ITMF and RWTH Aachen report: fragmented data blocks textile AI

Textile manufacturers should expect end-to-end AI workflows only within two to three years, according to a white paper published in October 2026 by the International Textile Manufacturers Federation (ITMF) and RWTH Aachen University's Institut für Textiltechnik (ITA).

The report, AI in the Textile Value Chain: A Strategic Guide to Artificial Intelligence for the Global Textile Industry, draws on the ITMF and ITA's AI in the Textile Value Chain webinar series. It argues the industry's progress will depend on "coordinated, industry-wide capability-building".

What is holding adoption back?

The report describes textile manufacturing as "not under-digitalised, but digitally fragmented". Decades of digitalisation have left data scattered across disconnected systems. Companies frequently overestimate the amount of usable, structured data available to support AI projects, it states.

ITA director professor Thomas Gries said: "Companies need a dedicated task force to embed data integration into their day-to-day and strategic technical and business processes. Partnering with external experts on a transparent assessment – a data or automation waste walk – and on effective implementation is not only decisive to future success, but essential to survive!"

The message for factory management is blunt: no AI project succeeds on top of fragmented, poor-quality data.

What is driving the pressure?

The paper identifies four forces pushing AI adoption across the textile supply chain:

• demographic change and shortages of skilled workers;
• market volatility;
• increasing customer and regulatory demands for quality;
• rising requirements for traceability and sustainability.

Several applications already deliver measurable results, according to the report. These include computer-vision systems for fabric and yarn inspection, predictive maintenance, energy optimisation and demand forecasting.

The paper also records a shift in the technology itself. AI is moving from systems that merely assist decision-making towards systems able to coordinate or execute defined tasks under human supervision. Agentic AI is beginning to manage multi-stage processes such as sourcing, while causal AI can adjust production settings directly.

Where does AI risk making things worse?

Participants contributing to the research warned that AI can compound, rather than resolve, underlying problems in incomplete or poor-quality data. Deployment on unreliable datasets therefore carries operational risk, not just cost.

The paper argues businesses should establish dedicated teams to integrate data into technical and commercial processes before attempting wider AI deployment. Trust and accountability were also identified as requirements. Industry participants cited four conditions for using AI in production and supply-chain activity:

• control of sensitive information;
• cyber security;
• transparency about system limitations;
• retained human responsibility.

ITMF director Olivier Zieschank said: "AI is not a shortcut around the structural challenges our industry faces, but it can help companies manage complexity and make the most of scarce expertise. Understanding where AI genuinely helps is a learning process. That is why exchange through associations like ITMF matters so much."

What should suppliers do now?

The paper recommends companies begin with contained use cases rather than broad implementation programmes. It also proposes shared data standards, data spaces that allow originators to retain ownership of their data, and sector-specific tools such as a controlled "Textile GPT", aimed particularly at smaller and medium-sized businesses.

For sourcing and production executives, the decision the report forces is sequencing: audit and integrate data foundations first, run narrow pilots with proven returns such as vision inspection or predictive maintenance, and defer end-to-end automation until the two-to-three-year window matures.