Network Architecture Changes Explain why 5G is More Expensive than 4G

One often hears it said that 5G networks are more expensive than 4G networks, which are in turn more expensive than 3G networks. It is likely an accurate observation, but might be explained quite simply by changes in network architecture requiring use of smaller cells. 

And the simple reality is that networks with smaller cells require many more cell sites. The general rule is that shrinking the cell radius by 50 percent requires four times the total number of radio sites. 

And one might argue that, overall, cell sizes are getting smaller.

Network Generation	Typical Macrocell Radius
3G	1 - 5 miles
4G	0.5 - 3 miles
5G	0.25 - 1 mile

And that has clear implications for mobile network costs, as many more cell sites must be supported to accommodate the physical characteristics of radio frequencies in higher ranges that do not travel as far as frequencies in the lower ranges. 

Frequency Band	Frequency Range	Typical Propagation Distance
Low-Band	600-850 MHz	Up to 25 miles (40 km)
Mid-Band	1-3 GHz	1 to 12 miles (1.6-19 km)
C-Band	3.7 GHz	0.17 * Low-Band range
High-Band	24-47 GHz	50 to 2,000 feet (15-600 m)

So, if for no other reason that sheer number of radio sites, mobile networks using the newer frequencies--which do not travel as far as signals in the lower bands--are required to build more cell sizes. 

And even if those smaller cell sites are not always as expensive as macrocell sites once used, there are simply so many more sites to support. 5G networks, which use higher frequencies, can require 15 to 20 sites per square kilometer in urban areas, compared to only two to five sites for 4G networks using lower-frequency signals. 

The other issue is that indoor coverage becomes more difficult as frequencies increase. And that can add capex or opex cost as well. 5G indoor solutions can cost more than double that of legacy 4G-only indoor solutions. And costs three to 10 times higher are also possible. 

"Rule of Three" in Global Capacity Markets: Why Hyperscalers "Do It Yourself"

Podcast of this article

Alphabet, Meta, Microsoft and Amazon account for 71 percent of all used international capacity on undersea cables, up from less than 10 percent 10 years ago, according to TeleGeography researchers. 

Some might find that shocking. Some observers might worry about market structures; economic impact or other social and political concerns. 

But such statistics would not be surprising in any number of industries. 

The "Rule of Three," for example, was developed by Bruce Henderson, the founder of the Boston Consulting Group (BCG). This rule suggests that in any mature industry, there will typically be three major competitors that dominate the market, while the rest struggle to maintain profitability.

Those top three companies often control 70 percent to 90 percent of the market, and, as a corollary, some would note, profitability also tends to track with market share. In many markets, profit margins for the number-one firm are double that of the number-two provider. Margins for the number-two provider are, in turn, double that of number three. 

That pattern suggests a 40-20-10 market share and profit margin pattern, sometimes reaching a 50-25-12 structure. 

Not all industries match the “Rule of Three” pattern, however. Industries where the Rule of Three tends to apply best are:

• Capital-intensive

• In consumer goods

• In technology hardware or software

By definition, internet capacity businesses are capital intensive and in the technology space, but also are platforms to support internet app and service businesses that are heavily consumer oriented, mass market products. 

The Rule of Three doesn't tend to work well in industries that are:

• Highly fragmented

• Regulated industries

• Rapidly evolving markets

• Niche markets

• Geographically dispersed markets

The main point is that global capacity markets, like other forms of communications and computing infrastructure, are good candidates for “Rule of Three” market structures. The market is relatively mature; capital intensive; based on technology and highly leveraged to consumer demand. 

Capacity demand also is asymmetrical, as suppliers are centered in North America and Northeast Asia, while users are disproportionately concentrated in North America; Europe and Northeast Asia, though global activity continues to spread. 

The Rule of Three also illustrate why supplier scale--and consumer preferences--have become so important for app and service providers in the internet era, for app or service providers who operate in consumer markets globally. 

Researchers tend to agree that some consumer apps get used more often than others, and therefore are the more-important drivers of capacity usage, as a result. Globally, a relative handful of apps are used by mobile and desktop users and the list of suppliers of those apps is a similar handful. 

