Life cycle assessment of Mistral Large 2
Share of compute (training and use) in each indicator
Life cycle assessment of Mistral Large 2
The analysis
- For climate and water, most of it comes from compute, hence from server electricity.
- For materials, most of it comes from the manufacture of servers and chips: 61%.
- This is the first life cycle assessment of a large model published with ADEME, using the AFNOR method.
The figures
- Climate: compute, 85.5 %.
- Water: compute, 91 %.
- Materials: compute, 29 %.
Assumptions
- Full life cycle of the Mistral Large 2 model, excluding users' devices for the response.
- A single model, a single provider; the study does not separate training from responses.
- Shares read from the infographic published by Mistral AI; the full report is not public.
- The other stages combine server hardware (11% of climate, 61% of materials), users' devices (3% and 7%) and the data centre buildings (less than 1% and 1.5%).
Continue with hardware
The whole Hardware topic →Charts and maps
Chart
HardwareManufacture of an AI server, by component42%of manufacturing: memory, the largest itemChartHardwareTraining emissions, 4 published assessments20timesthe emissions of GPT-3, for Llama 3.1ChartHardwareThe footprint of digital devices99%of a smartphone's footprint comes from its manufactureChartHardwareElectronic waste and AI22.3%of 2022 electronic waste collected and recycled
HardwareManufacture of an AI server, by component42%of manufacturing: memory, the largest itemChartHardwareTraining emissions, 4 published assessments20timesthe emissions of GPT-3, for Llama 3.1ChartHardwareThe footprint of digital devices99%of a smartphone's footprint comes from its manufactureChartHardwareElectronic waste and AI22.3%of 2022 electronic waste collected and recycled
Water
UsesAcross the hub
AI Impact DatabaseAll the figures, their methodology and their sources.
AI Ranking10 assistants rated out of 100: performance, ethics and environment.
AnalysesThe signed pieces: analyses, viewpoints, "Where does this figure come from?".
Charts and mapsYou are hereThe reference visuals, with their data and sources.
ToolsOpen calculations, your assumptions on the left, the result on the right.
A question about a figure?
An error, a source to add, a methodology question, a press request: write to us. Guillaume Pakula, Clément Reynaud, Sébastien Pierfederici and Alexandre Veilly will reply.
Write to us →
Projet Celsius, which runs this hub, also helps companies measure and reduce the footprint of their AI use.
Accounting for AI in your Bilan Carbone® ↗
