AI in sportswear: how it shapes performance product, community and demand
Sportswear brands combine technical product, athlete storytelling and huge launch volumes. AI helps in design, content and demand, but performance claims and athlete data raise the bar.
KEY TAKEAWAYS Summary by the editors
- In sportswear, AI matters in three places: performance product creation, personalised content for communities of athletes and fans, and demand planning for large, launch-driven ranges.
- Nike used generative AI in its A.I.R. project in 2024, generating hundreds of images from athlete preferences before designers refined concepts with 3D sketching and 3D printing, according to Dezeen.
- PUMA reported in September 2024 that it uses Google Cloud's Imagen on Vertex AI to create localised product imagery, citing better click-through rates and faster time to market.
- Performance claims must stay grounded in physical testing: generative AI can propose shapes and stories, but it cannot certify cushioning, grip or thermal performance.
- A sensible first step for a sportswear brand is to connect product, content and sell-through data by sport and region, because personalisation and forecasting both depend on that link.
AI in sportswear is used to speed up performance product design, to personalise content for communities organised around sports, and to forecast demand for launches and core franchises. The segment differs from general fashion because products carry functional claims, brands depend on athlete and community storytelling, and a few franchises drive large volumes, so errors in either credibility or demand are expensive.
Why does AI matter differently in sportswear?
Sportswear sits between fashion and engineering. A running shoe or football boot must deliver measurable performance, while the same brand sells lifestyle products driven by trends and culture. Demand is shaped by sports calendars, major tournaments, athlete moments and product drops, and core franchises are replenished season after season. McKinsey and The Business of Fashion describe sportswear in their State of Fashion 2026 report as a bright spot in China, where wider fashion demand faces headwinds.
This mix creates three distinct AI needs: faster iteration on technical product, content that speaks to very different communities (a trail runner, a padel player and a sneaker collector), and planning that separates hype-driven launches from steady core business.
What are the main AI use cases in sportswear?
| Use case | Why it matters in this segment | Example (only if verified) | Maturity |
|---|---|---|---|
| Generative concept design | Shortens the path from athlete input to prototype for performance product | Nike A.I.R. project, 2024: AI-generated images refined with 3D sketching and 3D printing (Dezeen) | Emerging |
| Localised product imagery | Same product must appeal to different sports and regional contexts | PUMA uses Imagen on Vertex AI for region-specific product imagery (Google Cloud, September 2024) | Emerging |
| Personalised recommendations and search | Large catalogues across many sports need relevant discovery | PUMA planned to extend Vertex AI Search for Retail to more subsidiaries (Google Cloud, 2024) | Established |
| Demand forecasting for launches and core | Hype launches and evergreen franchises behave differently | No verified detail published | Established |
| Generative copy and customer service | High content volume around launches and sizing questions | No verified sportswear case used here | Established |
| Simulation of performance properties | Could reduce physical prototyping cycles | No verified case used here | Experimental |
How is AI used in performance product creation?
Nike offers the clearest public example. In May 2024, Dezeen reported on Nike's A.I.R. project, in which the company worked with 13 elite athletes. Designers fed prompts based on athletes' preferences into AI models that generated hundreds of images, then refined the concepts using 3D sketching and 3D printing to reach prototypes. Nike executive John Hoke described the approach as converging athletes' ideas with the designers' own imagination.
The lesson for other brands is the division of labour. Generative AI was used to widen the range of ideas quickly at the start; engineering, material choice and testing remained human and physical. In sportswear, a concept that looks fast is not the same as a product that is fast. Any performance claim must come from lab and athlete testing, not from the model.
How does AI help sportswear brands serve communities?
Sportswear marketing speaks to many niche communities at once. Generative imagery allows the same product to be shown in contexts that fit each one. PUMA said in September 2024 that it uses Imagen on Vertex AI to create dynamic, personalised product imagery adapted to regions; Google Cloud's announcement gave the example of a lifestyle shoe shown on the streets of Ginza or a trail running shoe in the foothills of Mount Fuji. PUMA reported improved click-through rates, faster time to market for campaigns and a higher average order value, without publishing the underlying figures.
Common community-focused applications include:
- Sport-specific landing pages and recommendations based on browsing and purchase behaviour.
- Localised campaign variants for regional sports and climates.
- Assistants that answer technical questions on fit, cushioning or fabric, grounded in product data.
- Content tagging so that product imagery and stories can be found by sport, surface and use.
How does AI improve demand planning in sportswear?
Sportswear planning has to handle two very different patterns. Core franchises behave like replenishment business, with stable demand by size and colour. Launches and collaborations behave like events, with intense demand spikes and a risk of either selling out instantly or leaving heavy stock. Machine learning forecasting can combine sell-through, web traffic, sign-ups for launches, sports calendars and regional weather to separate these patterns, but it needs clean history at product, size and channel level.
Sportswear brands that sell heavily through wholesale partners, such as specialist sports retailers, face an extra challenge: they often see sell-in but not timely sell-out. Agreements to share sell-out data with key accounts can do more for forecast accuracy than a more advanced model.
What data and risks are specific to sportswear?
Data specifics include technical product attributes (drop, stack height, weight, membrane, compression level), sport and use classification, athlete and team assets with their usage rights, and fitness or activity data where apps exist. Each brings its own risks:
- Performance claims: AI-generated copy can overstate benefits; technical claims need verification and legal review.
- Athlete likeness and rights: generated images or voices of athletes require explicit contractual permission.
- Health and activity data: data from fitness apps can reveal health information and should be handled under strict privacy rules.
- Hype distortion: forecasting models can overreact to social buzz around a launch that never converts.
- Content sameness: generated campaign imagery can drift towards generic visuals that weaken brand identity.
What should a sportswear brand do first?
Begin by linking product attributes, content assets and sell-through data by sport and region. Personalisation, generative content and forecasting all rely on knowing which product is for which athlete and how it actually sells. Then choose one use case with a clear measure: generative product imagery for a single category, measured on time to market and conversion, or a forecast pilot for a core franchise, measured on forecast error and stock-outs.
Keep performance claims and athlete representation under human and legal control from the start. In sportswear, credibility with athletes and communities is the asset that AI must protect, not spend.
Frequently asked questions
How do sportswear brands use AI?
They use it for concept design, localised product imagery, personalised recommendations, customer service and demand forecasting. Verified examples include Nike's A.I.R. design project with generative AI and PUMA's use of Google Cloud's Imagen for product imagery. Results are rarely disclosed in detail.
Did Nike design shoes with AI?
In 2024 Nike used generative AI in its A.I.R. project to create hundreds of concept images based on the preferences of 13 elite athletes. Designers then refined these with 3D sketching and 3D printing into prototypes. The AI supported ideation; engineering and testing remained human.
Can AI predict demand for sneaker launches?
AI can combine signals such as sign-ups, web traffic and past launches to improve forecasts, but hype-driven launches remain hard to predict. Social buzz does not always turn into sales, so planners should treat model output as one input and keep buffers and allocation rules.
What are the risks of generative AI in sportswear marketing?
The main risks are overstated performance claims, unauthorised use of athlete likeness and generic visuals that weaken the brand. Brands should ground copy in tested product data, secure rights in athlete contracts and review generated content before publishing.
One edition every weekday morning. Read in five minutes. Free for industry professionals.
