Nature Sustainability study reveals surging power consumption of US hyperscale datacenters
11/17/2026 — 11/17, 15:15·1 sources (all)·1 reports (all)
Key takeaways
- A study in Nature Sustainability shows that US hyperscale datacenter electricity consumption has reached 68-99 TWh, with upper bounds approa…
Background & analysis
On November 17, 2026, Solidot reported on a study published in Nature Sustainability that puts the electricity use of hyperscale data centers in the United States on par with the total annual consumption of several individual states. The research examined 403 hyperscale data centers across the country, facilities that underpin cloud computing, LLM training, and the global rollout of AI tools, covering the period from May 2024 to April 2025.
Researchers analyzed satellite imagery of these facilities to estimate annual power consumption under different operating loads, then calculated carbon emissions based on the energy mix of the regional power grids where each site sits. Total electricity use across the hyperscale fleet ranged from 68-99 TWh depending on load, with the upper bound approaching the full yearly consumption of states such as Indiana, Michigan, and Tennessee. During the study window, the facilities accounted for 1.8% of total U.S. electricity use.
On the power side, roughly 54% of the electricity came from fossil fuels, split between natural gas at 38% and coal at 15%, while one quarter was drawn from grids with above-average carbon intensity. That consumption produced about 25.7 million metric tons of CO2 emissions over the period. The authors also noted that current power demand and emissions are likely far higher than the levels observed during the study. As of the Solidot report, the findings stand on the study's own estimates, with no further developments or responses recorded since publication.
AI-generated from 1 reports · updated 29 minutes ago
- Heat index
- 96
- All sources
- 1
- All reports
- 1
- First seen
- 2 hours ago
Reports on this story headlines open the original
According to Solidot citing a study in Nature Sustainability, power consumption by US hyperscale data centers—which support LLM training and AI tool deployment—reached between 68 and 99 TWh, with the upper limit matching the annual electricity use of entire states like Indiana. The research reveals that about 54% of their power comes from fossil fuels, generating roughly 25.7 million metric tons of CO2 emissions during the study period.
SolidotAI score 58
Other stories people are talking about
- Heat index 1447RisingNvidia board advances probe into Jensen Huang as staff urge his return24 sources
- Heat index 1138RisingCalifornia Sues OpenAI Over Agents Escaping Tests, Intruding into Hugging Face15 sources
- Heat index 950RisingAlibaba releases MoE model Qwen4-Max with 1M token context for low-cost, long-running tasks18 sources
- Heat index 650US Commerce Department issues interim final rule restricting remote access to AI chips via overseas data centers10 sources
- Heat index 585SurgeAnthropic IPO filing revision reveals $900.9B compute commitment7 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion

