IP Was a Business Model Change, Not Just a New Networking Platform

Almost always, big changes in networking architecture and platform change the range of possible business models and market shares in the application, connectivity and infrastructure businesses. 

For example, the move to disaggregated, open and virtual networks automatically creates new potential roles for system integrators. Where platforms could be purchased monolithically, new networks can be assembled from various suppliers. 

To note only the most-obvious possible changes, monolithic platform suppliers could lose some market share to new suppliers and network integrators who supply the network components and the complete networks. 

In other words, when we move to disaggregate functions and elements, we automatically create a new need for system integration. 

So we must now look for the emergence of new names in the system integration business, as it applies to core networks and access networks. At some point, if they are willing or forced to concede some roles and revenues, the legacy monolithic network suppliers also are in line to act as system integrators, using elements and software sourced from any number of possible suppliers. 

Consider an earlier change that produced precisely those results. Because “layers” are the technology architecture, disaggregation is both possible and desirable. 

Look at data center or server businesses. Hyperscalers now build their own servers, they do not buy them. They can do so because layers make it possible. Custom software can run on commodity hardware; and commodity hardware can be built “in house.”

Hyperscalers build and own their own wide area networks, they do not have to buy services from WAN suppliers on their core data center to data center routes. 

Hyperscalers build their computing fabrics from modular arrays of servers, not monolithic mainframes. Whenever possible, they virtualize both compute and storage operations, rather than dedicating hardware to those functions. 

In a related process, “everything” is moving to virtualized supply. Enterprises and consumers can buy “services” rather than owning their own hardware and software licenses. Customers can buy computing or storage features “by the instance,” as a service, rather than building and operating their own data centers. 

It is an under-appreciated fact that when the global telecom industry selected internet protocol as its next generation platform, it also--knowingly or not--chose a layered business model. 

IP is not simply a framework for moving bits around; it is a business and revenue architecture as well, separating logical functions in ways that allow whole industry segments to emerge in a disaggregated way. 

In other words, the salience of the term “over the top” is precisely the result of a “layered” approach to building communication networks. When we disaggregate edge devices and functions from transport layer functions, and those from application functions, the revenue streams and possible business models also are disaggregated. 

That is why Meta, Amazon, Netflix and others can build businesses using networks without owning networks. 

IP was not just a technology platform change. It was a profound business model change.

"Open:" How We Got Here

Among the various conversations people had at the #PTC’22 conference are those about where networks are going, where the business is going and where revenue is to be found. Among the topics, 5G and Wi-Fi 6, network slicing, edge networks and the complications of in-home environments have been prominent. 

Among the potentially most far-ranging were questions posed by Robert Pepper, Meta head of global connectivity policy. Use of open technology is simply the latest in a series of transitions that have happened in the networking business over the last 40 years, Pepper said. 

“Disaggregated network elements are 40 years in the making,” Pepper said. The industry transitioned from analog to digital; then hardware to software functions, he noted.

The “next transition is from integrated and proprietary to open and modular networks,” he said. 

There will be big repercussions for suppliers of networking infrastructure. Where telcos 50 years ago developed and made their own gear, they then switched to buying complete networks from a handful of global suppliers. That obviously created huge new businesses, but also made telcos “captive” to a few suppliers and “vendor lock in.”

Suppliers might like that state of affairs, but buyers (telcos) hate it, it is fair to say. In a broad sense, the shift to open and modular networks also represents a shift from vendor-led to operator-led infrastructure development and supply. 

It also is fair to note that there always are private interests that benefit from any wider shift in framework. Perceived benefit hinges on where a firm or industry segment operates in the complete value chain. 

Application supplier business models depend on ubiquitous, high-quality and low-cost  internet access. Access providers are not similarly situated within the value chain. For app providers, high-quality, low-cost internet access is a prerequisite for business. For connectivity providers, access is the business. 

For an app provider, internet access is a cost of doing business. For a connectivity provider access is the core revenue stream. What the former wants is lowest-possible cost and highest-possible quality, the latter wants highest-possible revenue with minimum-possible cost. 

You might argue it is in Meta’s interest for internet access to be universal and good, as it is in a connectivity provider’s interest to reap the highest revenue from access services, with the highest margins consistent with long-term sustainability. 

