The Rising Unsustainability of AI Graphics Cards Production

• URL: http://arxiv.org/abs/2607.01258v1
• Date: Fri, 05 Jun 2026 15:49:03 GMT
• Title: The Rising Unsustainability of AI Graphics Cards Production
• Abstract summary: This study estimates the environmental damages associated with graphics cards production over the past decade (2013-2025)<n>We analyze trends in energy consumption, carbon emissions and resource depletion.<n>Our finding highlights the need for greater transparency in life-cycle data, a persistent challenge in AI environmental assessments.
• Score: 3.255583064724235
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The rapid advancement of Artificial Intelligence (AI) has been accompanied by significant increases in computational and environmental costs, driven by large-scale investments in AI infrastructure, hardware, and software. In particular, graphics cards have become central to AI training, with frequent hardware updates required to meet escalating computational demands. However, the environmental damages of graphics cards production remain understudied. This study addresses this gap by estimating the environmental damages associated with graphics cards production over the past decade (2013-2025). We analyze trends in energy consumption, carbon emissions and resource depletion. We compile and provide a dataset documenting the environmental damages of NVIDIA workstation graphics cards production since 2013. Our analysis of this dataset reveals a steady increase in production-related impacts over the period. Our finding highlights the need for greater transparency in life-cycle data, a persistent challenge in AI environmental assessments. While operational efficiency improvements (e.g., energy-efficient training, carbon-aware computing) are often prioritized, our results underscore that production-related impacts are also escalating and cannot be overlooked. The AI community must move beyond incremental optimizations and confront the necessity of sufficiency. This shift may demand structural changes such as policy interventions, hardware design for longevity, and cultural shifts away from perpetual growth and increased performance.

Related papers

• A Survey on the Green Development of Large Models: From Resource-Efficient Architectures to Hardware-Software Co-Design [71.32556809242666]
  The rapid expansion of large-scale AI models has led to significant performance breakthroughs across diverse domains.<n>It has also raised critical concerns regarding computational costs, energy consumption, and environmental sustainability.<n>This survey aims to offer a holistic roadmap toward sustainable, scalable, and socially responsible development of large models.
  arXiv  Detail & Related papers  (2026-07-10T04:02:22Z)
• Improving AI Efficiency in Data Centres by Power Dynamic Response [74.12165648170894]
  The steady growth of artificial intelligence (AI) has accelerated in the recent years, facilitated by the development of sophisticated models.<n> Ensuring robust and reliable power infrastructures is fundamental to take advantage of the full potential of AI.<n>However, AI data centres are extremely hungry for power, putting the problem of their power management in the spotlight.
  arXiv  Detail & Related papers  (2025-10-13T08:08:21Z)
• The Environmental Impacts of Machine Learning Training Keep Rising Evidencing Rebound Effect [3.255583064724235]
  We estimate the environmental impacts associated with training notable AI systems over the last decade.<n>Our analysis reveals two critical trends: First, the impacts of graphics cards production have increased steadily over this period.<n>We show that the impacts of hardware must be considered over the entire life cycle rather than the sole use phase in order to avoid impact shifting.
  arXiv  Detail & Related papers  (2025-10-10T05:49:37Z)
• Life-Cycle Emissions of AI Hardware: A Cradle-To-Grave Approach and Generational Trends [4.68541999760349]
  This study presents the first publication of a comprehensive AI accelerator life-cycle assessment (LCA) of greenhouse gas emissions.<n>Our analysis of five Processing Units (TPUs) encompasses all stages of the hardware lifespan.<n>A byproduct of this study is the new metric compute carbon intensity (CCI) that is helpful in evaluating AI hardware sustainability.
  arXiv  Detail & Related papers  (2025-02-01T17:26:19Z)
• How Green Can AI Be? A Study of Trends in Machine Learning Environmental Impacts [2.640490842167383]
  optimisation strategies aim to reduce the energy consumption and environmental impacts associated with AI.<n>This paper investigates the evolution of individual graphics cards production impacts and of the environmental impacts associated with training Machine Learning (ML) models over time.
  arXiv  Detail & Related papers  (2024-12-23T08:24:44Z)
• Sustainable Diffusion-based Incentive Mechanism for Generative AI-driven Digital Twins in Industrial Cyber-Physical Systems [65.22300383287904]
  Industrial Cyber-Physical Systems (ICPSs) are an integral component of modern manufacturing and industries.<n>By digitizing data throughout product life cycles, Digital Twins (DTs) in ICPSs enable a shift from current industrial infrastructures to intelligent and adaptive infrastructures.<n>GenAI can drive the construction and update of DTs to improve predictive accuracy and prepare for diverse smart manufacturing.
  arXiv  Detail & Related papers  (2024-08-02T10:47:10Z)
• Beyond Efficiency: Scaling AI Sustainably [4.711003829305544]
  Modern AI applications have driven ever-increasing demands in computing. This paper characterizes the carbon impact of AI, including both operational carbon emissions from training and inference as well as embodied carbon emissions from hardware manufacturing.
  arXiv  Detail & Related papers  (2024-06-08T00:07:16Z)
• On the Opportunities of Green Computing: A Survey [80.21955522431168]
  Artificial Intelligence (AI) has achieved significant advancements in technology and research with the development over several decades. The needs for high computing power brings higher carbon emission and undermines research fairness. To tackle the challenges of computing resources and environmental impact of AI, Green Computing has become a hot research topic.
  arXiv  Detail & Related papers  (2023-11-01T11:16:41Z)
• Future Computer Systems and Networking Research in the Netherlands: A Manifesto [137.47124933818066]
  We draw attention to CompSys as a vital part of ICT. Each of the Top Sectors of the Dutch Economy, each route in the National Research Agenda, and each of the UN Sustainable Development Goals pose challenges that cannot be addressed without CompSys advances.
  arXiv  Detail & Related papers  (2022-05-26T11:02:29Z)
• Sustainable AI: Environmental Implications, Challenges and Opportunities [13.089123643565724]
  We characterize the carbon footprint of AI computing by examining the model development cycle across industry-scale machine learning use cases. We present an end-to-end analysis for what and how hardware-software design and at-scale optimization can help reduce the overall carbon footprint of AI.
  arXiv  Detail & Related papers  (2021-10-30T23:36:10Z)
• Compute and Energy Consumption Trends in Deep Learning Inference [67.32875669386488]
  We study relevant models in the areas of computer vision and natural language processing. For a sustained increase in performance we see a much softer growth in energy consumption than previously anticipated.
  arXiv  Detail & Related papers  (2021-09-12T09:40:18Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.