Slowing Device Sales are Both a Plus and a Minus for Mobile Operators

Slowing device sales are both a plus and a minus for mobile operators. On one hand, such sales also are associated with new account growth, and contribute to customer retention when the phones are financed over a period of years. 

On the other hand, devices are a significant cost driver and depress cash flow generation. Perhaps profit margins on handset sales are in the 10-percent range. That is offset by any subsidies the operator provides to customers in the form of installment payments. 

 

Pros	Cons
Can help to boost new account additions	Can be a major cost driver
Can make it more attractive for customers to sign up for a plan	Can lead to thin margins on device sales
Can help to lock in customers for longer periods of time	Can make it difficult for operators to compete on price

Say a mobile operator sells a $1,000 smartphone to a customer, but has to pay the device manufacturer $800 upfront. There is, in principle, a $200 profit margin on a device sale. But if that device is sold to a customer on an installment plan, and the installment payment period is three years, then  the $200 is earned over 36 months. This means that the monthly cash flow from device sales will be $5.55, offset by the carrier’s need to essentially carry the cost of inventory, in part, for up to three years. 

For example, AT&T's device installment payment inventory costs were $12.6 billion in 2022. Verizon's device installment payment inventory costs were $11.4 billion in 2022. T-Mobile's device installment payment inventory costs were $7.8 billion in 2022.

So device sales play a dual role in a mobile operator's business model. A subsidized phone is a significant value for many customers. On the other hand, device sales can also be a major cash flow issue for mobile operators.

Business Models for Generative AI?

As much as we presently struggle to define revenue models for generative AI or perhaps AI in a broader sense, past experience with major technology transitions provides a way of envisioning the possible adaptations. Consider whether generative AI is more like a "spreadsheet" or more akin to the internet.

The spreadsheet was invented in the 1970s and quickly became an essential tool for businesses of all sizes. It allowed businesses to store, organize, and analyze data in a way that was never before possible. This led to better decision-making, increased efficiency, and reduced costs, and also provided the rationale for buying personal computers. 

At a high level, generative AI might be the use case that spurs adoption of new or additional computing capabilities. 

Compare that to the internet, which eliminates distance and geography as key business constraints; virtualizes many formerly-physical processes and changes production costs (as for media, content, messaging, document exchange, voice communications, conferencing, retailing, advertising, teaching). 

Virtually all of us would consider the internet a bigger overall innovation than was the spreadsheet. 

The analogy one chooses will hint at degree of impact. 

Also, much of the direct new activity will be in “picks and shovels;” enabling infrastructure to support AI, such as computation, storage, advice and tech support, the creation of language models and generation of inferences will be direct revenue models. 

But most of the impact will be indirect, as was the impact of computing itself in the mainframe, minicomputer, personal computer, client-server, internet and mobile evolutions of computing. 

Infra suppliers (hardware, software and services) will create revenue enabling generative AI. But most entities will apply generative AI in their existing businesses, with mostly-indirect outcomes. Generative AI will be a means, not an end. 

Still, there might also be new lines of business created in the application and use case areas, as we have seen with cloud computing, e-commerce, search, social media and content streaming, all the result of the existence of the internet. As was the case for those innovations, we might not be able to predict their emergence or their revenue models. 

Generally speaking, for most industries the impact is likely to resemble the impact of spreadsheets (richer analysis, lower-cost customer service, faster modeling) more than the impact of internet disaggregation (products become services, death of distance). 

Pre-Internet Revenue Model	Internet Revenue Model
Hardware sales	Cloud computing
Software licenses	Software as a service (SaaS)
Maintenance and support contracts	Infrastructure as a service (IaaS)
Advertising	Technology supported by advertising (Google search, Facebook social media)
Subscription	Video streaming, content apps that displace newspapers, magazines, music streaming
Transaction	E-commerce
Affiliate marketing	Online travel agencies, price comparison websites
Leasing	Cloud computing
Concessions	Online gaming
Sponsorship	Esports

So most of the revenue models will fall within an existing range: usage fees, subscriptions, licenses, advertising, product sales, product leasing, advice, instruction, tech support and so forth. 

Technology	Revenue Model
Computing	Infrastructure gear and services sales, indirect support of advertising, commerce
Cloud computing	Infrastructure as a service (IaaS), platform as a service (PaaS), software as a service (SaaS)
Mobile access	Mobile data plans, mobile advertising
Broadband	Broadband internet access fees

"You Get to Keep Your Business" Will be the Fundamental Driver of 6G

Most 5G infra suppliers and mobile operators have been insistent that 5G would enable new use cases, novel applications and drive higher revenue, to some extent. So far, those proponents have been “wrong,” but only to the extent that they also were wrong about 3G and 4G. 

Though some important new use cases have emerged in each digital generation (from 2G on), most of the innovation has not been of the sort mobile operators can directly participate in as equity owners. 

