Fashion is an industry with only one constant: change. Not long ago, fast fashion revolutionised retail by doubling production. Now brands face a different kind of disruption: products they have already sold find a second life without them. The infrastructure being built around that second life is what has allowed this market to evolve this fast. That infrastructure is AI, and the B2B wholesale layer – connecting secondhand supply with retail demand – is where that transformation is happening first and fastest.

What the data is telling us

Before examining the technology, it is worth understanding the scale of the behavioural shift already underway.

Across luxury and high-street brands, searches for secondhand are now consistently outpacing searches for new. For instance, comparing "Mango Vinted" or "Zara Vinted" to "Mango new collection" or "Zara new collection" shows secondhand queries running 4-6 times higher in search volume throughout 2024 and into 2026, with Vinted searches hitting peak popularity in mid-2025 while new collection searches remained a fraction of that.

At luxury level, Hermès vintage searches significantly outpace new bag searches – more than doubling them at their recent peak – while even Chanel, where new and vintage searches had tracked closely for years, saw vintage interest reach near-parity with new in early 2026.

Google Trends measures search interest on a scale of 0 to 100, where 100 represents the peak popularity of a search term during the selected period.

What this data is showing is that consumer intent around secondhand has fundamentally shifted – people are starting their fashion journey at preloved, not arriving there as a fallback – and for brands, this is a signal about where they need to be present and what infrastructure they need to build to participate in that market.

Not only that, but how people talk about secondhand has shifted just as what they search for. Before 2020, the dominant language was negative: charity shop, hand-me-down, compromise, stigma; by 2024-2026, that vocabulary had been replaced almost entirely by the language of identity, aspiration and discovery: preloved, vintage find, curated, unrepeatable – confirming the shift in cultural perception, according to my analysis of media coverage, market reports and consumer communities before and after Covid.

Globally, secondhand clothing sales are forecast to hit 289 billion dollars this year – 105 percent growth from 2021 – expanding at twice the pace of the overall clothing market, according to the ThredUp Annual Resale Report 2026. And the AI layer appears to have accelerated this growth further. Search interest in "AI shopping" was effectively zero across all markets until mid-2024, began building through late 2024 as generative AI tools entered the mainstream, then surged from June 2025 onwards, growing over 3,000 percent in two years before sustaining near-peak levels.

But the most striking finding in the data is the correlation between the two. Both "AI shopping" and "secondhand clothing" were effectively flat for four consecutive years. Both began moving in the same month – July 2025 – surging simultaneously in August 2025 and sustaining elevated levels ever since. The data suggests that AI was not a mere coincidence with secondhand acceleration, but may have played a significant role in enabling its growth at scale.

Why secondhand cannot scale without AI – the B2B infrastructure problem

The consumer shift is evident and supported by data. What is less visible – and what matters more commercially – is the infrastructure problem that has historically made secondhand so difficult to scale – and why AI is not an optional layer here but a structural requirement.

The resale market is notoriously complex. Platforms must manage vast, unpredictable inventory that ranges in quality, size and authenticity, at a complexity that has no equivalent in new retail. To understand how all looks like operationally, I spoke to Sanket Agarwal, co-founder of Fleek, one of the fastest-growing AI platforms in secondhand wholesale and one of the primary sourcing platforms for Vinted resellers. He helped me grasp exactly why the B2B layer of this market required a fundamental technological rebuild.

The core problem, as Sanket explains, is one of scale with no equivalent in the traditional fashion business: "In classic retail, shops usually have a few defined SKUs, but in secondhand there is such a large variety of eras, brands, styles and wear that it leads to millions to billions of SKUs – essentially each piece is unique even if it is the same brand SKU". And that uniqueness is exactly what makes every single item so difficult to categorise, price and match to a buyer. Unlike Amazon or Asos, where AI operates on structured, consistent product catalogues, secondhand wholesale has no shared product data, no standardised SKUs, no taxonomy linking condition to buyer intent, which is what has made it so hard to scale, and precisely what makes AI so transformative here.

