A report from GSMA Intelligence has identified how AI (artificial intelligence) can drive energy efficiency in telecom networks.
AI for Optimized Site Selection
AI uses traffic patterns and population density to optimize site placement, reducing energy waste in low-traffic areas. Huawei is a leader in site planning with its AI-integrated products.
Extending Equipment Lifespans
AI helps reduce equipment wear and tear by optimizing energy use, extending equipment lifespans, and minimizing maintenance costs. European vendors like Ericsson and Nokia focus on recycling and extending equipment lifespans as part of the circular economy.
Smarter Energy Use in the RAN
The RAN, responsible for most network energy consumption, benefits from AI-driven techniques like real-time spectrum carrier selection based on demand. ZTE’s Reconfigurable Intelligent Surface (RIS) technology also optimizes beam usage around obstacles, reducing energy costs.
AI for Performance Monitoring and Digital Twins
AI replaces traditional drive tests with digital twins and virtual testing using real network data, reducing energy used by fleets, which account for 5-10 percent of operator energy consumption.
Efficient Data Center Energy Management
Although hyperscalers own most data center energy usage, operator-owned centers powering the core still use about 10 percent of their total energy. AI can dynamically shift traffic to underutilized sites, saving energy in dense urban areas.
Energy Savings at the Edge
AI helps retain traffic at the edge, reducing backhaul and cooling costs by keeping data closer to users. This also eases the load on larger data centers and infrastructure under high demand.
AI-Driven Energy Reductions Across Network Levels
AI interventions, such as RAN shutdowns, dynamic spectrum usage, optimized site planning, and digital twins, are at different stages of deployment but offer significant energy savings potential, with annual reduction rates from 15 percent to 50 percent.
Growth in Mobile Data Traffic
Global mobile data traffic is increasing by nearly 30 percent per year, driven by 5G adoption. 5G users consume 2-3 times more data than 4G users, with total data traffic expected to be six times higher by 2030 compared to 2023.
Infrastructure and Energy Demands
The increase in traffic requires additional infrastructure in the RAN, core, and backhaul, which leads to higher energy demands. While AI contributes to higher computing needs, the primary energy strain comes from handling larger traffic loads and processing demands, both in operator networks and at data centers.
Rising Energy Costs for Telecom Operators
Despite 5G’s efficiency improvements over previous generations, its higher traffic volumes and complexity result in increased power consumption. Operators are addressing these costs by optimizing energy use in base stations, core networks, and transitioning from copper to more efficient fiber infrastructure.
AI-Driven Pressure on Fixed Line and Transport Layers
With AI applications, energy demand is also rising in the fixed-line infrastructure, especially as copper (which is 5× less efficient than fiber) connects mobile networks to data centers, further adding to the energy pressures on network transport layers.
Energy Costs in Operators’ Expenses
Energy accounts for 15-20 percent of operators’ operating expenses (opex), and this figure is even higher in regions like Africa and India, where diesel is often used in rural areas. Reducing these energy costs can significantly improve profit margins and provide long-term financial stability.
Cost Savings from Energy Efficiency
Vodafone’s efforts to improve power supply efficiency and optimize network equipment saved over €100 million in energy costs from 2015 to 2020, before the integration of AI.
Impact of Energy Cost Reduction on Profit Margins
Reducing energy opex by 20 percent could raise an average operator’s EBITDA margin by 3 percentage points, highlighting the financial value of energy-saving initiatives. AI is expected to play an increasing role in these efforts as it becomes more integrated into network operations.
Energy Price Volatility and AI’s Role in Mitigation
While global energy prices remain elevated, operators can use AI to optimize energy consumption, particularly in the RAN, which consumes about 75 percent of network energy. Long-term investments in AI are essential to help operators control these rising costs and mitigate future energy challenges.
AI for Energy Efficiency & Cost Savings
Telecom operators are leveraging AI to optimize energy use, reducing operational costs and supporting global sustainability initiatives.
AI-driven solutions enhance network management, reliability, and customer satisfaction, which encourages further innovation within the industry.
Sustainability as a Competitive Advantage
Energy-efficient solutions are becoming a key selling point for vendors, particularly in regions with high energy costs or strict environmental regulations.
