Research

Environmental Data: Entity Profiles

Energy, carbon, water, and disclosure quality data for AI labs, cloud providers, semiconductor manufacturers, and data center operators. Each metric carries a confidence tier reflecting data quality.

Infrastructure Observatory

Interactive map of AI-related data center and power sites, each with environmental context (water stress, grid carbon intensity, drought risk) and documented community concern records.

Data Methodology

How we approach environmental intelligence: confidence tier definitions (A–E), disclosure quality framework (T1–T6), data sources, and estimation methods.

Infrastructure Methodology

Inclusion criteria, site classification definitions, environmental data sources (WRI Aqueduct 4.0, EPA eGrid, IEA), and community concern standards for the infrastructure observatory.

Data center energy estimation (working paper)

Documenting the capacity model, query-volume model, and capex-proxy estimation cascade used when direct disclosure is unavailable. Includes source citations for all constants.

Draft in preparation

AI training carbon accounting

Methodology for attributing Scope 2 emissions to AI training workloads when company reports aggregate all operations. Addresses the renewable energy matching problem.

Draft in preparation

Semiconductor supply chain impact

Environmental burden analysis of GPU and memory manufacturing: fab energy, water, and chemical inputs from TSMC, Samsung, and SK Hynix production nodes.

Draft in preparation