Is AI actually changing fashion, or just the marketing around it?

What happened

Here's the contrarian point worth knowing: most people assume “AI is disrupting fashion” means better forecasting, smarter design or automated trend-spotting. The data says otherwise. McKinsey and The Business of Fashion's 2026 State of Fashion report shows AI adoption in fashion is real, but concentrated almost entirely in customer service, product imagery and copywriting: the easy, low-risk applications, not the creative or strategic core of the business.

Meanwhile, the company actually proving AI-adjacent tech can reshape an entire operating model isn't using AI to design anything at all. Shein's LATR system tests new products in tiny batches of 100 to 200 units, watches real customer demand in real time, and restocks winners in as few as five days. AI sits underneath that, forecasting which products to scale, but the actual disruption is operational: smaller batches, faster feedback, automated reordering.

That's the argument worth having ready: the headline AI story in fashion right now isn't about smarter creativity; it's about who's rebuilt their operations fast enough to act on data before a trend dies.

Stats and figures

• More than 35% of fashion executives already use generative AI, mostly for customer service, images and copywriting, not design or forecasting (McKinsey and The Business of Fashion, 2025).
• Up to 90% of AI projects in fashion never scale past the pilot stage (McKinsey and The Business of Fashion, 2025).
• Shein restocks bestsellers in as little as five days, starting from batches of just 100 to 200 units (Modaes, 2026).
• The AI-in-fashion market is projected to grow from US$1.75 billion in 2025 to US$9.45 billion by 2030 (Research and Markets, 2026).

Business, operational and management takeaway

If you're asked what's changed operationally in fashion recently, the answer isn't “companies added AI”; it's “companies that redesigned their operating model around real-time data are winning, regardless of how much AI is actually involved.” Shein didn't succeed because it has better algorithms than competitors; it succeeded because it restructured production into small, testable batches so that data could actually change decisions fast. That's a transferable point beyond fashion: technology adoption without process change is why up to 90 percent of AI pilots stall; the bottleneck usually isn't the tool, it's the operating model around it.

Finance takeaway

If asked about capital efficiency or inventory risk, this is a strong, specific answer: small-batch, test-and-reorder models exist to solve fashion's oldest financial problem, deadstock and markdown losses from overproducing on a guess. Rather than committing large capital upfront on a forecast, companies like Shein commit small, test cheaply, and only scale spend once demand is confirmed. That's a genuinely useful framework to reference in any conversation about risk management or capital allocation under uncertainty, not just fashion specifically.

Marketing takeaway

If asked where AI is actually being used today, the honest and precise answer is: content generation and customer-facing tools, not strategy. That's worth saying plainly, because it shows you understand the difference between where AI adoption is loud (marketing) and where it's actually driving business results (operations and forecasting). The sharper point to make is that marketing's real opportunity isn't more AI-generated content, since that space is already crowded; it's using the demand data these operating models generate to make smarter decisions about what to promote and when.