Beyond the uniqueness issue, resellers struggle with variables such as lighting inconsistencies in photos or wear patterns. Authentication requires human expertise at the final stage, even when AI handles the initial scan. Pricing is a constant calibration problem. And layered over all of this, the traditional wholesale secondhand supply chain is not only messy, outdated, and incredibly complex, but it was constructed on personal relationships – trust between buyers and sellers built over years of informal dealing.

This is where Fleek comes in. The platform was founded in November 2021, born out of a problem co-founder Abhi Arora discovered in Brick Lane, London's vintage fashion hub, during the pandemic: the secondhand supply chain was built on chaos. Preloved clothing collected in western countries – around 90 per cent of all donations globally – is shipped in bulk to sorting centres in Pakistan, India and across Africa, where it is hand-sorted and sold back to western resellers, whether secondhand shops or the wholesalers supplying Vinted sellers. The hand categorisation was tedious and inaccurate, and the more granular and accurate the grading, the better the clothing sells – so the stakes of getting it wrong were high. Resellers often had no idea what they were getting, transactions happened over WhatsApp groups and informal networks where trust was everything and transparency almost nothing, and only a very small fraction of those donations ever made it back to being resold in western markets. The system was inefficient and structurally broken.

As Abhi mentioned in an interview for The Industry.Fashion, the platform was built to work directly with these wholesale suppliers, getting inventory listed, categorised, priced and sold through Fleek's own system. A reseller in London, Paris or New York can browse curated bundles or hand-pick items via video call and place an order. That order goes through one of Fleek’s quality control centres, where items are checked for quality and authenticity, and then it is dispatched to the buyer.

How does AI materialise on Fleek?

Fleek rebuilt the entire sourcing experience from the ground up. "At Fleek we had to reimagine our entire search and discovery experience, which is now driven by AI-first search technology. We are leveraging CLIP embeddings* to define semantic properties of fashion such as 'embellishments' or 'mushroom print' – a much harder task for pre-LLM** models". A buyer can now search by mood, style or aesthetic reference rather than product specification – the way people actually think about secondhand. Not only that, but the platform gives a pricing estimate, handles transactions, streamlines the supply chain, manages refunds and provides trust on both ends. The commercial results are visible: “more than doubling sales from 2024 to 2025”, connecting over 10,000 resellers with more than 1,000 wholesalers across 70 countries, having raised 50 million dollars in total funding, backed by investors including Andreessen Horowitz and Y Combinator. Sanket is direct on the opportunity for retailers still sitting on the sidelines: "Today, one out of every two individuals is looking for secondhand – it is good for the environment and good for business. We are already seeing Fleek's customers sell secondhand and first-hand clothes side by side".

*(CLIP stands for Contrastive Language-Image Pre-training – it is a model developed by OpenAI that was trained on hundreds of millions of image and text pairs simultaneously, so it learned to understand the relationship between visual content and language. Traditional image recognition asks "what object is this?" – it recognises a bag, a shoe, a jacket. CLIP goes further – it understands the feeling and character of what it sees. So instead of just recognising "jacket," it can understand "oversized 90s Japanese streetwear jacket with acid wash" or "mushroom print" or "embellished evening wear”).

**(LLM stands for Large Language Model – the type of AI that powers tools like ChatGPT, Claude and Gemini).

What this means for your business

Secondhand existed before AI – but without the infrastructure to source, authenticate, grade and price inventory at scale, the demand had nowhere to go efficiently. What Fleek proves at the wholesale level is that when you remove the structural friction, the commercial volume follows.

That said, the challenges have not disappeared. Logistics remain complex and costly – secondhand items cannot be restocked, and the quality of how an item is displayed still affects grading accuracy and drives returns. Consistency at scale is difficult to guarantee even with computer vision. Authentication at the final stage still requires human expertise. Margins across the industry remain under pressure, and most large resale platforms are still on the path to profitability rather than there. AI improves all of these problems significantly – but it does not eliminate them, and brands entering the space without a clear operational strategy are likely to find it harder than the market numbers suggest.

