Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-0433
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-07-20T10:47:07.256666Z
status: published
lens: p_space
sector: climate
headline: The carbon and water footprints of data centers and what this could mean for artificial intelligence
dek: Although there are ways to estimate the global power demand of artificial intelligence (AI) systems, it remains challenging to quantify the associated carbon and water footprints. The lack of distinction between AI and non-AI workloads in the environmental reports of data center operators makes it possible to assess the environmental impact of AI workloads only by approximating them through data centers' general performance metrics. The environmental disclosure of tech companies is, however, often insufficient t…
gain_title: (none)
problem_title: AI workloads' environmental impact is growing rapidly but remains hard to quantify because data center operators do not separate AI and non-AI reporting, with AI alone projected to reach 32.6-79.7 million tons CO2 and 312.5-764.6 billion liters of water in 2025.
trace_subject: (none)
gain_reading: (none)
gain_evidence: (none)
problem_reading: AI workloads' environmental impact is growing rapidly but remains hard to quantify because data center operators do not separate AI and non-AI reporting, with AI alone projected to reach 32.6-79.7 million tons CO2 and 312.5-764.6 billion liters of water in 2025.
problem_evidence: environmental impact of data centers is growing rapidly
quick_read: A December 2025 peer-reviewed article in Patterns examines how to estimate the carbon and water footprints of data centers and AI. It finds that lack of workload-specific reporting forces researchers to approximate AI impacts from general data center metrics, and that company disclosures often do not allow even total performance to be determined.

The work matters because it quantifies a rapidly growing environmental burden tied to AI expansion and highlights a transparency gap that limits accountability. What remains uncertain is the precise AI share of data center footprints without mandated disclosure of additional metrics.
limitation: Quantification remains approximate because operators do not distinguish AI from non-AI workloads and disclosures are insufficient to determine total data center performance.
tag: Evidence-backed problem
key_points: Peer-reviewed analysis in Patterns estimates AI systems alone could account for 32.6 to 79.7 million tons of CO2 and 312.5 to 764.6 billion liters of water in 2025. | Authors say environmental reports do not separate AI and non-AI workloads, forcing approximation through general data center performance metrics. | Tech companies' disclosures are often insufficient to determine even total data center performance, prompting call for policies mandating additional metrics.
rundown: The paper notes that while global power demand of AI can be estimated, carbon and water footprints are harder to pin down because operators' environmental reports combine AI and non-AI workloads.

It argues that current tech company disclosures are often insufficient to determine total data center performance, and suggests new policies mandating additional metrics to remedy transparency shortcomings as impacts grow.
sources:
- peer_reviewed | Patterns | https://doi.org/10.1016/j.patter.2025.101430 | 2025-12-17
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

9ac18e080b4c1913861aff49c8d35fd999f75051b6c08d398f46ce939792cd47

previous

0000000000000000000000000000000000000000000000000000000000000000