How Much Mobile Operators Care about Energy Consumption?

There was a noticeable increase in interest over the past several months in “Green 5G”. Energy consumption was top-of-mind for CTOs at MWC22. Different industry lobbyists released white-papers praising the energy efficiency of 5G [a few of which are misleading and even utterly wrong]. The jump in energy prices, especially in Europe, became a concern for their inflationary impact on telecom operators. Even US operators were not immune: Verizon cited their concerns on rising labor and energy costs in their most recent earnings call. Some, as in the recent Brooklyn 6G Summit, started to position 6G as the technology for sustainable networks which will compensate for the failures of 5G!

So, what to make of all this talk about sustainable telecom networks? What are the real issues related to power consumption? How important is this issue to the industry? What is the impact of the cost of energy on telecom operators? And, what lessons to learn from this, especially from a technology development perspective?

My thesis is that operators care about power consumption but for different reasons. The sensitivity of operators to energy requirements varies significantly. As such, there is no uniform or homogeneous response. This perhaps explains why sustainability and green telecom is a recurrent issue that peaks when a new technology is introduced. It also provides context for what technologies will matter in the future, and to whom.

The Two Aspects of Power Consumption

There are two aspects to energy in telecom networks. The first aspect I associate with capex and peak power. The second aspect I associate with opex and average power. Understanding these is important to understanding the context for energy in wireless networks.

The capex aspect: The capex aspect focuses on provisioning cell sites for the power load required to run 5G systems. This includes upgrading power rectifiers, backup batteries and even cables to support the additional load. The capex aspect is coupled with peak power for the cell site.

The capex/peak aspect is a common concern to all operators. 5G places a significant burden because it could easily double the power requirements at a cell site. This forces operators to make choices on the type of equipment to deploy. For instance, an Asian operator deployed 32T32R 5G radios because they can manage the respective power load. These radios were later swapped for 64T64R.

The opex aspect: This is the question of on-going energy expenses associated with the running the cell site; a question of average power. McKinsey estimates these costs to operators at 5% of opex [here].

What Operators Care About?

Traditionally, operators are most concerned with the capex/peak aspect. Upgrading existing cell sites to support 5G is a universal pain point that all operators need to resolve [and have been solving]. This aspect is financially lucrative for the value chain: Just think about the power subsystems that need to be upgraded. For instance, laggard operators who still use bulky lead-acid batteries will need to upgrade to lithium-ion in order to save space and obtain the necessary power density to run the cell site. For a perspective, Huawei estimated upgrading power systems to support 5G would cost $2.1 billion for China alone [here].

The operational cost of energy is what operators plan for and target to recoup through the new pricing plans for 5G services. Operational costs vary widely from country to country and between urban and rural areas. Consequently, operators in high energy cost markets are among the first to feel the pressure and would be the most motivated to address escalating energy requirements.

Why is this Important?

Energy consumption increases with every new generation. This is contrary to what industry lobbyist are telling regulators. Moreover, I don’t see a path where the total energy consumption of networks will get lower. A simple glance at what drives energy consumption in wireless technologies gives the idea:

• More spectrum and wider bandwidth: The use of wide channels increases power draw. For instance, the widest LTE carrier bandwidth is 20 MHz, whereas in 5G channel it is 100 MHz. If you consider emission requirements on a W/MHz basis, then this is 5x the power for the same frequency band. In a nutshell, more spectrum means more power.
• Higher modulation: The use of higher modulation rates, such introducing 256 QAM in downlink and 64 QAM in uplink in LTE Advanced reduces the overall efficiency of radios.
• High order MIMO antennas: 64T64R is the state of the art today: it consumes 25% more power than 32T32R. 128x, 192x, and 384x are on the roadmap of vendors [as Huawei showed recently in their industry analyst meeting].
• Densification: adding new cell sites to increase network capacity.

While total power consumption of wireless networks is set to rise, the incremental improvement in power efficiency is diminishing. The remote radio heads (RRH or RRU) account for nearly 2/3rd of macro base station power consumption. Power amplifiers (PAs) account for a very large percentage of the RRU energy consumption. But in 4G and 5G technologies, the improvement in PA technology has stalled, with relatively small margin of improvement with every successive generation.

