Global mobile data traffic has crossed 220 exabytes per month, but telecom operators are not increasing network spending at the same rate. The result is one of the most important changes in telecom economics: the cost of carrying each additional gigabyte is falling as networks become more efficient and traffic scales.

There is no standardized industry figure for the fully loaded network cost of 1 GB. Operators rarely disclose a metric combining radio access, towers, spectrum, fiber, core networks, energy, maintenance, depreciation and Capex.
But 2026 data from Jio, Airtel, Telefonica, Deutsche Telekom and other operators demonstrates how 5G, higher spectral efficiency, network automation and lower energy intensity are changing cost-per-GB economics.
Global Mobile Traffic Exceeds 220 EB per Month
Global mobile network data traffic exceeded 220 EB per month in Q2 2026, increasing 23 percent year over year and about 6 percent quarter over quarter.
Traffic has increased from slightly above 30 EB per month in Q2 2019, meaning networks are carrying roughly seven times as much data as seven years ago.
Video represented around 75 percent of mobile data traffic at the end of 2025.
The Ericsson mobile traffic analysis highlights why scale matters to operators.
Many telecom costs are fixed or semi-fixed. Spectrum licenses, towers, fiber routes and core platforms do not become proportionally more expensive every time another customer downloads 1 GB.
If traffic grows faster than network costs, cost per GB declines even when total telecom Capex remains high.
Jio Carries 241 EB of Data in One Year
Reliance Jio demonstrates this scale effect particularly clearly.
Jio carried 241.4 billion GB — or 241.4 EB — during FY2026, increasing 30.8 percent year over year. The operator had more than 524 million customers and 268 million 5G users at March 2026.
Its 5G network accounted for 55 percent of wireless data traffic, while Jio’s overall network carried around 60 percent of India’s data traffic.
By Q1 FY2027, average monthly data consumption had reached 43.7 GB per subscriber.
The economics are powerful. The same nationwide infrastructure is carrying progressively larger volumes of data, allowing spectrum, radio, fiber and core-network costs to be distributed across more gigabytes.
Jio’s FY2026 operating data shows why network utilization is increasingly as important as subscriber numbers.
Airtel Users Consume 34.4 GB per Month
Bharti Airtel provides another Indian benchmark.
Average mobile data usage reached approximately 34.4 GB per customer per month in Q1 FY2027, while India mobile ARPU increased to ₹264.
Rising consumption creates additional capacity requirements but also improves infrastructure utilization.
Airtel’s network emissions intensity has fallen to approximately 0.04 tonnes of CO2e per TB, indicating improving efficiency as traffic increases.
The operator is simultaneously investing beyond conventional mobile infrastructure. Airtel announced a $1 billion investment in Nxtra Data in March 2026, reflecting the growing connection between telecom networks, data centers and AI infrastructure.
Telefonica Cuts Energy per Traffic Unit by 92%
Telefonica provides perhaps the clearest published example of declining network resource cost per GB.
Its network infrastructure consumed 29 MWh per petabyte of traffic in 2025, equivalent to approximately 0.029 kWh per GB.
That represents a 92 percent reduction in energy consumption per unit of traffic since 2015.
Even more strikingly, Telefonica reduced total energy consumption by 12 percent between 2015 and 2025 despite network data traffic increasing twelvefold.
The operator attributes the improvement to fiber migration, 5G, legacy-network shutdowns, virtualization and automation.
Telefonica’s network efficiency data demonstrates an important telecom economics principle: traffic can grow dramatically without energy costs rising at the same rate.
Deutsche Telekom Cuts 5G Core Energy by Up to 65%
Deutsche Telekom is taking network efficiency further through software-controlled infrastructure.
Its new Full Stack Energy Efficiency approach delivered up to 65 percent lower energy consumption in initial live 5G core tests.
Instead of running computing and network components continuously, Deutsche Telekom dynamically adjusts resources according to actual traffic demand.
Future versions will use AI algorithms to predict traffic and determine when components should be activated.
The operator calls the principle “Zero Bits, Zero Watts” — minimizing electricity consumption when equipment is not carrying traffic.
