From Xiamen to Shanghai, mass graveyards of dirty bikes, all twisted frames and busted axles and handlebars, have become an unwanted emblem for hundreds of Chinese start-ups that once thrived on the back of easy money, hard graft and a light regulatory touch.
When the idea took hold in 2015, the bike rental companies' promise to attract China's booming middle class pulled in billions of dollars from investors even if they often charged cyclists very little or in some cases nothing to use their services.
Some, such as Mobike and Ofo, quickly expanded abroad.
However, both have subsequently slashed their overseas presence. Ofo's founder, Dai Wei, warned that it was teetering on the brink of bankruptcy. Wukong and Bluegogo have already folded.
They have now come to symbolise much of what has gone wrong across a swath of Chinese tech companies, especially those built around the idea of the sharing economy. The companies in trouble range from food delivery and shopping websites to transport apps.
"The transaction-oriented model is more or less done," says Jason Ding, partner at Bain & Co consultancy in Beijing.
"The bubble burst on the shared economy... It was pumped up by money on steroids. That's all gone."
The Chinese tech sector has developed a broad range of businesses in the space of four to five years. In 2018 alone about 100 tech start-ups became "unicorns" worth more than $1bn according to research group Hurun.
But the rapid expansion began to slow in the final quarter of last year. Capital is retreating or looking for deals in other regional markets, workers are rebelling and Beijing is grumbling about the power wielded by some of these companies.
Many of the tech businesses have found themselves with a fatal flaw of paying more to win customers than their customers bring in.
"I think the whole of China is trying to find a new business model," says one industry executive and investor.
While the US frets about China's rising tech prowess — a fear that has been a driving force behind the trade war and the efforts to choke off telecoms equipment company Huawei — the reality for many companies is less auspicious.
Instead, the trade war has served to highlight the weaknesses of many companies. "Since the trade conflict these gaps [between China and US tech capabilities] have become more evident," says Shirley Xie, who leads PwC's China and Hong Kong consulting practice.
"And tech companies absolutely get it."
For the last few years, there appeared to be a never-ending flow of capital into the sector. This allowed a company like Ant Financial to raise $10bn in a fundraising round last year that gave it a valuation of more than $150bn and spawned scores of so-called "PPT companies", whose prospectuses were based on little more than a PowerPoint presentation, say investors.
But investors have started to become squeamish. Aggregate deal value in the Chinese IT sector in the second quarter of 2019 was $2.2bn, compared with $26.4bn for the same period a year earlier, according to data provider Preqin.
An evaporation of state-backed renminbi-denominated funds was followed by a more disciplined approach among US venture capital firms in response to the high valuations and questionable business models of some tech groups — with the blow cushioned only by bloated coffers at blue-chip firms, such as Sequoia and Hillhouse, which each raised $US8b (S11.3b) last year.
The impact really started to be felt in the first quarter. Investors were spooked after a series of fund raisings that valued the start-up below the valuation of its previous round and a year of torrid performance by the 30-odd tech start-ups that listed in 2018 spooked.
According to one former banker and start-up executive: "Leverage has transferred from founders to investors."
Due diligence on prospective investments has significantly increased.
Another banker-turned-start-up executive recalls his fundraising last year: "People were calling us up saying, 'we have $100m ready'. There was no due diligence."
Nisa Leung, a managing partner at Chinese venture capitalists Qiming which boasts a large portfolio of tech and biotech start-ups, says due diligence frequently only used to begin after submitting a term sheet.
"The tide has changed," she adds. "People are tough on terms, and spend much longer on due diligence."
This new-found rigour is translating into extended financing rounds and "value adjustment mechanisms" designed to protect investors from falling values — for example, by pledging extra issuance of shares to balance any drop in value. That marks a sea change.
The first former banker and executive says: "The fact a sponsor who nine months ago was clawing to get into a deal is now insisting on value adjustment mechanisms ... shows there's no appetite among the global players for some of these companies."
Casualties of this new discipline include VIPKID, which employs native-English speaking teachers to teach children in China online and is struggling to raise funds, and Megvii, one of China's biggest artificial intelligence companies.
Megvii, which stole an early march in facial recognition, was deserted by earlier backers when it looked for more funds in May. Instead, it fell back on state-backed entities to provide the bulk of its $750m financing.
Fragile sentiment, aggravated by the escalating trade war, is spilling into the public markets — even as Chinese start-ups rush to access funds before investors shun the tech market.
Among the Chinese tech companies which listed last year, Uxin, a second-hand car marketplace, is down 73 per cent from its float price, while clothes-shopping site Mogu's shares are 81 per cent off their IPO level.
China tech's second jolt came from employees, some of whom earlier this year took to online developers' platform GitHub to voice their frustration at the "996" culture — working 9am to 9pm, six days a week.
Alibaba, JD.com and ByteDance were among those named and shamed in a campaign that caught the attention of the state-run Communist Youth League's newspaper, which said "dissatisfaction has reached a turning point".
