Abhi Arora, like many people, liked to buy vintage clothes. But he had never given too much thought to exactly how they arrived in second-hand shops. Then one day, Arora’s favorite vintage clothing shop near Brick Lane, in London’s East End, announced it was closing.
Surprised, Arora asked the shop owner why. She told him that she used to source her vintage items directly, going on buying trips to Asia and the Middle East to purchase directly from the wholesale suppliers of second hand clothes. The COVID-19 pandemic had put an end to those buying trips and gave rise to online brokers who charged high commissions. Some of brokers were also less than scrupulous. The Brick Lane vintage store owner had been scammed out of more than $15,000 by one of these brokers and, unable to recover the funds and with her margins shrunk by brokerage fees, she reluctantly decided to close her store, Arora said.
But that conversation gave Arora a brainwave. He teamed up with Sanket Agarwal, whom he had met in an online founders’ community and whose mother-in-law also sold vintage clothing and had also struggled with sourcing products. “We put two and two together,” Agarwal said, “and we said there’s a real problem in this industry where supply is not scaling as fast as demand.”
That insight became Fleek, an online marketplace the two cofounded in 2021 that connects second hand clothing wholesalers—most of them in India, Pakistan, and Dubai, where much of the world’s second-hand clothing is sorted and graded—to those who sell vintage clothing to retail customers, primarily in the U.S. and Europe. On Wednesday, the London-based startup said it has raised $25 million in new venture capital funding. The new funding round, Fleek’s Series B, brings to $45 million the total funding the company has raised to date. The company declined to disclose its valuation following the funding.
The new round is being led by Burda Principal Investments—an investor in popular second-hand clothing e-commerce site Vinted—with participation from eBay, FJ Labs, and H14, alongside existing investors Andreessen Horowitz, HV Capital, and Y Combinator.
Fleek currently connects roughly 2,000 verified wholesale suppliers with more than 50,000 vintage clothing retailers across over 100 countries. Those buyers range from large retail chains and even some brands to vintage shops, and professional resellers. The platform stands behind the transactions, so buyers don’t have to worry about being scammed.
The scale of the underlying trade is enormous and largely invisible. Up to 24 billion secondhand garments move through the global supply chain each year, the company says, traveling from donation bins in cities like London, Paris, and New York to sorting centers around the world. Yet the $200-billion-plus industry remains stubbornly manual: garments are assessed by hand, graded to inconsistent standards, and traded through disconnected networks with little pricing transparency.
Despite this seemingly huge volume, demand for vintage clothing is currently outstripping supply. Vinted scrapped selling fees to attract sellers, and Depop and eBay followed. “Supply is the biggest constraint” in the industry, Arora said.
New regulations may be about to help. EU rules that went into effect last year require member states to collect textiles separately for recycling, while other new regulations about to come into force prohibit retailers from simply junking unsold inventory. Those requirements should push more clothing into the channels Fleek serves.
“We backed Vinted when secondhand fashion was still considered niche,” Julian von Eckartsberg, managing director for Europe at Burda Principal Investments, said in a statement. “Fleek is building the infrastructure the next generation of fashion will rely on.”
Fleek spent its first three years building the marketplace itself. About a year ago, the founders began asking how AI could change the business, and the result is Fleek Sort—a custom computer vision-language model trained on four years’ worth of the company’s own transactions.
The AI model takes in photos from an ordinary smartphone—the more photos the better, although the model can still work form a single photo of the front of a garment—and from that it identifies the item’s brand, style, and category, flags defects such as rips or faulty seams, and estimates both what the item will sell for and how long it will take to sell. “The detection of the item, what price will it clear for, and how long will it take to clear the item—that’s the nutshell,” Agarwal says. Sellers and buyers can still negotiate their own prices, particularly since buyers are often purchasing hundreds or thousands of pieces at a time.
Fleek Sort is already in use by graders in sorting hubs in Pakistan, India, and Dubai, with pilots launching in the U.K., Europe, and the U.S. The company says it built the model in-house out of necessity: at the volumes involved—Fleek’s largest partner moves some 600,000 pounds of clothing a day—relying on third-party models would have been prohibitively expensive. The team is now extending the technology from photos to video and studying how it might eventually be applied to conveyor belts in sorting facilities, though Agarwal cautions that the development of robots who can sort and grade garments as fast as humans currently do in these massive sorting facilities is still a ways off.
Guaranteeing authenticity is still a challenge. The model can spot obvious fakes, but Fleek also runs quality-check centers in Karachi and Delhi that inspect items from brands and categories its data has flagged as frequently counterfeited.
The new capital Fleek has raised will go towards continuing to develop Fleek Sort, expanding the company’s engineering team, and continuing to grow its network of buyers and sellers, Arora said.
