Saturday, August 24, 2019

Ironically, Multi-Purpose Networks Now Hinge on Single-Purpose Revenue Models

Even if the great advantage of a network running internet protocol is that it is inherently an “any application” network, end user demand changes and the separation of app from access both mean that many fixed networks are on the verge of becoming single-purpose businesses in terms of revenue drivers. 

It is an unintended consequence of the move to IP, with serious implications for capex and opex strategies and business models.  

Fully 78 percent of U.S. consumers surveyed by IHS Markit think they will stream video on their 5G devices, for example, making video the killer app for 5G, according to IHS Markit. 

When asked to name activities they are likely to increase on 5G devices, consumers ranked video streaming first, ahead of video calling, social media, mobile gaming, virtual reality and augmented reality, IHS Markit says. 

You might notice something about that list of activities people believe they will engage in more, on 5G devices. Voice and text messaging are not listed. 


Keep in mind also that the global standards for 5G actually do not include voice support. Voice is simply assumed to be an IP-based app that runs on the network.

All of that tells you something about changes in the products people pay for when using mobile or fixed communications these days, and the way connectivity providers must structure their businesses.

Recall that a few decades ago, the big advantage of an internet protocol network was that it is, by design, media type agnostic. The old phrase “bits are bits” captures the flexibility of an all-IP network, able to carry any media type: voice, text messaging, video, voice, images or music. 

For those of you with long memories, recall that until the modern era, all networks were media-specific, single-purpose networks. Broadcast TV, broadcast radio, cable TV networks, telephone networks, telegraph networks, paging networks, cellular mobile networks, most satellite networks and microwave networks were created and optimized to handle one media type. 

IP changes all that. Add competition, product substitution, Moore’s Law, open source, virtualization and the economics of the networks business--revenue streams and business models--changes drastically.

The advantage of an all-IP network is that it can deliver, and in principle therefore make money from, providing a range of high-demand services (voice, internet access, television). 

On the other hand, legacy revenue streams are diminishing, reducing the value and magnitude of legacy services of many types, even while unit growth continues in some markets. 

The important implication is that many fixed networks increasingly are driven by a key revenue source, internet access or enterprise data access, the former for consumers, the latter for enterprise services. In other cases a key source of value for the fixed network is small cell backhaul. 

Many smaller internet service providers now build their revenue models almost exclusively on internet access. Cable TV operators now point out that internet access drives their revenue growth for consumers and businesses. 

Telcos--both mobile and fixed--are losing voice and messaging revenues that once drove the business. Video is a source of growth for some telcos, even as cable operators lose market share. 


Geostationary satellite constellations have relied on entertainment video delivery as the revenue mainstay. The proposed new low earth orbit constellations will aim to provide internet access. 

But you see the trend: many networks, though capable of supporting multiple services, are tending to find that growth comes from a single key service. Ironically, networks that can, in principle, deliver any media type are finding that revenue generation actually is coming from just one media type.

No comments:

Costs of Creating Machine Learning Models is Up Sharply

With the caveat that we must be careful about making linear extrapolations into the future, training costs of state-of-the-art AI models hav...