opex Archives – MTN Consulting

With each passing day, the 2G and 3G layers of telcos’ mobile networks are looming as heavy loads on operating expenses (opex). That’s due to multiple issues but especially energy consumption and related costs. With the existence of a 4G layer in these networks and the coming, if not already deployed, energy-hungry 5G layer, such loads become even heavier. Even before 5G has become widespread, energy costs averaged to about 4% of telco opex in 2021, based on an MTN Consulting study.

Quantifying base station energy costs by generation

Many telcos publish data on their energy consumption, and sometimes provide breakdowns for different parts of the network. But there are no existing estimates on the specific impact of maintaining 2G/3G networks alongside 5G. This blog attempts to quantify the effect of these older 2/3G mobile networks on a typical telco’s energy bill.

To accomplish this, first, we need to have a closer look at overall electrical energy consumption for mobile telcos, and then break down this consumption into parts, identifying what portion of total energy relates to the base station and what parts of the base station consume the most energy.

One complication is that a pure “mobile telco” is rare: most telcos providing mobile services also provide many other services, and operate network assets well beyond the mobile RAN. Some telcos providing mobile services began their lives long ago as fixed operators. Some started as mobile but acquired or built fixed assets to support converged offerings. Some have provided both fixed and mobile services from the start. Some provide cloud or other services mainly aimed at enterprise markets. The energy consumption patterns differ across operator types. Figure 1 illustrates this, for a few large telco groups.

Figure 1: Mobile network as % of total network usage, select telcos

Source: public reports and MTN Consulting estimates

To remove this confusion, we will consider mobile telecom companies that still rely exclusively on providing mobile telecom services. For such operators, the mobile network accounts for about 90% of total company energy consumption and costs. There is some limited variation around this 90% figure, due to vendor choice, network topology, and traffic mix, but 90% is a reasonable estimate.

Taking KDDI as an example from Figure 1, this company provides a range of services that are not mobile related; MTN Consulting estimates that KDDI’s mobile network accounts for only about 60% of total company energy consumption. But for Zain, this ratio is 93% as this company is almost exclusively focused on mobile services.

After concluding that about 90% of a mobile-only network provider’s energy consumption is from the mobile network, we need to dive deeper inside the mobile network to find the network elements that contribute most directly to energy consumption.

The mobile network consists of different parts like core, transport, and base stations. As shown in Figure 2, the base stations, or the mobile radio access network (mobile RAN), account for about 57% of network energy consumption for a mobile operator. Expressed differently, the mobile RAN accounts for about 51.3% (i.e. 90% * 57%) of total company energy consumption for a mobile-only operator, such as Zain.

Figure 2: Base station’s contribution to mobile network energy consumption

Source: IEEE Communications Surveys and Tutorials

Now we need to examine the base station and have a closer look at the base station elements and their corresponding energy consumption. As shown in figure 3, the base station element that consumes the largest portion of energy is the power amplifier (PA), which consumes around 75% of total base station energy consumption as shown in figure 3.

Figure 3: Base station energy consumption distribution by network element

Source: Journal of Energy

As shown above, the power amplifier element is the biggest energy consumer in a mobile-only telco network. The power amplifier accounts for about 38% of total company energy consumption: 75% * 57% * 90%. So if a mobile network operator turns off the 2G network layer, the bulk of energy savings will come from shutting down the power amplifier that corresponds to the 2G network.

Now that we have quantified the amplifier’s contribution to total energy use, will this 38% figure be enough to measure the benefits of shutting down the 2G and 3G layers? Actually, we still need one more number: the energy consumption of the power amplifier for each technology. In other words, what is the energy consumption percentage for each of these technology layers? Let us have a look at this in Figure 4.

Figure 4 is a presentation of the key components in a base station and their typical energy consumption, in three different network configurations. The columns show the configuration of a typical base station, and the rows are the affecting elements, mainly the power amplifiers. The gray colored boxes are the elements needed for the 2G and the 3G layer, blue colored boxes are the elements needed for the 4G layer, green colored boxes are the elements needed for 2G, 3G, and 4G, and lastly, the orange colored boxes are elements needed for the 5G layer.

Figure 4: Energy consumption of key components in a base station, across three network configurations

Sources: MTN Consulting; Huawei Technologies

The values inside the elements represent the maximum energy consumption of that element. So as shown in figure 4, in 2G/3G only, the base station consumes 3.9kWh. By adding a 4G layer onto the base station, you increase the energy consumption of this base station by 51%. By adding a 5G layer on top of the 2G, 3G and 4G base station, you can expect another 66% increase in energy consumption. The red arrows in Figure 4 indicate these increases in maximum energy consumption.

That 66% figure illustrates one thing that is scary about 5G: yes, it may offer revenue upside, but it also consumes lots of power to operate 5G, which costs money and has climate impacts.