What AI does is make those challenges manageable – not disappear entirely, but structured enough to build a scalable business on top of. It is now operating across every layer of the resale stack – at sourcing, platforms like Fleek use computer vision and semantic search to make bulk secondhand inventory discoverable at scale; at the brand level, Resale-as-a-Service platforms like ThredUp handle intake, grading, photography, pricing and fulfilment using AI automation, making it possible to launch a resale programme without building anything from scratch. Authentication, historically the biggest barrier to trust in secondhand, is being handled by computer vision models that triage suspect items before human experts review them. Dynamic pricing algorithms replace the guesswork that made secondhand margins unpredictable. The commercial case is already proven: Aymeric Déchin, CEO of Faume, told Vogue Business that customers who use a brand's trade-in service show 20 percent lower churn compared to those who do not. Collectively, these capabilities do something more significant than optimise individual transactions; they normalise secondhand as a reliable, trustworthy channel for both brands and consumers, and Fleek is only one example of it.

The regulatory layer is accelerating all of this. The EU's Ecodesign for Sustainable Products Regulation requires every fashion brand selling in Europe to attach a Digital Product Passport (DPP) to every garment from 2028 – a machine-readable identity recording materials, origin and ownership history. For AI, this is transformative: a garment with a passport can be authenticated, graded and priced automatically, because the data is already there.

One-third of industry executives called resale a priority for 2026, according to BoF/McKinsey State of Fashion 2026. That gap – between where the consumer already is, how AI is accelerating it, and where most of the industry is still focused (new) – is the opportunity, and it is closing fast. If you are still treating secondhand as secondary – or AI as optional – the data is clear: you are not behind the trend, you are behind the consumer.

Sources:
-The Guardian — Sarah Butler, “Secondhand Clothes Sales Forecast to Hit $289bn as AI Helps Shoppers Find Deals,” 2 April 2026.
-Retail Dive — Tatiana Walk-Morris, “US Resale Market Expected to Surpass $78 Billion by 2030,” 3 April 2026.
-Adobe — Vivek Pandya, “Generative AI-Powered Shopping Rises with Traffic to Retail Sites,” 21 August 2025.
-Modaes — “Inditex 2025 results: eight critical takeaways to watch,” C. De Agenlis / T. Alonso, 12 March 2026.
-Retail Boss — “Inditex Q1 2026 Results: Zara’s Best Quarter Yet,” Jenel Alvarado, 3 June 2026.
-Vinted Newsroom — “Financial Results 2025,” 2026.
-UNECE (United Nations Economic Commission for Europe) — UNECE and ECLAC propose measures to reduce environmental and health impacts of global trade of second-hand clothes’, 15 July 2024
-TheIndustry.fashion — “The Interview: Co-founder Abhi Arora on Building Second-Hand Wholesale Marketplace Fleek,” Camilla Rydzek, 16 April 2026.
-WWD — “How Vestiaire Is Using AI to Scale Its Business and Improve Customer Service, by Lisa Lockwood, June 14, 2024.
-The Impression — Vestiaire Collective Expands AI Capabilities With New Executive Hires.
-Vogue — “The Innovations Driving the Resale Renaissance,” byt Maghan McDowell November 19, 2024.
-GWI — How the circular economy is transforming fashion: Sustainable trends & insights by Stephanie Harlow, Senior Trends Analyst.
-McKinsey & Company — The State of Fashion 2026: When the rules change, November 17, 2025 by -Trellis — Circular boom(let): Resale and reuse surge as new fashion turnover slows, by Elsa Wenzel November 21, 2025 (Updated on November 24, 2025)
-Barclays Insights — The pulse of fashion: How the growth of the resale market has changed the game for retailers, by Melissa Pendlebury and Isabella Clough, April 2, 2026.
-Fashionista — “Fashion Resale Tech: AI and the Future of Evolution,” by Emma Raydar, June 4, 2025.