It started on Twitter. More than a decade ago, professors at the Massachusetts Institute of Technology (MIT) worked on an algorithm that could predict what would trend on Twitter four hours in advance. They then moved to cars. Analyzing the historical sales of a local dealership in the Midwest—from the colors of the vehicles to the time they were sold—their software predicted which ones would sell the most. They would publish these findings in a prestigious academic journal in 2011. The underlying technology would later be called one of IT’s Computer Science and Artificial Intelligence Laboratory’s 50 top innovations ever made.
Like many great ideas, it became a startup. Two of the MIT professors—Vivek Farias and Devavrat Shah—commercialized the technology into Celect, an AI company that could anticipate what customers would select, thanks to predictive analytics. “It’s very complicated behind the scenes,” Farias said in 2015. “But at the end of the day, it’s doing something very simple: tracking who wants to buy what and matching the product with the person.” It showed promise, securing millions in funding before being acquired by a multi-billion-dollar company that has been buying tech startups left and right to usher in its brand’s own digital transformation: Nike.
Thus, from viral tweets and cars on sale, the technology found itself in the hands of one of the most popular fashion brands worldwide in 2019. This gave the apparel and footwear giant access to a machine learning technology, a subset of AI, that could predict what customers would want—down to color, style and size—and ultimately manage the company’s sprawling inventory better. In an earnings call that year, Andy Campion, then-chief financial officer at Nike, had high praise for Celect. He said the acquisition accelerated the company’s digital demand-sensing capabilities “by at least three years.”
This was put to the test not soon after. Nobody anticipated the Covid-19 pandemic that disrupted supply chains in 2020 and kept nearly the entire world under lockdown. Most Nike-owned stores, or 90 percent, had to close for roughly eight weeks. But riding the boom of e-commerce, more shoppers started buying Nike online, increasing its digital demand. Nike said it saw its digital sales increase 75 percent in that quarter, with strong double-digit growth across all markets. It was responsible for 30 percent of Nike’s total revenues. In an earnings call in June 2020, Nike’s CEO and president at the time John Donahoe credited the Celect acquisition for allowing Nike “to get the right product as close to the consumer in the right time.”
All of this was before AI entered the public lexicon. The current popularity of AI could be traced to November 2022, per historical global data on Google Trends, which was about the same time OpenAI initially launched the groundbreaking ChatGPT. Before this, the public interest was low. After that point, it kept climbing. In the first quarter of 2026, AI-related web searches even reached 100, the highest score possible on Google Trends, indicating peak popularity.
So much has already changed since then. Just at the start of the decade, brands still had to find the best keywords to appeal to Google’s search algorithm. Now in 2026, AI agents are becoming the Internet’s newest gatekeepers, and companies have to find a way to be discovered by AI before they can even be seen by humans on the internet.
Nevertheless, data suggests that Nike’s AI strategy, fueled by its earlier tech acquisitions, is headed in the right direction. It is one of the very few fashion brands that managed to not only stay visible in the eyes of AI agents, but more importantly, stay on top. Nike gathered more mentions by AI than any other fashion brand in the first four months of the year, according to a landmark AI visibility study by Adobe company Semrush, which analyzed 126 million AI search prompts in the U.S. across different categories.
The same could not be said for the rest of the industry. It’s this new environment that prompts companies to become “AI mature” to stay competitive.
MIT’s four stages of AI maturity
AI maturity is a measure of AI adoption across an entire organization. MIT mapped out four stages of this, creating an industry-agnostic model that essentially starts with preparation, which includes educating the workforce and making company data accessible. This is followed by pilot projects, which, if successful, would then be scaled up across entire departments, until finally the company reaches the pinnacle of maturity with entire ways of working reimagined for an AI-ready future.
Generally speaking, most companies have trouble scaling up their pilot AI projects, said Stephanie Woerner, principal research scientist at the MIT School of Management, who co-developed the four stages. Not all companies have to reach the final stage of AI maturity, she said, but at some point, good business practice would require better efficiencies, and therefore companies would still have to mature their AI at scale.
This seems true for businesses in fashion, too. There is no exhaustive review of AI maturity in the industry, but surveys suggest that many companies hit a wall with their pilot projects, leaving them at stage two. A Sourcing Journal survey found over half of companies are still running AI pilot programs. About one out of five companies have scaled up their use of AI, and only 2 percent went as far as integrating AI in their decision-making processes.