If Meta is right, economics are moving in the direction of what is favorable for application creators. 

There are clear analogies in the data center or server businesses as well. Hyperscalers build their own servers, they do not buy them.  Hyperscalers build and own their own wide area networks, they do not buy services from WAN suppliers on their core data center to data center routes. 

Hyperscalers build their computing fabrics from modular arrays of servers, not monolithic mainframes. Whenever possible, they virtualize both compute and storage operations, rather than dedicating hardware to those functions. 

Moderator Gary Kim, a PTC volunteer and consultant, noted that when the global telecom industry selected internet protocol as its next generation platform, it also--knowingly or not--chose a layered business model. 

IP is not simply a framework for moving bits around; it is a business and revenue architecture as well, separating logical functions in ways that allow whole industry segments to emerge in a disaggregated way.

In other words, the salience of the term “over the top” is precisely the result of a “layered” approach to building communication networks. When we disaggregate edge devices and functions from transport layer functions, and those from application functions, the revenue streams and possible business models also are disaggregated. 

That is why Meta, Amazon, Netflix and others can build businesses using networks without owning networks. 

IP was not just a technology platform change. It was a profound business model change.

What if "Better Broadband" Actually does not "Cause" Economic Growth?

The prevailing wisdom that super-high-quality home broadband actually changes things is wildly and uncritically accepted as “truth.” That is not to deny that ubiquitous access to higher-quality broadband is to be preferred. Homes who can only get 25 Mbps will not have the same experience as households able to use 100 Mbps to 200 Mbps, when there are multiple users and multiple devices in simultaneous use. 

The issue there is bandwidth per user and device, in real time, with simultaneous use of various applications. Multiple users almost always benefit from “more bandwidth,” as is the case for every shared communications medium. 

But as a matter of science, it is impossible to actually quantify the outcomes from upgrading access--ubiquitously--from some lower level to some higher level. 

For most households, businesses and communities, almost nothing would change simply because bandwidth was increased from a moderate level (100 Mbps to 200 Mbps) to a gigabit per second, for example. 

More precisely, for any single user, trying to use any mix of applications, more bandwidth might help with experience, or might not. In other words, the benefit of “more bandwidth” depends on how many users in a home, how many online simultaneously, what they are trying to do, how many devices they are using and what the applications “need” in terms of performance. 

Downstream is one thing; upstream another thing. 

And, to be sure, our requirements drift upwards over time. That will not stop. But outcomes hinge on many things other than per-user bandwidth. We cannot actually say that student performance on homework is X percent better with Y increase in per-user bandwidth. Maybe it is; maybe it isn’t. 

And it is hard to see a true causal relationship between region economic growth and job growth, for example, as bandwidth is increased from X to Y. Regions that are growing slowly will still grow slowly, even with better broadband, because growth hinges on other matters, such as proximity to large markets. 

Regions losing population; facing industrial shutdowns or other changes in underlying conditions do not materially change simply because “better broadband” is available. 

Tourism, manufacturing or service businesses do not often relocate to distant or isolated regions because better broadband is available. There are other important reasons why a place is deemed fruitful for additional job growth or facilities. 

Better broadband does not causally change educational or industrial or professional skills possessed by the local population. You might argue that permanent work from home will change living locations. 

But most of those changes will still be tethered to population centers in key ways, simply because humans value the amenities that population density provides. So exurban changes will happen more than shifts of sizable numbers to very-remote areas. Shifts from urban to suburban likewise will be materially more important than shifts to very-rural areas. 

Yes, anecdotally, more people will spend more time in mobile modes when working away from the office. But the actual level of home broadband access quality--beyond a baseline level-- will arguably be a secondary consideration, at best, in most cases. 

Places that can be upgraded from less than 25 Mbps up to 100 Mbps are likely to be places not so desirable for workers for other reasons. Upgrades from 100 Mbps to some higher number likewise might be helpful and preferred, but not a driver of detectable performance and outcomes. 

One might argue that personal productivity now is higher with gigabit access than with dial-up. But that is a correlation. The applications I could use in the dial-up era were the real limitation, not the bandwidth. 

