True Circular Economy
Only 6.9% of the materials entering the world economy are secondary (recycled) material - and the share has been falling.
Open in the interactive tree →A circular economy means products last longer, get repaired, and their raw materials come back in the same quality. Today material is often downcycled or burned; alloys, composites and electronics are hard to separate. Growth consumes more primary raw material than recycling replaces.
As of October 2026
The Circularity Gap Report 2025 found that 6.9% of the materials entering the global economy in 2021 were secondary materials, down from 7.2% in 2018. The 2026 edition (April 2026, Circle Economy and Deloitte) estimates the avoidable value lost to linear material use at 25.4 trillion euros per year, almost 31% of global GDP.
What is missing
- Products designed from the start for disassembly and repair
- Sorting and separation processes for mixed materials, alloys and electronics at industrial scale
- Cheap energy for recycling, and recycled materials that match primary materials on price
- Data on material composition (product passports) and binding take-back systems
- Policy that charges raw-material use and waste at their true cost
Becomes possible once solved
- An economy without growing raw-material extraction
- Raw-material independence for countries without their own mines
- Much less waste in soils and oceans
- Closed loops for battery metals and rare earths
Open steps
- Sorting mixed waste streams High AI leverageSeparate mixed plastics, metals and electronics into pure streams at industrial speed and low cost.
- Recycling mixed plastics, textiles Medium AI leverageRecycle mixed or dirty plastics and blended textiles back to monomers at industrial scale with enzymes or catalysts.
- Keeping alloys and metals pure Medium AI leverageIdentify alloy grades and trace contaminants in scrap fast enough to avoid down-cycling.
- Robotic disassembly of products Medium AI leverageTake apart varied, damaged or unknown products with robots that adapt to each item.
- Material data for products Medium AI leverageCreate and read product material passports, estimating composition from labels, spectra and images when records are missing.
Where AI could help
Medium AI leverage. AI sorting, disassembly and material data close real technical gaps; prices, product design and policy still keep recycled shares low.
- Identify and pick materials on conveyor belts with vision models, raising purity of sorted streams
- Design enzymes and catalysts that depolymerize mixed plastics and textiles
- Design products for disassembly and read material composition from labels and spectra
- Optimize reverse logistics and match secondary materials to demand
Shown so far
- In April 2022 a machine-learning-engineered enzyme (FAST-PETase, Nature) broke down untreated post-consumer PET from 51 products, in some cases to monomers within 24 hours. source
- In January 2021 AMP Robotics said (company claim) its AI sorting robots pick about 80 items per minute and had processed over a billion recyclables in installations in more than 20 US states. source
Prerequisites
Unlocks
- A World Without Waste2050s?