We Need Better Recycling to Sustain It
Artificial intelligence is often discussed as though it exists entirely in the digital world. We see software, search results, generated images and automated systems, but rarely the physical infrastructure operating behind them. The AI boom is driving rapid growth in data centers, energy infrastructure and the use of plastics throughout the digital economy.
In reality, AI is built from steel, copper, concrete, silicon—and plastic.
The rapid expansion of artificial intelligence requires enormous data centers filled with servers, cables, cooling systems, electrical equipment and specialized processors. Plastics appear throughout this infrastructure: in cable insulation, connectors, equipment housings, cooling components, piping, protective films, packaging and countless other applications.
That should not be surprising. Plastics have become essential because they are lightweight, durable, moldable, insulating and inexpensive. Those same properties make modern electronics, transportation, medicine, food preservation and communications possible.
The question is therefore not how to eliminate plastics from the AI economy. It is how to use them more intelligently, recover more of their value and develop better solutions for the materials that conventional recycling cannot handle.
AI Has a Physical Footprint
The growth of AI is placing unprecedented demands on physical infrastructure. The International Energy Agency projects that global electricity consumption by data centers will increase from approximately 485 terawatt-hours in 2025 to around 950 terawatt-hours by 2030. Electricity use by AI-focused data centers is expected to grow considerably faster than data-center consumption overall.
Meeting this demand will require much more than additional computing equipment. It will require new power generation, transmission, substations, backup systems, cooling capacity and supporting industrial infrastructure.
As PowerGen Insights has emphasized, the digital economy cannot expand independently of the physical energy system supporting it. Reliable computing ultimately depends on reliable electricity.
The same principle applies to materials. Every new data center contains large quantities of highly engineered products designed to manage electricity, heat, water and information. Plastics are indispensable within many of those systems because alternative materials may be heavier, more expensive, less corrosion-resistant or unable to provide the same electrical and thermal performance.
Plastic is not an accidental feature of the AI revolution. It is one of the materials making that revolution possible.
Faster Innovation Can Also Create More Waste
The challenge appears at the other end of the equipment’s useful life.
AI is driving demand for increasingly powerful processors and specialized servers. As technology advances, older equipment may be replaced even when some of its components remain functional. That creates additional electronic waste containing valuable metals as well as circuit boards, wires, insulation, composites and multiple types of plastic.
The world generated approximately 62 million metric tons of electronic waste in 2022, but only about 22% was formally collected and recycled. That waste includes not only metals and glass, but also the plastics used to encase, protect and connect electronic components.
Researchers have specifically warned that the expansion of generative AI could contribute to a substantial increase in electronic waste unless circular practices are incorporated throughout the AI hardware value chain.
This does not mean technological development should stop. It means material recovery systems must advance alongside it.
Plastics Are Useful Precisely Because They Are Durable
Public discussions about plastic sometimes treat its durability as a design failure. In many applications, however, durability is the entire point.
We want cable insulation that does not quickly degrade. We want cooling-system components that resist corrosion. We want electronics protected from moisture, impact and heat. We want lightweight materials that can be manufactured consistently and transported efficiently.
Modern life depends on those characteristics. Attempting to eliminate plastics indiscriminately could create new costs, greater material consumption and, in some cases, inferior environmental outcomes. As Real Cycle has argued from the beginning, the problem is not that plastics exist—it is what happens to them after they are discarded.
The real problem is not that plastics are useful. It is that our recovery infrastructure has not kept pace with their use. The result is a system in which less than 9% of plastic waste is recycled globally, with even lower rates reported in the United States.
Clean, uniform plastics can often be mechanically recycled, and that should remain an important part of the solution. But many materials are mixed, contaminated, degraded, laminated or combined with additives and other substances. These characteristics can make them unsuitable for conventional recycling. Our overview of how mechanical, advanced and other forms of recycling work explains why these different waste streams cannot all be processed in the same way.
A credible circular economy must recognize those differences rather than pretending that every plastic item can enter the same recycling bin and emerge as an equivalent new product.
AI Can Become Part of the Recycling Solution
The encouraging news is that artificial intelligence may also help solve some of the problems its growth will intensify.
AI-assisted vision systems can improve the identification and sorting of materials. Sensors and analytical models can help distinguish polymers, detect contamination and characterize complex waste streams. Predictive systems can improve equipment performance, logistics and maintenance. Better data can also help companies trace materials from production through use, collection and recovery.
At an industrial level, intelligent systems can help facilities adjust operations based on changing feedstock composition, energy prices, equipment conditions and product requirements. This is part of the broader transition from basic waste processing toward more responsive and efficient circular infrastructure.
For clean and separable plastics, better identification and sorting can improve mechanical recycling. For difficult materials that cannot be economically returned to useful products through conventional methods, advanced recycling and resource-recovery systems can provide additional pathways. Companies such as Convergen Energy demonstrate how certain non-recyclable industrial by-products—including paper and plastic materials—can be diverted from landfill and converted into engineered fuels for industrial energy users.
These approaches should not be treated as competitors. Different waste streams require different solutions.
Building a More Circular AI Economy
The AI boom is creating an opportunity to rethink how major infrastructure is designed from the beginning.
Data-center developers and technology companies can consider repairability, component reuse, material identification and end-of-life recovery when selecting equipment. Suppliers can reduce unnecessary packaging and improve the recyclability of components. Recycling companies can develop systems specifically suited to the complex materials emerging from modern electronics and digital infrastructure.
Achieving this will require coordination among technology companies, recyclers, utilities, energy suppliers, infrastructure developers and commercial partners. It is not merely a waste-management problem. It is an integrated materials, energy and infrastructure challenge—the kind of intersection increasingly being examined through projects such as CBB Advisory.
The objective should not be a world without plastics or artificial intelligence. Both provide enormous practical benefits, and both will remain important parts of modern society.
The objective should be a world in which useful materials are not casually discarded after a single application, products are designed with recovery in mind and technologies exist to capture value from materials that were once considered unrecyclable.
Artificial intelligence is often described as a tool for optimizing complex systems. We should apply that same intelligence to the physical system supporting AI itself.
The AI economy will require more plastics. The opportunity is to ensure that it also drives better design, better collection, better recycling and a more genuinely circular use of resources.
About the Author
Gregory Merle (Greg Merle) is a materials scientist and engineer, energy infrastructure executive, and project developer with more than 20 years of experience in resource recovery, advanced recycling, industrial infrastructure, and environmental solutions. He serves as a director and advisor to Real-Cycle and is involved in initiatives focused on recycling innovation, waste reduction, environmental education, and practical resource recovery solutions.