While most AI innovations currently target the RAN (which accounts for 75 percent of operator energy use), they are expected to expand into core and edge networks in the future.
AI Infrastructure to Support Workloads
AI advancements enable telecom operators to handle AI workloads more effectively, fostering continuous innovation that benefits both networks and their users.
Impact of Regulatory Oversight
Business motivations for AI adoption center on cost savings, revenue growth, and regulatory compliance.
Increasing regulatory scrutiny over AI practices may influence how telecom operators deploy and manage AI, shaping their strategic choices.
Increased Energy Demand with 5G and AI
5G and AI have significantly raised energy consumption, with telecom networks and data centers each accounting for about 1 percent of global electricity. Rapidly growing data traffic, especially from AI-powered applications in video, gaming, and enterprise, is expected to further strain networks.
AI compute for training and inferencing large language models (LLMs) could push cloud energy consumption up by 30-60 percent by 2030, reaching 1.5-2.0 percent of global energy use.
The Scale of AI’s Energy Impact
AI’s incremental energy demand by 2030 could equal the annual consumption of a populous country like Egypt, emphasizing the significant challenge of managing energy and environmental costs tied to AI expansion.
AI for Energy Efficiency in Telecoms
AI can help telecom operators cut energy costs through smarter practices, like RAN shutdowns and dynamic spectrum management, with equipment vendors leading these innovations. Though energy demand may peak in the next few years, AI investments are expected to yield cost savings in operational expenses (opex) and capital expenses (capex).
Supporting AI Workloads for Enterprises
Telecom networks will be essential for enabling AI in enterprises, requiring higher and smarter network capacity to handle growing workloads driven by AI compute demands.
Long-Term AI Integration and Energy Impact
AI in telecoms will likely progress in three phases this decade: early adopters, broader industry integration, and eventual AI-native operations where AI and automation drive efficiencies across functions. Achieving energy neutrality, where AI-driven efficiencies offset increased power use, will require meticulous tracking, including granular benchmarking of energy consumption for AI tasks such as LLM training versus inference.
Sustainability Priority for Operators
70 percent of telecom operators view sustainability as a “very” or “extremely important” goal in network transformation, second only to network security (87 percent).
Energy Consumption in Telecom Networks
The Radio Access Network (RAN) consumes 76 percent of a network’s energy, with the network core and owned data centers using 19 percent, and other operations making up the remaining 5 percent.
Telecom Networks’ Global Energy Impact
Telecom operator networks (fixed and mobile) account for about 1 percent of global energy use, equivalent to around 340 terawatt hours (TWh), comparable to the energy usage of a large European country.
Increased Data Center Energy Demand from AI
AI-driven training and inference for large language models (LLMs) could increase data center energy usage by 60 percent by 2030.
Turkcell’s AI-Driven Sustainability Efforts
Turkcell has established AI Principles for ethical AI use in its operations and set sustainability targets: sourcing 100 percent renewable electricity by 2030 and achieving net-zero emissions by 2050.
In 2022, Turkcell’s energy consumption equaled that of Türkiye’s 16th largest city, with 5G projected to increase usage further. Traditional energy-saving methods, like carrier shutdown, provided limited benefits, as at least one carrier had to remain active to maintain coverage.
Turkcell developed an AI-powered algorithm to predict low-traffic sectors, allowing selective sector shutdowns without service disruption. This approach minimizes redundant energy use by precisely analyzing network coverage and geolocating data from call trace logs, enabling data-driven energy optimization.
Telenor’s AI-First Program for Sustainable Network Operations
Telenor’s AI-first program emphasizes responsible, secure, and sustainable AI use across business areas, with initiatives in data modernization and workforce upskilling.
Telenor is aligned with the mobile industry’s goal of reaching net-zero emissions by 2050, focusing on digital advancements that promote a greener, safer, and more accessible digital ecosystem.
In collaboration with Ericsson, Telenor has developed a machine learning (ML) solution to enhance network energy efficiency without sacrificing connectivity quality.
Using reinforcement learning, the AI agent autonomously adjusts network settings to control energy consumption, particularly through cell sleep-mode activations in the RAN. The AI system balances energy savings with optimal network performance, representing a step towards self-managing, energy-efficient networks.
Baburajan Kizhakedath
