Editor’s note: Electronics recycling will be featured in sessions at the 2026 E-Scrap: The Longevity Conference in New Orleans October 26-28.
While PCs and endpoint devices have been central to ITAD for some three decades, data-center decommissioning has since become a more consequential service line as AI infrastructure is changing the equipment entering retirement projects. Colocation adds further complexity.
This creates an attractive opportunity for ITAD providers, but I would be cautious about treating all data-center work as one market. The construction pipeline may eventually produce larger and more durable disposition volumes. Commercial success will still depend on access to contracts and equipment cycles. Ownership rights matter, as do recovery economics. Providers must also be able to work inside highly controlled facilities.
The work retains familiar ITAD responsibilities, including inventory and data security. Reuse remains central. Yet these projects require an understanding of facility systems, while responsibility may be divided among several parties.
AI infrastructure involves high-density compute. Liquid cooling introduces different technical demands, while batteries create safety concerns. Some equipment connects to infrastructure outside the ITAD provider’s normal scope.
Retirement volume is less certain, but growth is clear
To appreciate the expected growth of data centers, and the scale of the decommissioning work that may follow, consider this: The International Energy Agency reported that global data-center electricity demand rose 17% in 2025, compared with 3% growth in total global electricity demand. It projects that data-center electricity use will double by 2030. Consumption at AI-focused sites could triple.
Those projections do not directly measure future ITAD volume. Retirement activity will depend on refresh cycles, equipment efficiency and facility life.
Uptime Institute reported by Data Center Frontier showed modal rack density exceeding 11 kW in 2026, up from 9 kW in 2025. About 24% of respondents operated racks at 30 kW or more. Yet most had no racks at that level.
Data Center Knowledge reported that 22% of participants in Uptime Institute’s 2025 Cooling Systems Survey used direct liquid cooling.
Liquid cooling illustrates the difference
I have long argued to ITAD CEOs that endpoint disposition and data-center decommissioning require distinct operating strategies, resources and go-to-market approaches. Liquid cooling provides a useful example of why.
A removal plan may need to address cold plates and hoses, together with quick-disconnect fittings. Before work begins, the parties must establish ownership. They must determine whether the loop can be isolated and who may drain it. An ITAD provider may control the assets without having authority over surrounding infrastructure.
Ownership becomes especially complicated in colocation facilities. A tenant may own the servers. The operator may own the cooling plant, while part of the compute fleet may be leased. An ownership matrix can prevent costly mistakes.
Recovery requires component-level decisions
AI equipment also changes recovery economics. A retired server’s greatest value may sit in an accelerator rather than the complete unit. Resale depends on functionality and title. Firmware condition may restrict the buyer pool. Export controls can eliminate attractive channels.
This is another reason I would hesitate to describe data-center decommissioning as automatically more profitable than conventional ITAD. The value can be considerable, but realizing it requires technical knowledge, test capability, market access and disciplined control over the assets. Complexity creates opportunity only when the provider can manage it at a logical cost.
AI data still resides on media
Data handling extends beyond hard drives. Boot media and controller caches may contain sensitive information. AI environments may also hold model weights, prompts or credentials. These remain forms of data stored on media.
NIST SP 800-88 Revision 2 provides media-sanitization guidance. AI-related data does not automatically require physical destruction, but the owner’s decision must be documented before equipment leaves controlled custody.
Battery and power work must be defined
Battery work must likewise be defined in advance. The EPA warns that lithium-ion batteries can cause fires during transport and at recycling facilities. Shutdown may require qualified specialists. Knowing where the ITAD scope ends is part of competent project management.
Specialization may be the stronger strategy
I do not believe every ITAD provider needs to become capable of decommissioning every type of data center. An organization experienced in conventional enterprise facilities may not be ready for a high-density liquid-cooled AI environment. A provider equipped for hyperscale server harvesting may not have the field network or commercial model required for distributed edge sites.
For many ITAD firms, the stronger strategy is to identify segments that fit their technical capabilities and geographic coverage. They should become exceptionally good within that niche. The data-center opportunity is real. So are the boundaries of competence.
