Summary

AI-generated

This summary is written by a language model reading the source document. It is not the publisher's words and is not a substitute for the original.

Learn more about AI enrichment

The authors argue that global data center energy use remained relatively stable between 2010 and 2018, rising by only 6% despite a massive increase in internet traffic and storage. This near-plateau is attributed to improvements in IT device efficiency, server virtualization, and the shift toward hyperscale data centers. The paper critiques extrapolation-based models for overestimating energy growth and uses these findings to debunk claims that data center CO2 emissions rival those of the aviation industry. However, it warns that computationally intensive applications like AI could outpace future efficiency gains, necessitating investments in next-generation technology and renewable energy.

Key insights

AI-generated

These insights are written by a language model reading the source document. They are not the publisher's words and are not a substitute for the original.

Learn more about AI enrichment
  • Global data center energy use likely increased by only 6 percent between 2010 and 2018, consuming approximately 205 terawatt-hours (TWh) in 2018, which represents about 1 percent of global electricity use.
  • The relative stability of energy use despite growth in IP traffic and storage is driven by three efficiency factors: technological progress in the energy efficiency of servers and storage drives, the adoption of server virtualization software, and the migration of compute instances to hyperscale-class data centers with ultra-efficient cooling.
  • Extrapolation-based models tend to overestimate data center energy growth because they rely on rapidly growing market indicators, such as IP traffic, while failing to account for countervailing technological and structural efficiency trends.
  • Claims that data center CO2 emissions are equivalent to the global aviation industry are implausible; for such a claim to be true, data centers would need an emissions intensity of 4.4 kg CO2/kWh, which is more than four times the intensity of an average coal-fired power plant.
  • There is a risk that the demand from compute-intensive applications, specifically artificial intelligence (AI), may outpace historical efficiency gains in the coming decade, requiring new investments in heat removal, computing, and renewable power.

Cite the original document

APA
Masanet, E., & Lei, N. (2020). How much energy do data centers really use? Aspen Global Change Institute. https://www.agci.org/research-reviews/how-much-energy-do-data-centers-really-use
Chicago
Masanet, Eric, and Nuoa Lei. How much energy do data centers really use? Aspen Global Change Institute, 2020. https://www.agci.org/research-reviews/how-much-energy-do-data-centers-really-use.
Wikipedia
{{cite report |last1=Masanet |first1=Eric |last2=Lei |first2=Nuoa |title=How much energy do data centers really use? |publisher=Aspen Global Change Institute |date=22 March 2020 |url=https://www.agci.org/research-reviews/how-much-energy-do-data-centers-really-use |access-date=5 November 2026 |via=Climate Insights Directory (document agci-000004)}}
BibTeX
@techreport{masanet2020how, author = {Masanet, Eric and Lei, Nuoa}, title = {{How much energy do data centers really use?}}, institution = {Aspen Global Change Institute}, year = {2020}, month = mar, url = {https://www.agci.org/research-reviews/how-much-energy-do-data-centers-really-use}, urldate = {2026-11-05}, note = {Indexed by Climate Insights Directory, document agci-000004} }

Full text

Collected · Record updated