Research insights

Data Centres and Data Transmission Networks

Table of Contents

Global internet traffic surged by over 40% in 2020, primarily due to the explosion of video streaming, video conferencing, online gaming, and social networking. This surge builds on a decade-long trend of skyrocketing digital service demand. Since 2010, the global number of internet users has doubled, and internet traffic has grown an astonishing 15 times, averaging approximately 30% annual growth.

The Energy Demands of Data Centers

The backbone of the internet, data centers, process the lion’s share of global Internet Protocol (IP) traffic. As connectivity expands, so does the demand for data center services and, by extension, energy consumption – primarily electricity. An estimated five bits of data are transmitted within or between data centers for every bit of data sent across networks to users.

In 2020, data centers worldwide consumed an estimated 200-250 terawatt-hours (TWh) of electricity, making up about 1% of global electricity use. However, this figure excludes the electricity-intensive activity of cryptocurrency mining, which alone accounted for approximately 100 TWh in the same year.

Efficiency Gains Curb Data Center Energy Growth

While the demand for data center services continues to rise relentlessly, technological advancements in efficiency have played a vital role in limiting energy consumption. Innovations in server performance, storage systems, network switches, infrastructure, and the growing dominance of highly efficient cloud and hyperscale data centers have mitigated much potential energy demand growth.

Optimized through virtualization technologies, cloud and hyperscale data centers operate at high capacity, delivering more work with fewer servers. Many data centers boast exceptionally low power usage effectiveness (PUE) ratios. The most advanced hyperscale centers achieve PUE values as low as 1.1, meaning that only 0.1 kilowatt-hour is consumed for cooling and power provision for every 1 kilowatt-hour IT equipment uses.

Advancements in Data Transmission and Energy Demand

In 2020, global data transmission networks consumed an estimated 260-340 TWh of electricity, accounting for approximately 1.1-1.4% of global electricity use. Yet, rapid advancements in efficiency have decoupled this energy consumption from the explosive growth in data traffic.

Fixed-line network energy intensity in developed countries has halved every two years since 2000, while mobile-access network efficiency has improved by 10-30% annually. Such improvements are crucial in offsetting the escalating demands of data transmission.

Decoupling Traffic from Energy Use

Energy-efficient network designs have enabled significant telecom operators to handle growing traffic without proportionate increases in energy use. For instance, Sprint slashed its network energy intensity by over 80% between 2014 and 2019, keeping total energy consumption flat. Telefónica achieved a fivefold increase in data traffic from 2015 to 2020, while its electricity use decreased by 2%. Even in 2020 – a year of unprecedented internet traffic surges – Telefónica saw a 45% growth in data usage without exceeding its 2019 electricity consumption.

Balancing Demand and Sustainability

As global reliance on digital services accelerates, maintaining this balance between rising demand and sustainable energy use will require continued innovations in efficiency. Government and industry stakeholders' commitments toward energy efficiency, renewable energy adoption, and research, development, and demonstration (RD&D) will be essential to curtail energy demand and mitigate emissions over the coming decade.

Mobile Networks Driving Data Evolution

The landscape of data transmission is undergoing dramatic shifts as mobile devices increasingly dominate internet traffic. Between 2017 and 2020, global internet traffic more than doubled, a trend projected to repeat by 2023 if current growth persists. Mobile network traffic, growing at an annual rate of 50%, is outpacing wired and Wi-Fi-only devices, increasing by just 17%.

This shift has substantial implications for energy usage, as mobile networks consume more electricity per gigabyte (kWh/GB) than fixed-line networks. However, technological advancements are rapidly transforming mobile network efficiency.

The Shift to 4G and 5G

By 2022, 4G and 5G networks were expected to carry 83% of global mobile traffic, compared to less than 1% for older 2G networks. These newer technologies are significantly more energy-efficient: 4G is approximately five times more efficient than 3G and 50 times more efficient than 2G.

The energy profile of 5G networks is evolving. While a 5G antenna initially consumes about three times as much electricity as a 4G one, innovations such as sleep mode narrow the gap, potentially reducing the energy differential to 25%. By 2025-2030, 5G networks are projected to be 10 to 20 times more energy-efficient than their 4G counterparts.

Video Streaming and Gaming as Key Drivers

Consumer demand for video streaming and gaming continues to surge, accounting for an estimated 87% of global internet traffic by 2022. Although the average energy consumption for streaming is relatively low, peak usage during "prime time" hours could significantly increase peak internet traffic. It may necessitate expanded infrastructure, potentially raising overall energy use.

Emerging Technologies Increase the Load

Beyond entertainment, emerging technologies such as blockchain, machine learning (ML), and virtual reality (VR) contribute to the growing demand for data services. Blockchain, exemplified by Bitcoin mining, is a particularly energy-intensive application. In 2020, Bitcoin mining alone consumed an estimated 60-70 TWh of electricity, roughly 0.3% of global energy use. By 2021, this figure was on track to reach 100 TWh due to surging cryptocurrency prices.

