|
Work item:
|
L.CU_AI
|
|
Subject/title:
|
Guidelines for Circular Utilization of artificial intelligence Computing Power Resources
|
|
Status:
|
Under study
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
2027 (Medium priority)
|
|
Liaison:
|
ETSI EE
|
|
Supporting members:
|
State Grid Corporation of China, China Information Communication Technologies Group, China Unicom, China Mobile, ZTE, Nokia
|
|
Summary:
|
The deployment scale of AI computing equipment (such as GPU servers and dedicated AI accelerators) is expanding rapidly. As chips iterate and equipment ages, they may no longer meet the high computational demands of AI training tasks, but they can still effectively support applications with lower resource requirements, such as AI inference and edge computing, demonstrating significant continued-use value.
Improper disposal of downgraded AI computing resources not only leads to resource waste but also poses environmental risks and data security threats. Existing practices and standards primarily focus on general ICT equipment and lack specialized guidance for the graded and hierarchical circular management of AI computing resources.
The uniqueness of AI computing equipment lies in the fact that its value and service life are highly dependent on the evolution of AI models. Through AI model optimization techniques such as quantization and pruning, downgraded computing equipment can still operate efficiently in suitable scenarios. When combined with AI chip upgrades or expansion slot capabilities, this creates reuse possibilities for downgraded computing equipment. Therefore, this standard will establish a hierarchical architecture and circular utilization process guidelines for downgraded AI computing resources, promoting their orderly, efficient, and sustainable transition into the next life cycle.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2025-11-12 11:21:43
|
|
Last update:
2025-11-12 11:30:22
|
|
