From FLOPs to Footprints: The Resource Cost of Artificial Intelligence

As computational demands continue to rise, assessing the environmental footprint of AI requires moving beyond energy and water consumption to include the material demands of specialized hardware. This study quantifies the material footprint of AI training by linking computational workloads to physical hardware needs. The elemental composition of the Nvidia A100 SXM 40 GB graphics processing unit (GPU) was analyzed using inductively coupled plasma optical emission spectroscopy, which identified 32 elements. The results show that AI hardware consists of about 90% heavy metals and only trace amounts of precious metals. The elements copper, iron, tin, silicon, and nickel dominate the GPU composition by mass. In a multi-step methodology, we integrate these measurements with computational throughput per GPU across varying lifespans, accounting for the computational requirements of training specific AI models at different training efficiency regimes. Scenario-based analyses reveal that, depending on Model FLOPs Utilization (MFU) and hardware lifespan, training GPT-4 requires between 1,174 and 8,800 A100 GPUs, corresponding to the extraction and eventual disposal of up to 7 tons of toxic elements. Combined software and hardware optimization strategies can reduce material demands: increasing MFU from 20% to 60% lowers GPU requirements by 67%, while extending lifespan from 1 to 3 years yields comparable savings; implementing both measures together reduces GPU needs by up to 93%. Our findings highlight that incremental performance gains, such as those observed between GPT-3.5 and GPT-4, come at disproportionately high material costs. The study underscores the necessity of incorporating material resource considerations into discussions of AI scalability, emphasizing that future progress in AI must align with principles of resource efficiency and environmental responsibility.

https://arxiv.org/abs/2512.04142

From Efficiency Gains to Rebound Effects: The Problem of Jevons' Paradox in AI's Polarized Environmental Debate

https://dl.acm.org/doi/full/10.1145/3715275.3732007

As the climate crisis deepens, artificial intelligence (AI) has emerged as a contested force: some champion its potential to advance renewable energy, materials discovery, and large-scale emissions monitoring, while others underscore its growing carbon footprint, water consumption, and material resource demands. Much of this debate has concentrated on direct impacts—energy and water usage in data centers, e-waste from frequent hardware upgrades—without addressing the significant indirect effects. This paper examines how the problem of Jevons’ Paradox applies to AI, whereby efficiency gains may paradoxically spur increased consumption. We argue that understanding these second-order impacts requires an interdisciplinary approach, combining lifecycle assessments with socio-economic analyses. Rebound effects undermine the assumption that improved technical efficiency alone will ensure net reductions in environmental harm. Instead, the trajectory of AI’s impact also hinges on business incentives and market logics, governance and policymaking, and broader social and cultural norms. We contend that a narrow focus on direct emissions misrepresents AI’s true climate footprint, limiting the scope for meaningful interventions. We conclude with recommendations that address rebound effects and challenge the market-driven imperatives fueling uncontrolled AI growth. By broadening the analysis to include both direct and indirect consequences, we aim to inform a more comprehensive, evidence-based dialogue on AI’s role in the climate crisis.

The Hidden Cost of AI

 

Dr. Sasha Luccioni Projects

• Evaluating the carbon emissions of AI models – my longstanding project is getting a better idea of how much carbon is emitted by AI models and what are the factors that influence it - see my “BLOOM” and “Counting Carbon” articles.
• Stable Diffusion Bias Explorer – a demo for exploring the biases in text-to-image models like Stable Diffusion and Dall-E 2.
• The Data Measurements Tool – a tool for exploring and analyzing common datasets used for training and evaluating Machine Learning models.
• This Climate Does Not Exist – in which we use Generative Adversarial Networks (GANs) to visualize the potential future impacts of climate change.
• CodeCarbon – I am contributing to creating a calculator to quantify the CO2 emissions produced during the training of AI algorithms.
• Big Science – BigScience is a one-year long research workshop on very large language models as used and studied in the field of Natural Language Processing and more generally Artifical Intelligence research. I am co-chairing the carbon footprint working group within the project.