Mobile App	Desktop App/Service	Usage Metric	Supplier
Facebook	Facebook.com	2.96 billion monthly active users	Meta
WhatsApp	WhatsApp Web	2 billion monthly active users	Meta
Instagram	Instagram.com	2 billion monthly active users	Meta
YouTube	YouTube.com	2.6 billion monthly active users	Google
TikTok	TikTok.com	1 billion monthly active users	ByteDance
WeChat	WeChat for Windows/Mac	1.29 billion monthly active users	Tencent
Google Chrome	Google Chrome	3.2 billion users	Google
Gmail	Gmail.com	1.8 billion users	Google
Amazon	Amazon.com	300 million active customer accounts	Amazon
Netflix	Netflix.com	231 million paid memberships	Netflix
Spotify	Spotify for Desktop	489 million monthly active users	Spotify
Microsoft Office	Microsoft 365	345 million paid seats	Microsoft
Zoom	Zoom for Desktop	300 million daily meeting participants	Zoom
Twitter	Twitter.com	396.5 million users	Twitter, Inc.
LinkedIn	LinkedIn.com	900 million members	Microsoft

Consumer behavior in U.S. markets, in terms of minutes of daily usage, tend to confirm the concentration of app engagement, which is correlated with capacity demands to support that usage, especially since virtually all apps these days rely extensively on video content, which is the most-consumptive media format.  

App Name	Engagement Metric	App Owner
Facebook	33 minutes daily	Meta
TikTok	95 minutes daily	ByteDance
Instagram	29 minutes daily	Meta
WhatsApp	30 minutes daily	Meta
YouTube	45 minutes daily	Google
Twitter	31 minutes daily	X Corp.
Snapchat	28 minutes daily	Snap Inc.
Pinterest	14.2 minutes daily	Pinterest Inc.
LinkedIn	17 minutes monthly	Microsoft

In the U.S. market, almost by definition, video streaming apps are bandwidth-heavy, with social media following. Other usage, such as messaging and search, tend to require less bandwidth or capacity. 

source: financesonline, Hootsuite 

App or Service	Average Time Spent Per Day	Hyperscale Owner
Social Media
Facebook	30-45 minutes	Meta
YouTube	40-60 minutes	Alphabet
Instagram	30-45 minutes	Meta
TikTok	30-60 minutes
Search Engines
Google Search	10-20 minutes	Alphabet
Messaging Apps
WhatsApp	30-45 minutes	Meta
Messenger	20-30 minutes	Meta
Email
Gmail	15-25 minutes	Alphabet
Streaming Services
Netflix	60-90 minutes
Amazon Prime Video	45-60 minutes	Amazon
Disney+	30-45 minutes

One might note that internet apps and services are almost uniquely positioned to benefit from scale. As digital products, there are relatively few barriers (beyond government regulations) to global usage. Distance is not a major issue for product delivery and consumption.

Also, many internet-accessed apps benefit from network effects, where the value of the network grows with the numbers of users. If “most people” are using one particular social media app or messaging platform, that becomes a powerful reason to use it. 

The point is that we ought not be shocked by the disproportionate amount of global traffic and capacity needs generated by relatively few firms. That corresponds to Rule of Three dynamics in the businesses that generate the need for connectivity. 

Nor should we be surprised that app providers have themselves emerged in additional roles as leaders of the cloud computing, data center and transmission infrastructure functions. With scale, it becomes feasible to internalize formerly-external functions such as communications. 

And we tend to see similar dynamics with respect to chipsets, servers and software (ranging from operating systems to large language models). With scale, it is increasingly valuable or efficient to shift former externally-procured assets to in-house production. 

In the pre-internet era, global communications networks were primarily needed to support voice calls, so the suppliers were telcos who created and supplied the voice calling function. That has changed. 

These days, the dominant media type is internet data, supporting every consumer and business app and service. And since cloud computing is the architecture, every app is supported by remote processing and storage. And that, in turn, creates the need for high-capacity connectivity networks. 

The point is that different drivers for international connectivity should not come as a surprise. Internet-delivered apps and services have displaced voice calls as the driver of international connectivity. 

And the primary or significant drivers of consumer app use are relatively few. Also, low-latency, high-bandwidth, reliable connectivity now is a fundamental business requirement. And do-it-yourself now is highly feasible for volume users of chipsets, servers, rack space and connectivity. 

So should it come as a shock that a disproportionate share of international capacity is generated by just a few app and service suppliers, or that those users find they can afford to internalize connectivity functions? 

Hardly.