Can AI make fashion more sustainable?

KEY TAKEAWAYS Summary by the editors

1. AI can support sustainability in fashion mainly by improving forecasts and buying decisions, so that companies produce closer to real demand.
2. Since 19 July 2026 large companies may no longer destroy unsold clothes, clothing accessories and footwear in the EU, which raises the cost of overproduction.
3. The European Commission estimates that around 5 million tonnes of clothing, about 12 kg per person, are discarded in the EU each year, and only 1% of material in clothing is recycled into new clothing.
4. AI has its own footprint: the IEA estimates data centres used around 415 TWh of electricity in 2024, about 1.5% of global consumption, rising to around 945 TWh by 2030.
5. AI only makes fashion more sustainable if companies act on its insights, for example by buying less or differently, rather than using it only to sell more.

AI can make fashion more sustainable when it helps companies produce closer to real demand, use materials more efficiently and manage product and supply chain data for new EU rules. It cannot make fashion sustainable on its own: the gains depend on data quality and on decisions to make fewer unwanted products, and AI itself consumes energy.

Why is overproduction fashion's central sustainability problem?

Every garment that is made but never worn carries the full environmental cost of fibre, dyeing, sewing and transport without delivering any value. According to the European Commission, textile consumption in the EU has the fourth highest environmental impact after food, housing and mobility, and around 5 million tonnes of clothing are discarded every year, about 12 kg per person. Only 1% of material in clothing is recycled into new clothing.

The regulatory context is also changing. Under the Ecodesign for Sustainable Products Regulation, the ban on destroying unsold clothes, clothing accessories and footwear has applied to large companies since 19 July 2026, and medium-sized companies follow in 2030. Exceptions exist for items that are unsafe or damaged, among others. Surplus that once could be disposed of quietly now has to be sold, donated or prepared for reuse, and companies must report on what they discard.

How can AI help reduce overproduction?

The most direct lever is better planning. Machine learning models that forecast demand by style, colour and size can help buyers commit smaller initial quantities and replenish what sells, rather than placing large speculative orders. Size curve optimisation reduces the classic surplus of extreme sizes in some stores and shortages in others. Pre-order and wholesale order data can provide early demand signals before production is final.

These gains are real but conditional. A forecast only reduces waste if the business accepts lower volumes and possibly lower peak sales. If AI is used mainly to push more product through more channels, the net effect on resource use may be negative.

Read also

AI in denim: how finishing, laser technology and water use are changing

What are other sustainability use cases for AI in fashion?

• Digital sampling: 3D and generative design tools can reduce the number of physical prototypes shipped between brands and factories.
• Resale and recommerce: image recognition and automated pricing help resale platforms list second-hand items faster. The State of Fashion 2026 report forecasts the secondhand market to grow two to three times faster than the firsthand market through 2027.
• Sorting and recycling: sensor and machine learning systems identify fibre composition in used textiles, a precondition for fibre-to-fibre recycling.
• Data for regulation: AI tools can help collect and check supplier, material and origin data needed for the future Digital Product Passport and green claims rules.

What data does AI need to support sustainability goals?

Sustainability use cases rely on the same foundation as commercial ones: consistent product master data, reliable sales and stock history and structured supplier information. Forecasting models need to know which styles, colours and sizes sold at which price and in which channel, including what was left over at the end of the season. Without surplus and markdown data, a model cannot learn where overproduction occurs.

Material and supplier data are often the weakest link. Fibre composition, origin and certificates are frequently held in documents rather than databases, and impact data for specific materials is often generic. AI tools can help extract and structure this information, but they cannot replace primary data from suppliers. Companies that invest in this data now serve several goals at once: better planning, credible reporting and preparation for the Digital Product Passport.

What is the environmental footprint of AI itself?

AI runs in data centres, and its energy use is growing. The International Energy Agency estimates that data centres accounted for around 1.5% of the world's electricity consumption in 2024, or 415 TWh, and that this is set to more than double to around 945 TWh by 2030. Not all of this is AI, but AI is a major driver.

For a fashion company, the footprint of a forecasting model running a few times a week is small compared with the materials it helps to avoid. Large-scale image and video generation for marketing is a different case. A sober approach compares the energy and cost of an AI application with the waste it realistically prevents, and avoids claiming sustainability benefits that have not been measured.

Read also

How does AI detect defects in fabrics and garments?

So can AI make fashion more sustainable?

AI is a useful tool, not a strategy. It works best where it supports decisions that already point in a sustainable direction: producing less speculatively, keeping products in use and knowing what is in the supply chain. Companies that want measurable results should start with a baseline, for example surplus stock at season end or the number of physical samples, and track whether AI-supported decisions change it. It is equally important to check for rebound effects: if better forecasts mainly lead to more collections, more drops and more volume, the environmental gains disappear.

1. Measure current overproduction and surplus by category before any AI project.
2. Use forecasting to reduce initial buys and increase replenishment where lead times allow.
3. Build reliable product and supplier data, which serves both AI and regulation.
4. Report results in measured terms and avoid vague claims.

Frequently asked questions

How does AI reduce waste in fashion?

Mainly by improving demand forecasts and buying decisions, so that companies produce quantities, sizes and colours closer to what sells. It can also reduce physical samples through 3D design and support sorting for recycling.

Is AI bad for the environment?

AI uses electricity, and the IEA expects data centre consumption to more than double by 2030. Whether a specific use is worthwhile depends on how much waste or resource use it actually avoids compared with its own footprint.

Can fashion brands still destroy unsold clothes in the EU?

Since 19 July 2026 large companies are banned from destroying unsold clothes, clothing accessories and footwear, with limited exceptions such as unsafe or damaged items. Medium-sized companies will be covered from 2030.

Can a brand market products as sustainable because AI was used?

Only with care. Environmental claims must be substantiated, so brands should communicate measured outcomes, such as lower surplus, rather than presenting the use of AI itself as a sustainability benefit.

GuideThe complete guide to AI in the fashion supply chain and sustainabilityRead the complete guide