alibaba is too big.' It was that a core mechanism sustaining the dominance was an illegal abuse of position - and it imposed a fine of RMB 18.228 billion, roughly EUR 2.4 billion, the largest antitrust penalty in Chinese history at the time.⁴ The fine stung. The remedy mattered more: the exclusivity gear had to come out, and the ecosystem was forced open. A flywheel doesn't bill you for the gear that breaks. A regulator does.

The wheel was already losing speed

The most damning evidence that the flywheel was less self-sustaining than advertised isn't the fine. It's that the wheel was slowing before the fine landed. Alibaba's e-commerce share by sales fell from about 76% in 2015 to about 61% in 2019 - a fifteen-point drop while the company looked unassailable.⁶ Then the timing turned brutal: in March 2021, while the SAMR case was still active, Pinduoduo surpassed Alibaba as China's largest e-commerce platform by active end users.⁶ A self-reinforcing loop is supposed to make a challenger's job harder every year. Pinduoduo took the users anyway - and by 2024 became the largest of the two by market value.⁶ Same wheel, different reading: a flywheel that needed a lock to hold its merchants was, underneath, a wheel that was leaking riders the whole time.

But isn't the flywheel still spinning fine?

The fair objection is that Alibaba is hardly broken. Group revenue still grew 8% in FY2024 to RMB 941 billion¹; China commerce retail revenue rose 5% in the December 2024 quarter to RMB 129.5 billion, with customer-management revenue up 9% on GMV growth and take-rate improvement³; the cloud and AI ring is compounding fast.² If the moat was fake, where's the collapse? The honest answer: there is no collapse, and the point was never that the whole flywheel was a fraud. The gravity is real - 46 million people pay for 88VIP because the experience genuinely keeps them², and that part of the wheel never needed a contract to hold. The argument is narrower and harder to dodge: a meaningful share of the spin came from a gear that turned out to be a legal liability, not a network effect, and the company is now growing in the single digits in its core retail business against a rival it could not lock out. The flywheel didn't stop. It was revealed to be smaller than it looked once you subtract the part regulators could take away.

A flywheel is a story companies tell about momentum - that the wheel, once spun, keeps itself going. Alibaba's wheel does keep going; nearly a trillion renminbi in annual revenue doesn't run on inertia alone.¹ But the most expensive lesson in its history is that not everything that makes a wheel spin is the wheel's own weight. Some of it is a hand on the brake of every competitor. Pull that hand away - by court order, in 2021 - and you learn the true diameter of the thing. The genius of a flywheel is that it's self-reinforcing. The danger is mistaking a lock for a law of physics. Alibaba spun a real one. It also spun a fragile one. And it took a regulator to show where one ended and the other began.