There is also the question of how much AI should be in fashion. Should it also be involved in design? The range of answers fashion leaders give suggests that the industry is not necessarily on the same page on this issue. There are companies that have already incorporated AI in their design process to brainstorm, such as Gap and H&M. Shein, another example, uses machine learning to track ever-changing customer preferences in real time, fueling its ultra-fast-fashion business.
But there are also those who draw the line, such as Prada Group CEO Andrea Guerra, who acknowledged AI as a “fantastic tool” for improving efficiencies and customer experience. “Creativity is heart, brain, lungs, people, culture—it’s not AI,” he told CNBC last July. “It’s a huge accelerator and it’s a huge simplifier. Please, not in creativity.”
In essence, it’s technological complexity and human resistance that stand in the way of meaningful AI adoption. This is especially complicated for legacy companies, which have already “built systems on top of systems” even before AI was added to the equation. Now they are being asked to reimagine a formula that has worked for them for so long. “It’s difficult,” Woerner told Sourcing Journal. “To go and really get value from your AI, you have to do things differently, and it’s very hard for people to unlearn what has worked in the past.”
The industry is fashionably late
If AI adoption was a race, then Nike started running a long time ago. But in all fairness, it was not until recent years that the fashion industry realized it has to even be in this race to begin with. “Fashion was a little bit behind on AI adoption if we compare it to other industries,” Sebastian Bak, managing director and partner at Boston Consulting Group, told Sourcing Journal. “The use cases or the opportunities that were very obvious when AI exploded seemed to be a little bit less relevant in such a creative and dynamic industry, as opposed to financial institutions and a few others.”
Part of the reason for this lag has to do with what the fashion business inherently is. At its core, the industry capitalizes on desirability, explains Bak, who is part of a BCG leadership team that focuses on AI and retail. It’s what pushes people to buy garments, may it be a luxury Gucci dress or the trendy clothes that sell cheap on Shein. It can also be a color preference or a style that feels personal, but ultimately, it’s desire that draws people into fashion. In this context, AI seemed like a cold instrument that wouldn’t go well with retail therapy.
Nevertheless, money talks, and in the case of the fashion industry, the biggest potential from AI adoption is not just better internal efficiencies but a better bottom line, too, Bak said. The biggest potential for returns on investment (ROI) is seen in how AI is used in pricing, supply chain management and marketing. Fashion brands are even spending more money on AI now, with AI investments doubling so far this year, Bak said, although he deferred from sharing actual figures. “I would say several companies are getting extremely high ROIs out of this, and there are some companies that are getting no value—and both are true.”
Lessons from fashion’s past ‘Panaceas’
Companies across industries, including fashion, throw the word AI and its variations around with almost wanton abandon, describing their operations as AI-enabled or AI-powered, or even declaring themselves AI natives, which refers to companies whose operations are so entrenched in AI that they literally cannot operate without it. It has even given rise to a new deceptive marketing practice called AI washing, similar to greenwashing. As loud they may be, they still have little to show for it. A 2025 MIT report found 95 percent of the companies it reviewed were still “getting zero return,” despite investing $40 billion in generative AI.
In 2021, Dr. Byuongho Ellie Jin, a professor at North Carolina State University, co-wrote a paper about how the 4th industrial revolution was affecting the fashion industry. Whereas previous industrial revolutions focused on improving production efficiencies in factories to cater to the most people, AI and other technologies were bringing forth a new revolution that prioritized hyper-personalization, getting to know the individual customer like never before. They then issued a warning.
“When technologies are first introduced, they are often treated like a panacea emphasizing the merits,” the paper read. For example, when dotcom emerged in the early 2000s, the researchers noted how developing and operating a website was considered a huge skill. But now, everyone can make a website. Companies had to learn the hard way that merchandising skills that meet consumer needs are more important than a website. “The global fashion industry should learn from this history,” it said.
Half a decade later, Dr. Jin told Sourcing Journal she did not expect AI would develop this fast when she worked on the paper. But in the end, like the advancements that preceded it, AI is a tool. It’s not about having it. “It’s how you use it,” she said.
This article was published in Sourcing Journal’s Fall Summit report. Click here to download.