In other words, most of the new value and revenue from new use cases has flowed to third-party app developers. And if you think about it, that is what is “supposed” to happen when a layered app architecture is assumed. 

By definition, the internet is “permissionless.” App creators do not require a formal business relationship with an internet access provider to reach users and customers. 

Eventually, some new 5G use cases will develop. But infra suppliers and mobile operators have routinely “over-promised and under-delivered” in the area of new apps, use cases and value, for every digital mobile generation.

An SKT white paper says 5F failed to achieve its goals, among which were the rapid development of new use cases, apps and services that collectively would fuel mobile operator revenue growth. There was no “killer service.”

SKT also essentially argues that 5G “over-promised and under-delivered.” Customers expected much more than what was delivered. 

As was the case for 4G, 6G will enable “services that were difficult to fully implement with 5G.” Anybody who followed 4G will get this. The promises of one mobile generation often are not realized--if at all--until the subsequent generation. 

In other words, some use cases hoped for in the 3G era did not develop until 4G. Perhaps some 4G use cases will flourish during 5G. Perhaps some 5G innovations will happen when 6G arrives. 

Maybe the industry is simply collectively wishful, without sufficient basis in fact. What a given network can do is not the same as assurance customers will value the innovations, or pay to use them. 

Quite to the contrary, the very architecture of internet-based apps and services militates against the ability of access providers to capture the value of app development. 

Perhaps a comparison with home broadband will illustrate why the “over-promising” always happens. Over time, home broadband has moved capacity upwards from kilobits per second to megabits to gigabits per second. As with mobile platforms, home broadband networks have used different media to support those advances. 

But nobody actually argues that “faster home broadband” will directly lead to new use cases and value supplied by the internet service provider. People understand that virtually all of the development will be fueled by third parties. The faster internet access only enables use of those innovations. 

Mobile operators might argue that they have a more-embedded role, as they offer managed services including voice and messaging. True, but some fixed network suppliers also offer voice, as well as internet access. 

The point is that a mobile service provider, in its role as an ISP, supplies “internet access” but not apps. And the primary value of 5G is that it supports more capacity than did 4G, as 4G enabled more capacity than 3G. 

Such capacity increases are essential. But ISPs are not primarily the producers of application value. 

To be sure, ISPs and their infra suppliers have to argue that wonderful new apps will be possible. Otherwise, it is hard to convince regulators to grant use of more spectrum. But everyone also understands that the new apps will mostly be produced by third parties. 

5G and 6G are vital, nonetheless. As with home broadband networks, capacity must continually be increased. 

But the hard truth is that 5G mostly means “you get to keep your business.” It is a means of supplying needed capacity, primarily. Someday, 6G will be required to enable mobile service providers to stay in business.

But the claimed benefits will extend quite a bit beyond that. They always do. 

Prosaic though it might be, the next-generation mobile networks are the functional equivalent of increasing home broadband and fixed network capacity from kilobits per second to megabits to gigabits. “More capacity” is the value. 

4G, 5G, 6G and beyond are the means by which mobile operators are able to supply faster speeds and more capacity over time. It means they get to stay in business. But it generally does not mean the mobile operators themselves will be creating new apps and use cases. 

So expect 6G to be yet another example of “failure.” Proponents will again over-promise. To get additional spectrum, they almost have to do so. 

But do not be fooled. They need more capacity. The way they will get it is partly by adopting 6G. It is important; they need to do so. But most of the hype about new value, apps and use cases--as produced by the mobile operators themselves--will fail. 

The architecture ensures it. The whole point of internet access is to enable people and machines to use apps available to internet-connected devices. We need more capacity, over time. New mobile networks are how we get there. 

But think of 5G and 6G as a necessary precondition for remaining in business, as faster fixed network access also is a fundamental requirement. Proponents will emphasize bells and whistles. Ignore all that. It is about remaining in business, as that business requires more capacity over time.

When Possible, Choose a High-Valuation-Ratio Business to Operate In

When creating a new category of products, allowing for some differences in perceived utility, value and valuation often can vary significantly between an older, legacy product and the new product that is a substitute. 

To be sure, a fast-growing category of product will have a higher valuation than a slow-growing or declining product. But it also makes sense, from the standpoint of attaining the highest-possible valuation, to position an asset as in the “new” category rather than the older category, when a valuation differential exists. 

Product	Industry	P/E	EV/EBITDA
Cloud computing	Software	25	20
Ownership of data devices and software	Hardware	15	10
Linear video	Media	10	5
Streaming video	Media	30	25
Traditional media	Media	5	2
Internet media	Media	20	15

Granted, it is hard to separate the value of “fast revenue growth” from other attributes driving financial value. Newer products often grow faster precisely because they do offer higher value or utility; have different and better cost structures or other value-creating features. 

The point is that key decisions made early in a new product’s positioning within a category can have an important valuation impact later on. This might be true when any single firm has revenue earned from multiple lines of business, for example.