Deutsche Telekom also reported that overall energy intensity declined about 16 percent to 48 kWh per TB in 2025.
Its 5G core energy-efficiency program shows how AI and automation could become direct tools for reducing network Opex.
Vodafone Shows Why Network Scale Matters
Vodafone operates approximately 158,000 mobile sites across Europe and Africa and serves around 279 million mobile customers across controlled operations.
Its FY2026 revenue reached €40.5 billion, with adjusted EBITDAaL of €11.4 billion.
Vodafone also delivered around €700 million of European Opex savings between FY2023 and FY2026.
A network of 158,000 sites has enormous fixed costs covering power, leases, maintenance, transmission and equipment. Increasing the traffic handled by each site without proportional cost growth can therefore materially reduce unit economics.
Network sharing can provide another lever by allowing operators to distribute infrastructure costs across larger traffic volumes.
NTT DOCOMO Targets 4x Capacity Efficiency
Hardware improvements are also changing cost-per-GB economics.
NTT DOCOMO announced in July that Ericsson’s RAN Processor 6672 could provide up to four times higher capacity efficiency while reducing energy consumption by more than 50 percent compared with the previous generation.
DOCOMO serves more than 91 million subscribers and uses 280 MHz of combined spectrum across its 3.5 GHz, 3.7 GHz and 4.5 GHz bands.
This represents the next stage of network optimization: increasing traffic capacity without proportionally adding radios, processing equipment or electricity consumption.
Capex Remains High Despite Falling Cost per GB
Lower unit costs do not mean telecom operators can reduce investment.
Verizon expects $16 billion-$16.5 billion of 2026 Capex.
AT&T expects annual capital investment of approximately $23 billion-$24 billion between 2026 and 2028.
Orange invested €3.2 billion in eCAPEX during H1 2026, equivalent to 15.2 percent of revenue.
Deutsche Telekom’s H1 2026 cash Capex reached approximately €8.4 billion including spectrum investment.
Operators are investing ahead of traffic growth. Today’s spectrum, radios and fiber can support several years of additional data consumption, making a simple annual Capex-divided-by-traffic calculation potentially misleading.
GSMA: Data Energy Intensity Falls 15% Annually
The broader industry trend supports the operator examples.
GSMA data covering 18 operators shows that the energy intensity of data transmission declined by an average 15 percent annually between 2019 and 2025.
Meanwhile, mobile traffic more than quadrupled between 2019 and 2024.
Renewable electricity represented 24 percent of operator electricity consumption in 2024, up from 10 percent in 2019.
Energy remains only one component of network cost, but the trend illustrates how technology improvements allow networks to carry substantially more traffic without equivalent growth in resources consumed.
So What Does It Cost to Carry 1 GB?
There is no credible universal 2026 dollar figure.
A proper calculation would need to allocate energy, RAN Opex, tower leases, fiber and microwave backhaul, spectrum costs, core-network operations, personnel, maintenance, depreciation and network Capex.
The basic equation is:
Network cost per GB = allocated network costs ÷ total data traffic
But the numerator and accounting treatment vary significantly between operators.
The more useful comparison is therefore the direction of travel.
Jio carried 241 EB in FY2026. Telefonica reached 29 MWh per PB. Deutsche Telekom demonstrated up to 65 percent 5G-core energy savings, while new DOCOMO equipment promises more than 50 percent lower energy consumption alongside higher capacity efficiency.
Cost per GB Is Becoming a Scale-and-Efficiency Game
Telecom economics in 2026 can increasingly be summarized as:
More traffic + better spectrum utilization + higher-capacity 5G + lower energy intensity + automation = lower network cost per GB.
Operators will continue spending billions on spectrum, fiber, radios and core networks because total traffic keeps increasing.
The strategic objective is therefore not simply cutting network expenditure. It is carrying more gigabytes over every dollar invested in spectrum, sites, fiber, equipment and electricity.
As mobile traffic moves toward the multi-zettabyte era, cost per GB will become an increasingly important measure of network efficiency — even if operators do not publicly report the metric itself.
FASNA SHABEER
