A data centre orders copper before it produces much scrap. That simple timing difference sits inside a growing investment argument: the expansion of AI infrastructure could create more business for companies that recover materials and keep equipment in use.
Tazia Smith, chief executive of Closed Loop Capital Management, outlined that opportunity in Business Insider’s reporting published on 12 November. Her view is an investor’s thesis, rather than evidence that the recycling industry has already earned higher margins. Testing it means following the physical material and the commercial contract through several steps.
The demand backdrop has independent support. The International Energy Agency’s 2026 analysis describes fast-growing data-centre infrastructure and pressure on equipment supply chains. That supports the relevance of additional sources of supply. It does not establish orders, capacity utilisation or profits at any particular recycler.
New construction and old scrap run on different clocks
Copper installed in cables and electrical equipment becomes a stock of material inside working infrastructure. A wave of new construction initially adds to that stock. It cannot simultaneously supply an equivalent wave of retired equipment from those same projects. The future recovery opportunity depends on when equipment is replaced and what can be collected.
That does not leave recyclers waiting for today’s AI facilities to age. They can draw material from other buildings, manufacturing operations and older electronics. The important distinction is between manufacturing scrap, generated during production, and material recovered after a product has been used.
The US Geological Survey’s 2026 copper summary illustrates the difference. It estimates that US copper recovery in 2025 included 160,000 tonnes from old, post-consumer scrap and 760,000 tonnes from new manufacturing scrap. These are national copper flows, not data-centre recycling figures, and the estimates refer to 2025 rather than current-year output.
For a recycling investment, the relevant bridge is therefore between an identifiable supply of material and a buyer’s need. A forecast of more AI construction supplies only one side. Collection agreements, manufacturing relationships and access to existing equipment determine whether a processor has enough input to serve that demand.
Material quality determines which buyer can use the output
A pile of collected equipment is not yet a shipment of metal that an industrial customer can use. Collection, separation, processing and any necessary refining stand between the two. Different input streams require different work, and not every part of the collected weight becomes the saleable output.
This is a mechanism, not a claim about a named company’s recovery performance. An investor would need actual data on recovered yield, contamination, processing cost and the specification accepted by customers. Capacity described in tonnes of incoming material cannot be compared directly with tonnes of finished copper without that conversion.
The IEA’s minerals outlook identifies copper as having an established recycling stream and says greater secondary supply requires investment in collection and recycling infrastructure, alongside sustained demand for recycled material. Existing recycling activity makes the opportunity credible. It also means a new entrant competes with established routes for both feedstock and buyers.
A supply-security premium is possible if reliable local material helps a customer avoid disruption. Whether the processor captures it depends on product quality, delivery performance and contract terms. The strategic value to the customer is not automatically the seller’s profit margin.
A higher copper price can also raise the recycler’s bill
Commodity exposure passes through both sides of a recycling business. A processor that buys scrap and sells recovered metal may receive more for its output when metal prices rise. But suppliers can demand more for the input too. The economically relevant amount is the spread after processing and other costs.
Contract design changes that exposure. In an illustrative service arrangement, a customer might retain ownership of the material and pay a processing fee. In a purchase-and-resale arrangement, the recycler may finance inventory and face price changes while material moves through the plant. Those are different business models, not descriptions of an undisclosed Closed Loop contract.
Working capital matters alongside the margin. Paying for material before receiving the customer’s payment ties up cash. Faster throughput or better payment terms can improve cash conversion without changing headline metal prices. Conversely, growing sales can consume cash if inventory and receivables rise faster than collections.
Keeping equipment in use is a separate source of value
Material recovery is only one route. Closed Loop Foundation’s overview of used electronics explains that equipment no longer useful to its first owner can retain considerable value. Reuse and repair can preserve a functioning product before its materials are recovered. This is background on the business mechanism, not a newly announced AI contract.
The strongest counterargument to a narrow scrap bottleneck is therefore that a capable operator can earn from services and existing equipment across several industries. It may benefit from AI-related demand without relying on retired AI servers as its main input. But the value of usable equipment still depends on condition, compatibility and a willing buyer.
The opportunity becomes financially persuasive when secure feedstock, qualified output and contract economics appear together. Evidence of repeat orders, realised processing margins and cash conversion would strengthen it; weak collections or expensive inventory would weaken it. AI demand can open a market. Turning that market into a profitable material loop remains an operating achievement.