The key point here is that the absolute energy requirements of mobile networks will keep increasing. This is why I don’t think the energy per bit metric promoted by lobbyists and 5G propagandists is useful, because they mask the core problem and have little purpose than perhaps create confusion and celebrate false achievements!

A Service Provider Example

I’ll use Vodafone as an example because they are among the most transparent in reporting on sustainability (I hope they continue!). The tables below provide a snapshot. The numbers are from Vodafone’s 2021 annual and sustainability reports.

Energy use (GWh)	3/31/2021	% of Total
Total	5,832	100%
Network base station sites	4,239	73%
Technology centres (i.e. switch centres and data centres)	1,358	23%
Offices	201	3%
Retail	33	1%

Vodafone’s radio access network accounts for the majority of energy use at 73%. Ericsson’s Sustainability Report provides a similar figure: 75% [here].

	3/31/2021	Energy cost, % of
Group Revenue, €m	43,809	1.73%
Group service revenue, €m	37,141	2.05%
Cost of sales, €m	30,086	2.53%

To put the numbers in perspective, a single cell site consumes about the same as 2.4 US residential consumers (10.7 kWh according to the EIA). The average cost of electricity is a relatively small part of the COGS for Vodafone. A sudden and large increase in energy costs would have a noticeable impact on margins. However, in the big picture, there are larger expenses that are easier to reduce if one is looking to cut costs!

Update: Vodafone published their FY2022 (FY ending March 31, 2022) numbers after I wrote this article. They reported €846m in energy expenses: 11% increase over FY2021. Energy accounted for 1.86% of revenue and 2.8% of COGS. In retrospect, this is not a large increase if one accounts for roughly 3% increase in site count and some additional 5G sites.

Large Regional Variations

There are large regional and national variations in the cost of energy. For instance, the cost of power in North America for industrial and commercial consumers is typically lower than that in many Asian countries. However, events over the past two years have changed the order of electricity prices, so we see now Europe countries paying among the highest prices. In fact, I have seen reports of operators paying as much as 20%-40% of opex for electricity [here].

The point here is that MNOs in high energy cost countries will be much more motivated to explore energy saving solutions to optimize power consumption. Before Covid, it used to be the Asian operators, but now, European operators ought to have the most incentives as they stand to lose the most from the rise of energy prices.

Energy Saving Technologies

I will be brief in summarizing energy saving technologies in wireless networks since this deserves an article of its own. For the purpose here, I classify energy saving technologies into two categories:

1. Hardware technologies: Hardware impacts both the capex/peak aspect and the opex/average aspect of power consumption. Innovation in equipment design remains a fundamental element in saving energy. Unfortunately, there is a diminishing return for hardware innovations. The improvement in efficiency related to hardware over successive generations is shrinking.
2. Software technologies: This is related to the analytics and intelligence that runs on top of the network to optimize operation depending on the traffic and other patterns [see more here]. Software techniques primarily impact the opex/average aspect of power consumption at a cell site. The key question here is whether the business case makes sense for the operators to implement such technologies. The answer lies in the operating environment and the cost of power.

From this, I would suggest that operators pressed with high energy costs will be the first movers to deploy software technologies.

Final Takeaways

1. Traditionally operators are most concerned about the capex spending related to deploying a new technology such as 5G. This one-time expense typically exceed the energy opex which is factored into the business plan.
2. Operators in high energy cost areas are most sensitive to increases in energy prices. When this is coupled with a sudden rise in energy prices, such as in Europe, it becomes very problematic and would adversely impact margins.
3. On the other hand, operators with low energy cost areas can tolerate higher energy costs. Where energy makes a relatively low percentage of opex, they will have other areas where it’s easier to cut costs.
4. Operators in high energy price areas have the most incentives to validate the business case for software-based energy saving solutions.
5. Operators don’t exhibit the same urgency and interest in sustainability largely because of their operating environment.
6. Efficiency of wireless equipment is diminishing with each successive generation. Software solutions will have more value over time, but the adoption will start in regions with high cost of energy where the business case is easier to validate.

Lastly, green telecom will be increasingly subject to misleading and false marketing as companies, industry associations and lobbyists seek to promote their interest and push their agenda. There’s a lot at stake.

---

Contact me if interested to discuss this topic further, or the power consumption of data centres or crypto currencies [see here for bitcoin power consumption analysis].