Long hours may be expected by start-ups from Silicon Valley to Shenzhen, but it is exacerbated in China by the shifting terms of engagement as start-ups seek to cut costs and contain losses.
Bonuses have shrunk at companies like Didi Chuxing and perks — everything from fruit bowls to gym membership — have quietly been discontinued.
"Tech is maturing here," says Seamon Chan, co-founder and managing partner of Palm Drive Capital and a previous investor in Alibaba.
"So people are asking if it is sustainable or not. Can you sustain 996 for the next five or 10 years? 20 years? It's very important to pace themselves."
Plenty of founders, including Alibaba's Jack Ma, see it differently.
"If you join Alibaba, you should get ready to work 12 hours a day, otherwise why do you come to Alibaba? We do not need those who comfortably work eight hours," he said in a transcript published in April on the group's WeChat account.
The gripes of workers and investors point to a bigger issue: a growing number of start-ups that look like a busted flush. From shared bikes and food delivery to online English classes, companies are finding that their consumer-focused business models do not add up because of intense competition.
Pinduoduo, a shopping site, typified the cash-burning model dependent on selling large volumes of goods.
The Shanghai-based company, run by former Google executive Colin Huang, spent $1.50 for every $1 in revenues it earned in the last quarter of 2018.
Pinduoduo shook up the cosy effective duopoly enjoyed by Alibaba and Tencent-backed JD.com but has since been joined by rivals, such as Xiaohongshu and Songshupinpin, offering their own twist on ecommerce.
Likewise there has been an explosion in online companies teaching English, some of which keep prices lower by not using native English speakers as teachers or hold group classes rather than individual tutorials.
"As competition has increased, in China especially, everyone's cost of acquisition has risen to a point where now few business-to-consumer companies will be profitable," says Toby Mather, co-founder and chief executive of Lingumi, an online education business in China and the UK.
"We are entering a B2C acquisition winter unless you have a genuine network effect like Instagram and Facebook."
Meitu, which listed in Hong Kong in 2016, is moving into skincare gadgets. This presents companies with a dilemma, says a tech banker who has helped to take scores of them public.
"Should they focus on growth and keep spending, or should they come out and say 'I can prove I am cash flow positive but only have growth of 5-10 per cent?'"
The banker adds: "The best is growth and earnings. And that's Alibaba and Tencent. So why bother with the smaller cap [companies]?"
Alibaba and Tencent are leading the charge on one initiative to cut customer acquisition costs: a focus on corporate sales including small and medium-sized enterprises.
Tencent's restructuring late last year — its first in six years and hailed as "a new starting point" for the group's next 20 years — created a unit focused on business and industrial customers.
But it is a trickier transition for smaller companies.
Serving enterprise customers often involves bigger spending on research and development, say consultants, unlike the simpler proposition of selling to customers via a one-size-fits-all app. It also requires different skillsets in distribution, products and services, HR models and organisation structure.
"Everyone is going through some form of transformation," Ms Xie says. "More and more companies are going outside for advice. This space is hotter than I've ever seen it and will be for the next 20-30 years."
One investor points to SenseTime, China's biggest AI start-up valued at $4.5bn after a latest funding round, to demonstrate the struggle to reach the next level. SenseTime grew its business on the back of selling facial recognition services including for surveillance purposes in cities, largely selling to government bodies.
"They are engineers but don't have great depth of experience in figuring out what customers need because they grew with one customer," the investor says.
Perhaps the biggest manifestation of the struggle to evolve comes in the ability of many companies to innovate new product lines.
At the China High Tech Fair in Shenzhen earlier this year, little seemed to have changed from previous years. Robotic arms place Oreo cookies in boxes, surveillance cameras glide round booths and the ubiquitous dancing robots provide a backdrop for selfies.
UBTech, valued at $5bn and backed by Tencent, sells its robots to schools, shopping malls, airports, hotels.
But at the trade fair an employee of the company, which is preparing for an initial public offering, said "a lot of orders are from government [bodies] and SOEs [state-owned enterprises]. They're more willing to buy local products or support high-tech development".
Some tech start-ups that seemed to be sparkling with new ideas when they listed have faded over time.
Meitu, which listed in Hong Kong in December 2016, has subsequently licensed its smartphone brand to former rival Xiaomi and slashed staff. It listed as a smartphone manufacturer with beautifying filter apps; now it is making a push into skincare gadgets.
Similarly the latest offering from Good Doctor, an online medical platform spun off from insurer Ping An in a $1.1bn listing last year, was a decidedly retro WeightWatchers-style milkshake.
For bullish observers, these problems are no more than teething pains as the sector evolves. While there will continue to be casualties in the tech winter, Chinese entrepreneurs are expected to rise to the task. "The talent is here. The motivation is here. And the demand is definitely here," says PwC's Ms Xie.
But sceptics point to the damage done by the tendency for the industry to be dominated by one single trend. "Three years ago it was the sharing economy. Now it's 5G and the industrial internet of things, so a lot of companies are joining in the game," a longstanding China tech executive says. "But nobody knows what is the business model for IIoT ... everyone is confused for the future."