ai
Wasted large language models: A life cycle thinking approach
- Source
- arXiv — Computers and Society
- Published
- Last verified
- 20 Aug 2026
- Confidence
- High
- Evidence
- Original document retained
- Reading time
- 1 min
- Country
- International
- Relevant to
- Research & Evidence, Policy & Regulation, Technology & Data
Executive summary
What happened, and why should leadership care?
Large Language Models (LLMs) are identified as having a significant and growing carbon footprint due to their development and use. Despite efforts to improve energy efficiency, overall consumption has not decreased, primarily due to rebound effects like Jevons Paradox. A new approach, termed 'life cycle thinking,' is proposed to manage the environmental impact of LLMs, treating them as products that can generate waste and suggesting the application of the EU's Waste Framework Directive's waste hierarchy.
Why this matters
Why is this strategically important?
The environmental footprint of Large Language Models represents a growing concern for technological sustainability and corporate social responsibility. Unchecked consumption, driven by rebound effects, poses a risk to green technology initiatives and regulatory compliance. Adopting life cycle thinking and waste management principles for digital assets could establish new industry standards and mitigate long-term environmental liabilities.
Key insights
What should be noted from the evidence?
- Large Language Models (LLMs) possess an increasing carbon footprint from their development and operational usage.
- Energy efficiency advancements in LLMs have not led to reduced consumption due to rebound effects such as Jevons Paradox.
- Additional measures beyond efficiency improvements are necessary to address the environmental impact of LLMs.
- Life cycle thinking, conceptualizing LLMs as products that can become 'waste,' is proposed as a potential solution.
- The EU's Waste Framework Directive's waste hierarchy (prevention, reuse, recycling, recovery, disposal) is suggested as a framework for managing LLM-related 'waste'.
Evidence and confidence
How far can this assessment be trusted?
High confidence. Named institution, original document retained and analysis corroborated.
Analysis is prepared editorially by Aziz Shuaib Ausi. The original publication remains the authoritative record, and executive judgement remains entirely human.
Source
Where does this originate?
Reported by arXiv — Computers and Society · International. This briefing summarises the publication for executive use; the document itself is not reproduced here.
Read the original publication