Climate impacts of AI hardware manufacturing rival those of data centers
AIハードウェア製造の気候影響はデータセンターと同等 (AI 翻訳)
Suh S, Xie X, Xu M
🤖 gxceed AI 要約
日本語
本論文は、グローバルなAIハードウェア製造とデータセンター運営の温室効果ガス排出量を、企業・施設レベルで推計。2024年の市場ベース会計において、AIハードウェア製造の排出量（5-7 Mt CO2e/年）がデータセンター運営（5-6 Mt CO2e/年）を上回ることを示した。AIの気候影響を理解するには、データセンターの電力使用だけでなく、ハードウェア製造と企業の脱炭素化コミットメントを考慮する必要がある。
English
This paper compiles a facility- and corporate-level GHG emissions inventory for global AI hardware manufacturing and AI data center operations. Under market-based accounting, AI hardware manufacturing emitted 5–7 Mt CO2e/yr in 2024, exceeding the 5–6 Mt CO2e/yr from data center operations. The findings highlight that understanding AI's climate footprint requires looking beyond data center electricity use to hardware manufacturing and corporate decarbonization commitments.
Unofficial AI-generated summary based on the public title and abstract. Not an official translation.
📝 gxceed 編集解説 — Why this matters
日本のGX文脈において
日本では、半導体製造やAI関連産業の拡大に伴い、サプライチェーン全体の排出量管理が重要となる。本論文の知見は、SSBJやScope3報告の枠組みにおいて、AIハードウェア製造段階の排出を考慮する必要性を示唆する。
In the global GX context
Globally, this paper underscores that AI's climate impact extends beyond data center electricity to include hardware manufacturing, urging companies to account for Scope 3 emissions under ISSB/CSRD. It also emphasizes the role of renewable electricity procurement in reducing reported emissions under market-based accounting.
👥 読者別の含意
🔬研究者:This paper provides crucial empirical data on the full lifecycle emissions of AI hardware, informing lifecycle assessment and climate modeling.
🏢実務担当者:Companies in the AI supply chain should prioritize measuring and reducing manufacturing emissions, and consider renewable energy procurement to lower market-based emissions.
🏛政策担当者:Regulators should include AI hardware manufacturing in climate disclosure requirements, as its emissions are comparable to operational emissions.
📄 Abstract（原文）
<title>Abstract</title> <p>Data centers have been the focus of AI’s electricity consumption and greenhouse gas emissions1, while the manufacturing of AI hardware has received much less attention. Yet the advanced AI hardware that improves use-phase energy efficiency in AI data centers requires increasingly complex and energy-intensive processes and components2. Here, we compile a facility- and corporate-level annual greenhouse gas emissions inventory for global AI hardware manufacturing and AI data center operations. We estimate that, under market-based accounting, which accounts for renewable electricity procurement, AI hardware manufacturing emitted 5–7 Mt CO2e yr⁻¹ in 2024, exceeding the 5–6 Mt CO2e yr⁻¹ from AI data center operations. Under location-based accounting, however, operational emissions were substantially larger, at 20–21 Mt CO2e yr⁻¹ versus 6–8 Mt CO2e yr⁻¹ for AI hardware manufacturing. These findings show that understanding AI’s climate footprint requires looking beyond data center electricity use to hardware manufacturing and corporate decarbonization commitments.</p>
🔗 Provenance — このレコードを発見したソース
- Research Square https://doi.org/10.21203/rs.3.rs-9963675/v1first seen 2026-06-26 04:36:12
🔔 こうした論文の新着を逃したくない方は キーワードアラート に登録（無料・3キーワードまで）。
gxceed は公開メタデータに基づく研究支援データセットです。要約・翻訳・解説は AI 支援で生成されています。 最終的な解釈・検証は利用者が原典資料に基づいて行うことを前提とします。