Beyond Energy: When Sustainability Dimensions Reshape LLM Serving Decisions

2026-09-28

AI Summary

This study addresses the inconsistent optimization decisions arising from isolated assessments of sustainability dimensions—energy, carbon, water, and biodiversity—in large language model (LLM) serving. To overcome the limitations of single-energy-perspective optimization, we propose PRISM, a unified framework integrating multi-objective optimization, life cycle assessment, and regional routing techniques. For the first time, PRISM quantifies cross-dimensional regret values and identifies life cycle boundary conditions, revealing the decisive influence of spatiotemporal deployment on non-energy dimensions. Experimental results demonstrate that, compared to the strongest baseline, PRISM reduces worst-case regret by 50.2%, offering a systematic solution for the sustainable deployment of LLM serving.

Abstract

Large language model (LLM) serving has environmental impacts across energy consumption, carbon emission, water consumption, and biodiversity loss. Yet these dimensions are largely evaluated in isolation, leaving it unclear when and how they lead to different optimization decisions. We present PRISM, a unified framework for characterizing and optimizing LLM serving across energy, carbon, water, and biodiversity impacts. Our analysis reveals a fundamental distinction: computing configurations determine energy consumption, whereas where and when LLM serving is deployed determine its carbon, water, and biodiversity impacts. Under a fixed deployment choice and operational-only accounting, all dimensions preserve the same energy-based configuration ranking. Deployment rankings can diverge across dimensions, while embodied impacts can break configuration invariance when they exceed a lifecycle crossover boundary. PRISM identifies these conditions, quantifies cross-dimensional regrets, and balances the four dimensions. In regional-routing experiments, PRISM reduces median worst-case regret by 50.2% relative to the strongest baseline.

Problem

• Large Language Model Serving
• Sustainability Dimensions
• Environmental Impact
• Multi-dimensional Optimization

Innovation

• Large Language Model Serving
• Sustainability Optimization
• Multi-dimensional Impact
• PRISM Framework
• Cross-dimensional Regret