For this blog, though, what we really need to know is the contribution to total energy consumption of the legacy network elements as you upgrade to newer technology. These values are presented in figure 4 in the yellow arrows. So, in 2G/3G/4G base stations, about 40% additional energy is consumed by holding on to the 2G and the 3G layer. Similarly, in a combined 2G/3G/4G/5G base station, roughly 24% extra energy is consumed because of still holding the 2G and the 3G layers in this base station.

We now have the estimates we need to find the impact of supporting the 2G/3G network layers alongside 4G/5G for a typical mobile operator.

Approaches to 2G/3G network shutdowns vary depending on current network design

Right now telcos are wrestling with the best way to deal with legacy networks, while they upgrade to 5G. The ideal solution depends on the current position of the operator’s network. In the following table, you can see some potential paths to 2G/3G network shutdowns and the impact on energy consumption. In the first scenario, where the telco currently has only 2G and 3G network layers, an upgrade straight to 5G would result in an approximate 50% reduction in energy costs. In the last scenario shown on the bottom of the table, where an operator is simultaneously operating 2G, 3G, 4G, and 5G networks, the ideal solution is to shut down the 2G and 3G layers. This would save an estimated 40% in base station energy consumption.

Of course, energy costs are not the only factor in planning 2G/3G network shutdowns: spectrum, regulatory, legacy service revenue, and other factors also matter. But telcos nowadays are very focused on reducing their opex burden whenever possible, given weak revenue growth. As such, energy costs are a central focus of most telcos.

Table 1: Energy usage impact of 2G/3G shutdown scenarios

Mobile network’s current scope	Likely migration path for 2G/3G shutdowns	% reduction in energy use for typical base station
2G and 3G network	Upgrade straight to 5G, bypassing 4G, then shut down both 2G and 3G	50.0%
2G and 4G network	Upgrade to 5G then shut down the 2G layer	30.1%
3G and 4G network	Upgrade to 5G, then shut down the 3G layer	33.2%
2G, 3G and 4G network	Upgrade to 5G, then shut down both 2G and 3G layers	39.9%
3G, 4G and 5G network	After upgrading all sites to 5G, shut down the 3G layer	33.2%
2G, 4G and 5G network	After upgrading all sites to 5G, shut down the 2G layer	30.1%
2G, 3G, 4G and 5G network	The operator is ready to shutdown both the 2G layer and the 3G layers	39.9%

Source: MTN Consulting

The above table represents cost savings for the “typical” mobile-only telco we described early in the blog. In follow-up blogs, we expect to detail the impacts of 2G/3G network shutdowns for a few specific mobile operators in different regions of the world.

The author, Samir Ahmad, is a telecommunications and IT consultant based in Amman, Jordan. Samir has a Master’s in Telecommunication, Electrical, Electronics, and Communications Engineering, from the University of Sydney, and a B.S. in Electrical Engineering – Communications & Electronics, from the Jordan University of Science & Technology. Prior to entering the consulting field in 2017, Samir worked for Zain Jordan for 8+ years, most recently as Expert, RF Planning and Optimization.

One of the many telecom stats we track is “labor costs”, i.e. what telcos spend in salaries and benefits to support their workforce. Not a lot of other analyst firms track labor costs, if any. It’s not an easy one to track, as telcos aren’t required to report it, and the data can be hidden. But it is essential to understanding the telco’s business and challenges they face.

What’s important to know about labor costs and the telco workforce in general?

Labor costs represent a lot of money. The capex spent by telcos on their networks gets lots of attention, and for good reason. It’s a huge cost, at roughly 17% of revenues across the globe. Yet spending on the workforce is nearly as much as capex. In 2021, telco capex was $326 billion (B), while labor costs totaled $273B. Some telcos spend significantly more on labor costs than capex. Saudi Telecom (stc), for instance, spent $2.6B on labor costs in 2021, 60% more than that year’s $1.6B capex figure for the company. The labor cost to capex ratio exceeded 1 for a number of other large telcos in 2021, including: Chunghwa (1.26), Orange (1.10), Singtel (1.2), Telecom Argentina (1.20), Telefonica (1.23), and Telstra (1.26).

Labor costs are not just salaries. When we say “labor costs”, we mean to capture the fully-loaded cost of an employee. That includes salaries and wages, short-term benefits, retirement benefits, any required government contributions, and share-based compensation. Labor costs are around 20-60% more than just direct salaries, depending on the company. Some companies report the breakout of the various categories, but many do not. For Verizon and AT&T, we estimate the labor cost to salary ratio as 1.4.

Headcount is falling. As MTN Consulting detailed in its 4Q21 market review for the telco sector, headcount in the telco industry continued to fall last year. Total employees dropped 2% in 2021, to 4.69 million. Only 5 of the top 20 telcos increased headcount in 2021. The 2021 decline follows a much worse 3.7% drop for the industry in 2020, when COVID forced cutbacks, office closures and acceleration of timelines for digital transformation and automation programs. Prior to COVID, telcos were already in staff-cutting mode, but COVID sped up the process. Looking down the road a bit, telco headcount should fall to ~4.4M by 2026. Figure 1 illustrates changes in 2021 for the top 20 telco employers.

