Greenpeace: emissions from AI chip production grew 357% in 2024, driven by a heavy reliance on fossil fuels, outpacing a 351% rise in electricity consumption
Aaron Clark / Bloomberg :
Context & Ripple Effects
The report shifts attention from the electricity used to run AI systems to the emissions embedded in the chips that supply them. That extends an earlier concern that chip fabrication accounts for much of a device's carbon footprint, particularly where manufacturing depends on fossil-fuel-heavy grids.
It also adds a supply-chain dimension to the emissions pressure already visible at AI buyers: Google had reported a five-year rise in greenhouse-gas emissions tied to data-center expansion.
First-order effects
- The reported rise makes AI-chip procurement an immediate emissions exposure for Amazon, Microsoft, Google, HP and other buyers whose key suppliers rely heavily on fossil fuels.
- Because emissions grew faster than electricity consumption, the report puts the carbon intensity of manufacturing power—not just chip volumes—under closer scrutiny.
Second-order effects
- Cloud providers and device brands face stronger pressure to seek cleaner-power commitments and emissions data from chip suppliers, alongside their own data-center energy plans.
- Chipmakers already working to curb fabrication emissions may face a sharper competitive distinction between capacity backed by cleaner grids and capacity reliant on fossil fuels.
Third-order effects
- If AI demand continues to expand through fossil-powered manufacturing hubs, the sector's climate constraint will increasingly span the full compute supply chain rather than data-center operations alone.
- This points toward AI infrastructure being evaluated as utility-scale industrial buildout, where access to clean power can shape both expansion risk and emissions performance.
The trend: AI's environmental footprint is moving upstream from data-center electricity use to the energy mix and industrial emissions of the semiconductor supply chain.