# Global database of cement production assets and upstream suppliers

## TL;DR

This paper introduces a global database for cement production assets and suppliers to better understand climate and sustainability impacts.

## Contribution

The paper presents a novel method combining geospatial computer vision and large language models to create a comprehensive cement production dataset.

## Key findings

- The database includes plant age and raw material sourcing patterns as key variables.
- The method improves modeling of cement production utilization rates and greenhouse emissions.
- Geospatial and language technologies enable a holistic view of global cement production.

## Abstract

Cement producers and their investors are navigating evolving risks and opportunities as the sector’s climate and sustainability implications become more prominent. While many companies now disclose greenhouse gas emissions, the majority from carbon-intensive industries appear to delegate emissions to less efficient suppliers. Recognizing this, we underscore the necessity for a globally consolidated asset-level dataset, which acknowledges production inputs provenance. Our approach not only consolidates data from established sources like development banks and governments but innovatively integrates the age of plants and the sourcing patterns of raw materials as two foundational variables of the asset-level data. These variables are instrumental in modeling cement production utilization rates, which in turn, critically influence a company’s greenhouse emissions. Our method successfully combines geospatial computer vision and Large Language Modelling techniques to ensure a comprehensive and holistic understanding of global cement production dynamics.