https://www.sashaluccioni.com/projects/

https://huggingface.co/sasha

Measuring the environmental impact of delivering AI at Google Scale

The transformative power of AI is undeniable - but as user adoption accelerates, so does the need to understand and mitigate the environmental impact of AI serving. However, no studies have measured AI serving environmental metrics in a production environment. This paper addresses this gap by proposing and executing a comprehensive methodology for measuring the energy usage, carbon emissions, and water consumption of AI inference workloads in a large-scale, AI production environment. Our approach accounts for the full stack of AI serving infrastructure - including active AI accelerator power, host system energy, idle machine capacity, and data center energy overhead. Through detailed instrumentation of Google's AI infrastructure for serving the Gemini AI assistant, we find the median Gemini Apps text prompt consumes 0.24 Wh of energy - a figure substantially lower than many public estimates. We also show that Google's software efficiency efforts and clean energy procurement have driven a 33x reduction in energy consumption and a 44x reduction in carbon footprint for the median Gemini Apps text prompt over one year. We identify that the median Gemini Apps text prompt uses less energy than watching nine seconds of television (0.24 Wh) and consumes the equivalent of five drops of water (0.26 mL). While these impacts are low compared to other daily activities, reducing the environmental impact of AI serving continues to warrant important attention. Towards this objective, we propose that a comprehensive measurement of AI serving environmental metrics is critical for accurately comparing models, and to properly incentivize efficiency gains across the full AI serving stack.

Informe sobre los centros de datos de Aragón- El precio de las nubes

https://tunubesecamirio.com/downloads/informe_centro_de_datos_aragon.pdf

Data Center Map

https://www.datacentermap.com/

Sam Altman Admits That Saying “Please” and “Thank You” to ChatGPT Is Wasting Millions of Dollars in Computing Power

https://futurism.com/altman-please-thanks-chatgpt

https://insighttechtalk.com/tech-news/stop-saying-please-thank-you-ai-cost/ 

https://arxiv.org/abs/2601.22357. https://arxiv.org/pdf/2601.22357 

Being polite is free - or is it? In this paper, we quantify the energy cost of seemingly innocuous messages such as ``thank you'' when interacting with large language models, often used by users to convey politeness. Using real-world conversation traces and fine-grained energy measurements, we quantify how input length, output length and model size affect energy use. While politeness is our motivating example, it also serves as a controlled and reproducible proxy for measuring the energy footprint of a typical LLM interaction. Our findings provide actionable insights for building more sustainable and efficient LLM applications, especially in increasingly widespread real-world contexts like chat. As user adoption grows and billions of prompts are processed daily, understanding and mitigating this cost becomes crucial - not just for efficiency, but for sustainable AI deployment.

AI’s water and electricity use soars in 2025

https://www.theverge.com/news/845831/ai-chips-data-center-power-water

carbon.txt is an open-source tool designed to make sustainability data easier to discover and use on the web.

https://www.thegreenwebfoundation.org/news/carbontxt-v0-4/

How Much Energy Does It Take to Store 1 Terabyte of Data in the Cloud?

 https://www.ecoflow.com/us/blog/energy-cost-cloud-storage?

A common estimate for the total energy consumption of 1 TB of data in a typical cloud storage service is between 40 and 70 kWh per year

National Data Center Moratorium Now!

 https://www.foodandwaterwatch.org/wp-content/uploads/2025/12/National-Data-Center-Moratorium.pdf

Climate change is devastating mining of minerals needed to fight it

https://www.nature.com/articles/d41586-025-03560-0?utm_source=Live+Audience&utm_campaign=a5a43a1778-nature-briefing-anthropocene-20251114&utm_medium=email&utm_term=0_-33f35e09ea-50015160

If the US Has to Build Data Centers, Here’s Where They Should Go

https://www.wired.com/story/heres-where-to-build-data-centers-to-keep-emissions-down/

A Small Town Is Fighting a $1.2 Billion AI Datacenter for America's Nuclear Weapon Scientists

 https://www.404media.co/a-small-town-is-fighting-a-1-2-billion-ai-datacenter-for-americas-nuclear-weapon-scientists/?ref=daily-stories-newsletter

https://cartography-of-generative-ai.net/

 

https://cartography-of-generative-ai.net/

The popularisation of artificial intelligence (AI) has given rise to imaginaries that invite alienation and mystification. At a time when these technologies seem to be consolidating, it is pertinent to map their connections with human activities and more than human territories. What set of extractions, agencies and resources allow us to converse online with a text-generating tool or to obtain images in a matter of seconds?

Sam Altman’s AI empire will devour as much power as New York City and San Diego combined.

 

https://fortune.com/2025/09/24/sam-altman-ai-empire-new-york-city-san-diego-scary/ 

Artificial Intelligence Index Report 2025

 

https://hai.stanford.edu/assets/files/hai_ai_index_report_2025.pdf 

Sugerencias para el Real Decreto sobre Centros de datos en España

 https://tunubesecamirio.com/2025/09/11/sugerencias-para-el-real-decreto-sobre-centros-de-datos-en-espana-en-castellano-e-ingles/ 

Sustainable HPC: Modeling, Characterization, and Implications of Carbon Footprint in Modern HPC Systems

 

https://arxiv.org/pdf/2306.13177v1