This article was originally published on Grist as part of the Climate Desk collaboration.
In 2023, fast-fashion giant Shein is ubiquitous worldwide. Small packages depart from [thousands of suppliers] and are transported by air to millions of customer mailboxes in over 150 countries. Influencers promote the company’s trendy styles through “#sheinhaul” videos on social media, garnering [billions of views].
Data is generated, collected, and analyzed at every stage. To manage this information, the fast-fashion industry has started adopting emerging AI technologies. Shein uses proprietary machine learning applications—essentially pattern recognition algorithms—to measure customer preferences in real-time and predict demand, which it then meets through an ultra-fast supply chain.
As AI accelerates the production of affordable, fashionable clothing, brands like Shein face [increasing pressure] to become more sustainable. The company has committed to reducing its CO2 emissions by 25% by 2030 and achieving net-zero emissions by 2050.
However, climate advocates and researchers argue that the company’s rapid manufacturing and pure online business model is inherently high-emitting, and using AI software to catalyze these operations may exacerbate its emissions. These concerns were amplified with the release of Shein’s [third annual sustainability report], which showed a near doubling of CO2 emissions from 2022 to 2023.
“AI has turned fast fashion into ultra-fast fashion, with Shein and Temu leading the way,” said Sage Lenier, Executive Director of Sustainable and Just Future. “They wouldn’t exist without AI.” (Temu is a rapidly rising e-commerce giant that rivals Shein in product variety, pricing, and sales.)
In its 12 years since inception, Shein has become renowned for its uniquely efficient manufacturing capabilities. It reportedly generated over [$30 billion] in revenue in 2023. Although estimates vary, a new Shein design can reportedly be produced within [10 days], with up to 10,000 new items added daily to the website. The company offers as many as [600,000 products] for sale at any given time, averaging around $10 each. (Shein declined to confirm or deny these reported figures.) A market analysis found that 44% of Gen Z in the U.S. purchase at least [one] Shein item monthly.
This scale has significant environmental impacts. According to its sustainability report, Shein emitted 16.7 million metric tons of CO2 in 2023—more than four coal-fired power plants’ annual emissions. The company is also criticized for [textile waste], [high levels of microplastic pollution], and exploitative labor practices. Polyester, a synthetic fabric known to release microplastics into the environment, accounts for 76% of its total textiles, with only 6% being recycled.
A recent investigation found that workers in Shein supplier factories regularly work over 75 hours per week, even after the company pledged to improve working conditions in its supply chain. Despite claims of improved labor conditions in Shein’s sustainability report, third-party audits of more than 3,000 suppliers and subcontractors showed that 71% scored C or lower on the company’s A-to-E rating system—[at best, average].
Machine learning plays a crucial role in Shein’s business model. While Peter Pernot-Day, Shein’s Global Strategy and Corporate Affairs lead, told Business Insider last August that AI [was not central to its operations], he expressed a different view during a retail conference earlier this year.
“We are using machine learning technologies to predict demand in what we believe is the most cutting-edge way,” he said. Pernot-Day informed attendees that Shein’s 5,400 suppliers have access to an AI software platform that provides updates on customer preferences, allowing them to adjust production content in real-time.
“This means we can produce small batches of each garment,” he explained. “It means we waste very little and have minimal inventory overstock.” The company claims it typically stocks between 100 to 200 units per style on average, contrasting sharply with more traditional fast-fashion brands that often produce thousands of items per quarter months in advance based on trend forecasts. Shein calls its model “on-demand production,” while a tech analyst described it as “real-time” retail during [a 2021 interview with Vox].
At the conference, Pernot-Day also noted that the technology helps the company capture “micro-trends” customers want to wear. “We can detect these and act on them in what I think is a truly pioneering way,” he said. A designer recently filed a class-action lawsuit in New York Supreme Court alleging that Shein’s AI market analysis tools were used for “massive systematic, digital copyright infringement against small designers and artists,” by scraping designs from the internet and sending them directly to factories for production.
In an email statement to Grist, a Shein spokesperson reiterated Peter Pernot-Day’s claims about reducing waste and improving efficiency through technology and stated that the company’s increased emissions in 2023 were due to business growth. “We do not believe growth is at odds with sustainability,” the spokesperson said.
An analysis by Business of Fashion of Shein’s sustainability report found last year’s emission increase was nearly [twice] its revenue growth rate—making Shein one of the highest-emitting companies in the fashion industry. In contrast, Zara’s emissions increased only half as much as its revenue. For other industry giants like H&M and Nike, sales grew while emissions decreased from the previous year.
Shein’s high emission levels are particularly due to its reliance on air transport, according to Sheng Lu, a professor of fashion and apparel studies at the University of Delaware. “AI has wide applications in the fashion industry, but AI is not necessarily bad,” said Lu. “The issue lies in the nature of Shein’s specific business model.”
Other major brands ship goods overseas in bulk, favoring sea transport due to lower costs and maintaining suppliers and warehouses across many countries to reduce delivery distances for consumers.
According to its sustainability report, 38% of Shein’s climate footprint comes from transportation between facilities and customers, with another 61% coming from other parts of the supply chain. While headquartered in Singapore with suppliers in several countries, most clothing is produced in China and shipped individually packaged via air to consumers. In July this year, the company sent approximately [900,000] packages daily to the U.S.
A Shein spokesperson told Grist that the company is developing a decarbonization roadmap to address its supply chain footprint issues. Recently, it has [increased inventory in US warehouses], enabling faster delivery times for American customers and increased use of carbon-efficient cargo ships.
“Controlling carbon emissions in the fashion industry is a very complex process,” Lu said, noting that many brands use AI to improve operational efficiency. “It depends on how you use AI.”
Research shows that using certain AI technologies can help companies become more sustainable. “This is the missing link,” said Shahriar Akter, Deputy Dean of Business and Law at the University of Wollongong in Australia. In May this year, Akter and colleagues published a [study] finding that when fast-fashion suppliers use AI data management software to comply with major brands’ sustainability goals, these companies are more profitable and emit less. One key application of the technology is closely monitoring environmental impacts such as pollution and emissions, said Akter. “This kind of tracking was not available before the advent of AI tools,” he added.
Shein told Grist that it does not use machine learning data management software to track emissions, one of the AI uses in Akter’s research. However, its widely publicized use of machine learning software for demand prediction and waste reduction is another application identified in the study.
Regardless, there is still a long way to go for the company to achieve its goals. Grist calculated that Shein saved emissions equivalent to about 3% of its total annual carbon footprint in 2023 through measures like providing solar panels to suppliers and choosing sea transport over air freight.
Sustainable and Just Future’s Lenier believes there is no ethical use of AI in the fast-fashion industry. She argues that largely unregulated technology allows brands to exacerbate harmful impacts on workers and the environment. “Workers in fast fashion factories are now under immense pressure to produce more products faster,” she said.
Lenier and Lu both believe achieving a more sustainable fashion industry hinges on convincing consumers to buy less. If companies use AI to boost sales without changing unsustainable practices, their climate footprint will grow accordingly, Lu noted. “This is the overall effect of being able to offer more marketable products and encouraging consumers to buy more than before,” he said. “Of course, the overall carbon impact would be higher.”
Source: WIRED