Figure 1: Total employees and YoY % change of top 20 telcos, 2021

Source: MTN Consulting

Labor cost per employee is rising. Amidst this drop in headcount, average labor costs have been rising. From $49.2K in 2017, the average telco employee in 2021 cost $57.6K, or 17% more. Prior to 2018, labor costs per employee were flat for most of the last decade, hovering around US$50K per employee. The inflationary pressures of late 2021 may have played a small role, but more important is the changing nature of a telco. With deployment of software-based platforms in the network, and digitization of a whole range of processes across the company, a different type of employee is required. Some may be younger, but their skills are in demand and can be costly. There is rising competition for these employee types, including from cloud providers like GCP and AWS.

Labor costs a large part of opex. The rising cost per employee is interesting, but even more important may be labor cost’s contribution to opex. As a percentage of opex, excluding the non-cash items of depreciation & amortization, labor costs can exceed 30%. That was the case for a number of large telcos in 2021, including BCE (labor costs equal to 32.6% of opex ex-D&A), BT (31.7%), KPN (30.8%), Swisscom (39.8%), and Telecom Italia (35.9%). Some of these companies face constraints in how they structure their workforce, for instance, union rules limiting layoffs or locking in salary increases. On average, labor costs account for about 22% of opex ex-D&A.

The labor cost burden isn’t equal across markets. Telcos across the globe face similar price levels for their technology inputs. Prices vary somewhat, of course, but the variation is rarely on the order of 3-5x. More important is the variation in technology choices made across different markets, and the way they finance capex. For labor costs, though, the variation in the cost of an employee can be huge. Labor markets are highly localized, even with a more remote/hybrid workforce than in the past. The average employee at UAE-based Du, for instance, cost US$199.8K in 2021, nearly 6x that of another UAE-based telco, Etisalat ($34.0K). The reason for that is most of Etisalat’s workforce is in lower cost countries such as Pakistan, Egypt, and Morocco, whereas Du operates solely out of high-cost UAE.

Telcos investing in upskilling. As telcos deploy more software in their networks and digitally transform all aspects of their operations (including sales & customer support), many are investing heavily in upskilling their employee base. Vodafone, for instance, says “the transformation into a new generation connectivity and digital services provider requires new skills and capabilities in our organization, such as software engineering, automation and data analysis.” Vodafone invested an average of 470 Euros in FY2021 on “training each employee to build future capabilities.” Similar things are occurring at many other telcos, including Deutsche Telekom, which is investing in “upskilling and reskilling programs with a focus on digital skills”. There is also a growing focus on hiring younger employees with skills more appropriate to the digital age.

Propensity to adopt automation varies widely. Vendors talk a lot about how their solutions allow customers to do more with less: automate tasks and processes which previously required manual intervention. This has always been a part of the telecom industry, from the days when telcos migrated away from manual telephone switchboards. It continues to be important as telcos aim to lower their cost of operations, deploy services more rapidly, and maintain network quality. The importance of automation varies widely across country and operator, however. Companies which face high unit labor costs tend to be more eager to adopt automation, all else equal. When labor costs are a relatively high portion of overall opex, that eagerness multiplies. Figure 2 below illustrates the issue.

Figure 2: Labor cost burden variation across 30 large telcos, 2021

Source: MTN Consulting
Notes: Size of bubble is indicator of relative revenues. Red star icon represents the global average.

For 30 large telcos, the above figure shows labor cost as a % of opex (ex-D&A) on the x-axis, and labor cost per employee on the y-axis. The companies in the top right quadrant tend to be more open to automation, while the bottom left (low labor costs) are the opposite. Swisscom is a bit of an outlier, as its labor costs are so high. That’s a reason for Swisscom’s adoption of Red Hat’s Ansible Automation Platform in 2018, for instance. Other big telcos with an economic inclination to automate include: Telefonica, DT, Telstra, NTT, Orange, BT, BCE, and Telecom Italia.

One thing that telcos won’t be automating anytime soon is the CEO. The top few managers in many leading telcos continue to earn millions of US$ per year, and there often seems to be little relationship between these sky-high pay packages and the company’s performance. Light Reading detailed this situation recently in an insightful article. LR notes that the ratio of CEOs’ pay packages with the median employee in 2021 was 312:1 for T-Mobile, 231:1 for AT&T, 166:1 for Verizon, and 106:1 for Telefonica. Swisscom’s CEO Urs Schaeppi had to make do with a relatively paltry margin of 14:1.

Cover image credit: Scott Webb on Unsplash

Tag: opex

Blog Details

Quantifying the energy cost savings from 2G/3G network shutdowns

Blog Details

Telcos are upgrading their workforce, but it comes at a price