Editor’s note: Electronics recycling will be featured in sessions at the 2026 E-Scrap: The Longevity Conference in New Orleans October 26-28.
In a lengthy conversation with me, veteran technology analyst Bob O’Donnell suggested that artificial intelligence may prove to be as consequential for retired hardware as it is for new hardware. The same forces driving unprecedented investment in AI infrastructure are beginning to influence the economics of servers, PCs, memory and storage, with implications for refurbishment, asset recovery and secondary markets that the ITAD industry cannot afford to ignore.
AI infrastructure spending sits at the center of O’Donnell’s argument. Hyperscale cloud providers, technology vendors and enterprises are collectively investing hundreds of billions of dollars to expand computing capacity and support increasingly demanding AI workloads. While much of the attention has focused on GPUs and advanced processors, O’Donnell believes the effects extend beyond those components.
High-performance memory is one of the clearest examples. Demand for advanced DRAM continues to rise as AI systems require larger memory footprints and more sophisticated architectures. Demand for leading-edge processors and accelerators also remains strong across multiple segments of the market.
The pressure is influencing the value of hardware components beyond AI accelerators themselves. Memory, storage and other supporting technologies are benefiting from demand throughout the hardware supply chain, creating an environment in which certain components may retain value longer than many industry participants have historically expected.
For operators in the ITAD and component trading sectors, that trend reinforces the importance of testing, grading and remarketing hardware at a component level. Equipment once viewed primarily through a recycling lens may increasingly offer opportunities for reuse, refurbishment and component harvesting.
Much of the early AI market centered on cloud-based services, with organizations sending workloads directly to hyperscale providers. Increasingly, enterprises are confronting the economic realities of large-scale AI adoption. As usage grows, so do the costs associated with cloud-based models.
Many organizations are beginning to move toward what O’Donnell describes as a hybrid AI architecture. Under that approach, workloads are distributed across cloud platforms, enterprise data centers and end-user devices instead of relying exclusively on hyperscale infrastructure.
A server that no longer meets the requirements of a hyperscale provider may remain well suited for enterprise inference workloads, internal AI applications or mixed-use computing environments. As organizations build their own AI capabilities, hardware that might previously have exited the market could find a viable second life elsewhere.
A broader market for used enterprise equipment could emerge, and certain categories of hardware could remain economically useful for longer periods. During the discussion, O’Donnell pointed to growing interest in enterprise AI infrastructure as evidence that organizations are increasingly willing to deploy hardware that may not represent the latest generation of technology but remains capable of supporting internal workloads.
The PC market presents a similar dynamic. While overall shipment growth remains modest, manufacturers are placing greater emphasis on systems with larger memory configurations, greater storage capacity and AI-related capabilities. Rising component costs are making it increasingly difficult to deliver the low-cost PCs that once occupied the entry level of the market.
According to O’Donnell, that trend could create new opportunities for refurbished equipment. If new systems continue moving upmarket, demand for affordable computing is unlikely to disappear. A growing share of that demand may be met by professionally refurbished devices.
Refurbished PCs could become an increasingly important part of the entry-level computing market. That possibility places greater emphasis on refurbishment quality, component upgrades and product consistency. Memory and storage upgrades, in particular, may become increasingly important tools for extending useful life and improving resale value.
The discussion also touched on the growing role of leasing and device-as-a-service programs offered by major technology manufacturers. Dell, HP and Lenovo continue expanding programs designed to maintain closer control over hardware throughout its lifecycle while providing customers with greater flexibility in how technology is acquired and refreshed.
For independent ITAD providers, those programs present both opportunities and challenges. They may accelerate refresh cycles and increase the volume of newer equipment entering secondary channels. They may also concentrate greater control over asset flows within OEM ecosystems.
Historically, enterprise technology followed relatively predictable refresh patterns. Artificial intelligence is introducing new variables into that equation. The pace of innovation, shifting workload requirements and the growing role of enterprise AI infrastructure could influence how quickly certain categories of equipment enter secondary markets and how they are valued once they arrive.
The significance of AI, in the broader context of electronics and IT equipment recycling, refurbishing and ITAD, may extend beyond the technologies being deployed today. The infrastructure supporting artificial intelligence is beginning to influence decisions about acquisition, deployment, reuse and retirement throughout the hardware lifecycle.
Whether those trends persist remains to be seen. If O’Donnell’s assessment proves correct, AI may influence how technology is purchased and deployed, and how it is valued once it enters secondary markets.
