Recalibrating global data center energy-use estimates

By E. Masanet, Arman Shehabi, N. Lei, S. Smith, Jonathan Koomey. 2020

1,184 citations recorded by the source corpus.

Growth in energy use has slowed owing to efficiency gains that smart policies can help maintain in the near term Data centers represent the information backbone of an increasingly digitalized world. Demand for their services has been rising rapidly (1), and data-intensive technologies such as artificial intelligence, smart and connected energy systems, distributed manufacturing systems, and autonomous vehicles promise to increase demand further (2). Given that data centers are energy-intensive enterprises, estimated to account for around 1% of worldwide electricity use, these trends have clear implications for global energy demand and must be analyzed rigorously. Several oft-cited yet simplistic analyses claim that the energy used by the world's data centers has doubled over the past decade and that their energy use will triple or even quadruple within the next decade (3–5). Such estimates contribute to a conventional wisdom (5, 6) that as demand for data center services rises rapidly, so too must their global energy use. But such extrapolations based on recent service demand growth indicators overlook strong countervailing energy efficiency trends that have occurred in parallel (see the first figure). Here, we integrate new data from different sources that have emerged recently and suggest more modest growth in global data center energy use (see the second figure). This provides policy-makers and energy analysts a recalibrated understanding of global data center energy use, its drivers, and near-term efficiency potential.

Subfields: Computer Networks and Communications, Electrical and Electronic Engineering, Information Systems, ML Systems & Hardware

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Networks and Communications

In Electrical and Electronic Engineering

In Information Systems

In ML Systems & Hardware

Citations in the corpus

Evaluating Large Language Models Trained on Code (2021, background): “8 While data center energy usage has become much more efﬁ-cient in recent years (Masanet et al., 2020), the production, use, and disposal of semiconductors still imposes environmental and human costs.”
Carbon Emissions and Large Neural Network Training (2021, background): “A recent paper in Science [Mas20] found that global datacenter energy consumption increased by only 6% compared with 2010, despite computing capacity increasing by 550% over the same time period [Mas21].”
Measuring the Carbon Intensity of AI in Cloud Instances (2022, background): “In 2018, it was estimated that global data center energy use represented close to 1% of global energy usage [27].”
The Computational Limits of Deep Learning (2020, methodology): “Server management can also make data centers more efficient, for example via techniques such as server consolidation, dynamic voltage scaling, and intelligent workload distribution [8, 80].”
The real climate and transformative impact of ICT: A critique of estimates, trends, and regulations (2021, background): “Blockchain is underwritten by energy: the algorithm, if based on ‘‘proof of work,’’ creates high levels of replication and redundant computation.70 The methodology and assumptions behind Mora et al.’s70 projections of blockchain’s future energy use have been questioned by Masanet et al.,11 but proof”
TPU v4: An Optically Reconfigurable Supercomputer for Machine Learning with Hardware Support for Embeddings (2023, background): “57 [52] as users closed their older data centers and switched to WSCs in the cloud [34].”
Digitalization and energy consumption. Does ICT reduce energy demand? (2020, neutral): “however is currently controversial (Hintemann and Hinterholzer, 2019; Masanet et al., 2020).”
Aligning artificial intelligence with climate change mitigation (2022, neutral): “Servers in large data centres are typically replaced every 3–4 years, which can result in higher operational efficiency 56,73 ; however, shorter lifespans could also increase the share of life-cycle emissions from manufacturing, which can be mitigated by reusing servers and equipment (such as older…”
A Holistic Approach for Designing Carbon Aware Datacenters (2022, background): “Data centers world-wide are estimated to consume 205 TWh of electricity in 2018 [47], exceeding the annual consumption”
Carbon-Aware Computing for Datacenters (2021, neutral): “D EMAND for computing resources and datacenter power worldwide has been continuously growing, now accounting for approximately 1% of total electricity usage [1].”
Usage impact on data center electricity needs: A system dynamic forecasting model (2021, background): “Masanet et al. [1] conclude that despite a massive growth of data storage (25-fold with only 3-fold increase in energy), IP traffic (10-fold growth with only a marginal increase in energy used), and data center compute instances (6.5-fold with 25% energy usage increase), the total energy…”
ACT: designing sustainable computer systems with an architectural carbon modeling tool (2022, background): “For example, data centers have seen a 5×, 8×, and 3× energy efficiency improvement in compute, storage, and networking, respectively [55].”