In the past two months, reports about the difficulty of getting deposits back from Ofo have kept surfacing. What began as an almost instant refund turned into a seven-day wait, then fifteen days, then a process that reportedly required repeated phone calls just to receive the money after those fifteen days. After that came attempts to convert deposits into financial products by passing them to a third party. Now the most direct sign of trouble has appeared: the refund button itself has been grayed out and can no longer be clicked.

Rumors have also spread online that Ofo offices in a number of cities have already been emptied out. Teams once responsible for collecting, transporting, and repairing bikes have reportedly been disbanded. In practical terms, that means Ofo bikes in many places have been left without management. Yellow bikes can be seen hanging from trees, tossed into rivers, abandoned in bushes, or left by roadsides under a thick layer of dust. In several cities, municipal workers have begun clearing them away because they have become a serious nuisance for both the streetscape and daily travel. News from Hangzhou even reported that a batch of Ofo bikes was dismantled in bulk and treated as scrap metal. Images of shared bikes compressed by machines into square blocks of metal have circulated widely online, and their destination is obvious enough: recycling yards.

Ofo is not the first bike-sharing company to collapse. Before it came other names such as Bluegogo, along with many smaller local brands that never had much chance to survive. The remarkable part is how quickly this happened. From the rise of bike sharing to today’s widespread retreat, only about two years passed. Online figures have claimed that more than 3,000 bike-sharing brands have already disappeared. Ofo, once one of the two dominant players, now appears critically weakened. For a long time, Ofo and Mobike together controlled about 95% of the national market. In terms of scale, Ofo was even somewhat larger than Mobike in many places. If Ofo falls, that would mean the collapse of roughly half of the shared-bike market.

The reason so many bike-sharing companies kept falling was never hard to see: the business model was unclear from the start. In the early scramble for market share, companies competed with free rides and sometimes even cash rewards for riding. In other words, they paid users to use a service that already had weak margins. Many small brands died in this phase alone. Competition in service quickly turned into competition in financing. Smaller companies had far more difficulty raising money, and after following the larger players for just a few months, they were pushed to the margins. In a business built around network scale, the smaller the fleet, the fewer the users; the fewer the users, the less viable the business. For many operators, failure became unavoidable.

The original expectation was that once small competitors were eliminated, the remaining giants would dominate the market and eventually earn outsized profits. Operators thought this, and venture capital did too. That is why the market became so frenzied. In roughly two years, more than 100 billion yuan was reportedly poured into the sector. But the outcome moved in the opposite direction. Once free riding gave way to paid riding, daily active users kept declining. Bike damage and theft remained serious. At the same time, city regulators tightened oversight and increasingly required bikes to be parked only in designated areas. That created a need for large numbers of workers to gather and move bicycles around, which drove operating costs sharply higher.

When the economics were finally added up, the numbers stopped making sense. In some cases, the cost of retrieving and repairing bikes in remote locations was reportedly higher than the cost of simply putting brand-new bikes on the street. As a result, more and more damaged or stranded bikes were abandoned. Municipal departments collected these bikes and asked the companies to reclaim them, but reclaiming them meant paying collection fees and fines. If taking them back cost more than buying new bikes, there was little incentive to do so. That is how bike graveyards emerged in one city after another, with piles of bicycles eventually dismantled and sold as scrap.

There was another structural problem as well. During the early land-grab phase, companies flooded cities with too many new bikes. Later, many cities placed limits on additional deployment. That turned bike sharing into a stock market rather than a growth market: operators could no longer expand freely and had to rely on the existing fleet. As damage accumulated, the usable number of bikes kept shrinking. The result was a late-stage shortage in many places, where people commuting during rush hour often could not find a shared bike at all.

Many people explained the failure of bike sharing in China by pointing to low public civility, arguing that high rates of vandalism and theft drove up costs. That is one reason Mobike and Ofo both looked overseas for growth. But when they entered markets such as the UK and Japan, they ran into many of the same problems. Damage rates remained high there as well, despite those countries generally being seen as having stronger public order. Once the model failed to prove itself globally, the idea that bike sharing could keep expanding indefinitely became much harder to defend. Ofo already stands close to the edge, and Mobike has not looked particularly strong either. In Beijing, there was a time when Ofo and Mobike filled the sidewalks outside nearly every subway station. Now even finding a Mobike during commuting hours can be difficult.

Bike sharing was one of the earliest and most visible symbols of the so-called sharing economy, and for a time it was its main pillar. Other highly publicized ideas—shared umbrellas, shared clothing, shared phones—had already faded from the market much earlier. At this point, only a few categories still appear to be holding on, mainly shared bikes and shared power banks. The broader boom in the sharing economy now looks as though it has passed. Once the trend loses momentum, the companies lifted by it have to come back down.

For the past two years, the sharing economy and internet finance were often described as two of China’s most successful forms of innovation, both seen as globally advanced. But this year has brought repeated collapses in P2P lending platforms, while the sharing economy has also moved toward its closing chapter. What once looked like a wave of bold innovation has too often ended in losses for everyone involved: users lose money, suppliers are hurt, and founders disappear.

That may be the clearest lesson left behind. Real innovation is not easy. Whether the innovation lies in technology or in a business model, it rarely succeeds in one leap. Painful failures are often part of the learning process, and industrial upgrading is usually far slower and harder than the excitement of a boom makes it seem.