Lisa Biber-Freudenberger

arXiv2025

More than Carbon: Cradle-to-Grave environmental impacts of GenAI training on the Nvidia A100 GPU

The rapid expansion of Artificial Intelligence (AI) has intensified concerns about its environmental sustainability. Current assessments focus on operational carbon emissions using secondary data, overlooking impacts in other life cycle stages. This study presents a comprehensive multi-criteria life cycle assessment of AI training, examining 16 environmental impact categories using primary data from the Nvidia A100 SXM 40 GB GPU. Results for GPT-4 training show that the use phase dominates 10 categories, contributing 96% to climate change and fossil fuel depletion. Manufacturing dominates 6 categories, including human toxicity (94%) and freshwater eutrophication (81%). The GPU chip is the largest contributor in 10 categories, particularly climate change (81%) and fossil resource use (80%). While primary data produces modest changes in carbon estimates, substantial variations emerge elsewhere, e.g. minerals and metals depletion increases by 33%. This analysis expands the Sustainable AI discourse beyond carbon emissions, challenging current sustainability narratives.

cs.CY↗

arXiv2025

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.

cs.CY↗