Efficiency Will Not Lead to Sustainable Reasoning AI

TL;DR

This paper argues that efficiency improvements alone cannot ensure the sustainability of reasoning AI, as performance continues to scale with exponential compute investments beyond physical limits.

Contribution

It highlights the need for explicit limits and governance in AI development to address sustainability challenges in reasoning AI.

Findings

Efficiency gains are nearing physical limits.

Performance scales with exponential compute investments.

Sustainable reasoning AI requires new policy approaches.

Abstract

AI research is increasingly moving toward complex problem solving, where models are optimized not only for pattern recognition but for multi-step reasoning. Historically, computing's global energy footprint has been stabilized by sustained efficiency gains and natural saturation thresholds in demand. But as efficiency improvements are approaching physical limits, emerging reasoning AI lacks comparable saturation points: performance is no longer limited by the amount of available training data but continues to scale with exponential compute investments in both training and inference. This paper argues that efficiency alone will not lead to sustainable reasoning AI and discusses research and policy directions to embed explicit limits into the optimization and governance of such systems.