This research introduces a novel approach to reduce the environmental impact of growing LLM infrastructure by managing CPU aging for extended hardware lifespans.

• Embodied carbon challenge: Manufacturing CPUs for LLM inference servers represents a significant environmental cost that needs better amortization
• Aging-aware management: Technique balances workloads across CPU cores to prevent premature failures while maintaining performance
• Extended lifespan: Strategic core management helps amortize embodied carbon emissions over longer server lifetimes
• Sustainable scaling: Enables more environmentally responsible growth of cloud LLM infrastructure

This engineering innovation matters because it addresses both reliability and sustainability challenges as organizations scale their LLM deployments in the cloud.

Aging-aware CPU Core Management for Embodied Carbon Amortization in Cloud LLM Inference