Notably, a pure-play fiber-to-home network carries a distinctly-different valuation from a telco that has both copper and fiber access assets. Data centers tend to have the highest valuations, but, so far, edge computing facilities have a much-lower valuation. 

Asset Type	P/E	EV/EBITDA
Cell Towers	20x	15x
Data Centers	30x	20x
Subsea Fiber Networks	15x	10x
FTTH Networks	10x	5x
Telco and Mobile Service Provider Networks	5x	3x
Edge Computing Infrastructure	3x	2x

The practical import would seem to be that a unit of revenue earned by a firm with a high valuation is worth more than an equal unit of revenue earned by a firm with a  lower valuation ratio. In other words, Equinix earns a higher valuation on a unit of connectivity revenue than does a telco or mobile service provider offering the same product. 

When one has a choice, choose to operate in a segment of the digital infrastructure business that carries a higher valuation. Additional assets in other segments with lower valuations will then tend to be credited with the higher valuation.

Fixed Wireless Dominates U.S. Home Broadband Net Additions in 2Q

With the caveat that nobody knows how long the trend will hold, in the second quarter, home broadband account additions in the U.S. market were dominated by fixed wireless, according to the latest data from Leichtman Research Group. 

Of 841,000 net account additions, 893,000 accounts were added by fixed wireless providers. In other words, fixed network provider accounts actually declined, while fixed wireless grew. 

Skeptics always argue that, eventually, fiber connections will limit fixed wireless demand. Fixed wireless optimists tend to argue that enough capacity can continue to be added to sustain fixed wireless as a viable market offering for quite some time, and perhaps almost indefinitely in a percentage of markets. 

The business strategy would be to continue upgrading fixed wireless speeds, for example, to appeal to 20 percent of the market. In that scenario, the objective is not to match fiber-to-home speeds but only to support features most relevant for about 20 percent of the market that does not want to buy the fastest, or faster, tiers of service. 

In terms of geography, rural areas and out-of-region locations are likely to remain the places where fixed wireless makes most sense. In such geographies the cost to supply will be far lower than the cost of building new optical access networks. The leading exceptions might be markets where FTTH leased access is generally available. 

In the near term, new mid-band spectrum is likely to provide the needed capacity expansion. Long term, millimeter wave spectrum will be the key supplier of capacity growth. To be sure, small cell networks using low-band and mid-band spectrum will help, in some cases. 

Still, longer term, only millimeter and higher frequency spectrum will add enough capacity to allow fixed wireless offers to keep pace (again, preserving key appeal for about 20 percent of the market) with other fixed network alternatives. 

The big advantage of milliwave spectrum is capacity; the main drawback is coverage. That will pose a continuing issue for rural millimeter wave network deployments. The conventional thinking is that denser urban markets are where millimeter will continue to offer the most-interesting business cases: relatively high amounts of capacity in areas where distance is not a primary issue. 

 

source: ABI Research, RCR 

At least so far, fixed wireless has been, far and away, the clearest new use case for 5G.

AI Will Drive Data Center Capabilities: How Much Seems the Only Real Issue

Though artificial intelligence and generative AI training and inference operations are widely expected to drive data center requirements for processing power, storage and energy consumption, it seems perhaps unlikely that edge computing will get a similar boost, principally because AI inference operations and training are not latency-dependent. 

And the value of edge computing is latency reduction as well as bandwidth avoidance. And while it still makes economic sense to store frequently-requested content at the edge (content delivery use cases), AI operations will likely not be so routinized that this adds too much value. 

Operations requiring large language model access likely will still need to happen at larger data centers, for reasons of access to processing and storage resources. Think about processing to train AI models for self-driving cars, fraud detection, and other applications that require the analysis of massive datasets.

To be sure, support of self-driving cars also involves perhaps-stringent latency requirements. The problem is simply that the requirement for high-performance computing and access to data stores is more crucial for performance. So processing is likely to be located “onboard.” Again, the key observation is the split between on-device and remote data center functions. Edge might not play much of a role. 

The debate will likely be over the value of regional data centers, which some might consider “edge,” but others will say is a traditional large data center function. 

Operations that can be conducted on a device likewise will not benefit much, if at all, from edge compute capabilities. Think real-time language translation, facial recognition, and other applications that require quick responses.

And Digital Bridge CEO Marc Ganzi believes AI will mean a new or additional market about the size of the whole public cloud computing market, eventually. 

If public cloud now represents about 13 gigawatts of capacity, AI might eventually require 38 gigawatts, says Ganzi. 

The whole global data center installed base might represent something on the order of 700 gigawatts, according to IDC. Other estimates by the Uptime Institute suggest capacity is on the order of 180 GW. 

According to a report by Synergy Research Group, the global public cloud computing industry now represents 66.8 gigawatts (GW) of capacity. 

So AI should have a significant impact on both cloud computing capacity and data center requirements over time.