Machine learning is another major contributor. Training the most significant ML models now requires exponentially increasing computational power. For instance, Facebook reported that its compute demand for ML training was growing at 150% annually, with inference workloads rising by 105% annually – far outpacing the 40% annual growth in its data center energy use.

Balancing Growth and Sustainability

As mobile networks, video streaming, and emerging technologies drive exponential growth in data demand, energy use in data centers and networks could escalate without continued efficiency improvements. Addressing these challenges requires coordinated efforts between governments, industry leaders, and researchers.

Key strategies include:

  1. Developing energy-efficient technologies such as low-power 5G antennas and optimized data transmission protocols.

  2. Leveraging renewable energy sources to power data centers and networks.

  3. Establishing policies to manage the energy impact of emerging technologies like blockchain and ML.

Sustainable innovation will be essential to balance the world’s increasing reliance on digital services with the need to mitigate energy consumption and environmental impact.

ICT Companies Leading Renewable Energy Adoption

Information and communications technology (ICT) companies are at the forefront of renewable energy investment, driven by the need to manage power price volatility, enhance their environmental credentials, and strengthen their brand reputation. Over the past five years, ICT firms accounted for nearly half of global corporate renewable energy procurement.

Hyperscale data center operators are particularly prominent in this area, using power purchase agreements (PPAs) to secure renewable energy. Companies like Google (12 TWh in 2019), Apple (1.7 TWh in FY 2020), and Facebook (7 TWh in 2020) now purchase or generate enough renewable electricity to offset 100% of their operational energy consumption.

Maximizing the Impact of Renewable Energy Investments

While these achievements are noteworthy, purchasing renewable energy equal to annual demand doesn’t guarantee that data centers are powered by clean energy at all times. Solar and wind generation variability often misaligns with real-time energy demands, and renewable energy purchases might support grids in entirely different regions.

Companies like Google and Microsoft are pioneering carbon-free electricity goals to address these challenges, aiming to match their energy needs with zero-carbon sources 24/7 by 2030. Advanced technologies, such as machine learning, can facilitate these efforts by shifting computing tasks to times and locations where low-carbon energy is abundant.

Collaboration between data center operators, regulators, and utilities will be critical to ensure renewable energy projects maximize local grid benefits and reduce overall greenhouse gas emissions.

Recommended Actions for Sustainable ICT Growth

1. Bolster Data Collection and Transparency

Improved data collection and transparency are crucial for energy analysis and policymaking. For instance, the U.S. Energy Information Administration (EIA) gathers information on connected devices and servers. At the same time, the U.S. Energy Act of 2020 mandates updated studies on data center energy use and the development of new efficiency metrics.

2. Commit to Efficiency and Climate Goals

ICT companies are adopting ambitious climate and efficiency targets, including science-based goals to reduce greenhouse gas (GHG) emissions by 45% between 2020 and 2030. In Europe, the Climate Neutral Data Centre Pact aims to achieve climate-neutral data centers by 2030, aligning with the EU’s digital strategy for sustainable ICT operations.

3. Enhance Data Center Efficiency and Grid Flexibility

Data centers have significant potential to improve energy efficiency while offering flexibility to power grids. Governments can incentivize efficiency through standards, regulations, and programs encouraging demand-side flexibility. Best practices and incentive strategies, such as those detailed by Huang and Masanet (2015), provide actionable blueprints.

4. Drive Renewable Energy Integration

Collaboration between governments, grid operators, and data centers is essential to optimize renewable energy investments. Encouraging energy storage and demand-side response capabilities can further align renewable energy generation with consumption needs.

5. Advance Energy-Efficient Network Policies

Accelerating the phaseout of inefficient legacy networks, implementing energy efficiency standards for network devices, and supporting international technology protocols are critical steps. Governments and operators can adopt metrics and incentives to drive efficiency across transmission networks.

6. Invest in RD&D for Next-Generation Technologies

Emerging technologies like AI, blockchain, virtual reality, and 5G are accelerating demand for data services. As efficiency gains from existing technologies plateau, research, development, and deployment (RD&D) of next-generation solutions are vital to maintaining sustainability.

7. Address Lifecycle Environmental Impacts

As data centers transition to clean electricity and greater operational efficiency, the focus must also include reducing embodied emissions from the manufacturing, transportation, and disposal of ICT infrastructure. Companies and policymakers should prioritize sustainable practices across the supply chain to minimize lifecycle environmental impacts.

Balancing Growth with Sustainability

The ICT sector’s growing reliance on renewable energy and advanced efficiency strategies underscores its role as a leader in sustainable innovation. However, meeting future energy demands while curbing environmental impacts will require continued collaboration, investment in cutting-edge technologies, and robust regulatory frameworks to ensure the sector evolves responsibly.