Today’s applications are so much richer that I cannot separate bandwidth from application richness. To the extent I might claim to be more productive, it is mostly because my applications and devices allow me to do more, in less time, not that my bandwidth--per se--allows me to do so. 

Better bandwidth is--virtually all of us agree-a good thing. But its ability to change outcomes--economic; job creation; educational outcomes--generally is overstated and hard to prove. Parental support for and involvement in their children’s education counts for much more. 

The general economic growth profile of a region matters more than the speed of broadband. The presence of large pools of workers with the right skills matters more than broadband. The quality of life of a region for such workers also arguably matters more than broadband.

In fact, we cannot disprove the thesis that highly-educated residents; fast-growing regions and industries; high incomes and high wealth “cause” better broadband, not the reverse. 

U.K. ISPs More than Double Gigabit Internet Access Availability in a Year

Big internet service providers are used to slings and arrows shot at them. Sometimes the criticisms of their performance seem undeserved. Consider U.K. gigabit-per-second internet access availability. 

U.K. ISPs already had said they would cover between 70 percent and 80 percent of households  of the country with gigabit-capable infrastructure by 2025, without government assistance. 

In September 2020 about 27 percent of homes could buy it. A mere year later, 46 percent of homes could buy gigabit services, according to Ofcom. 

Ofcom estimated availability had reached closer to 60 percent by the end of 2021, largely as a result of Virgin Media’s upgrade to DOCSIS 3.1, Total Telecom says. 

That is a dramatic change for a single year’s work. 

Yes, some households will be harder to upgrade, and yes, there will be additional government support to do so. 

To help deliver on these targets, in 2019, the government pledged £5 billion in public funding to help connect the most difficult-to-reach 20 percent of households, Total Telecom says. Most will be allocated no earlier than 2026, however. 

The point is that we would probably all find it hard to point to a single year when progress that dramatic was made. Basically, the major ISPs more than doubled the availability of gigabit per second service in a year’s time.

Not Every Acquisition Works Out

Not every acquisition works. Not every asset disposition is driven mostly by profit taking. Sometimes loss limitation is at work. And though many institutional investors or private equity firms have one business model for telecom infrastructure assets, service providers often have a different model. 

That difference in models explains why many institutional and private equity firms now are buyers of assets while many service providers are asset sellers. WindTre might be next. Lumen and Telefonica are among recent sellers. So was Cincinnati Bell.

Telecom Italia could move as well. 

Because we can” or “because we should” might explain a good deal of asset disposition behavior in the connectivity business these days. 

Optus owner Singtel, for example, is said to be mulling the sale of a stake in its Australian access facilities, a move that would allow Singtel to raise cash. 

Such opportunistic moves--as always--are driven by a combination of seller need, buyer interest and a broader rise in the value of optical fiber access and transport assets for investors in search of alternative assetshttps. 

Low interest rates mean lots of capital is available, while high valuations for other traditional assets also are driving investor interest in lower-valuation, higher-return financial vehicles and something more akin to a private equity approach to investing by institutional investors such as pension funds. 

Buyer interest has grown the value of optical fiber assets or the ability to create them,  while sellers are enticed by such higher valuations to monetize access network assets as they earlier monetized cell tower assets. Singtel itself sold a majority stake in its Australia cell towers in 2021. 

No doubt owner's economics still are important. But the issue is whether full ownership is required to reap that value. In a growing number of cases, partial ownership seems to be viewed favorably.  

In other areas, co-investment deals are changing the economics of optical fiber investment. 

For a number of reasons, the business model for telco and cable TV fiber to home is changing. A higher degree of government subsidy support; a desire for investment in FTTH facilities as alternative investment and competitive dynamics in the home broadband industry all mean the business case for FTTH improves. 

As one example,Cable One is part of a joint venture with GTCR LLC,  Stephens Capital Partners, The Pritzker Organization and certain members of the management team to build optical fiber to premises networks by Clearwave Fiber.

Clearwave Fiber holds the assets of Cable One’s subsidiary Clearwave Communications and certain fiber assets of Cable One’s subsidiary Hargray Communications. 

At the same time as capital investment requirements are changing, there is a shift in the assumptions about business model. 

In the late 1990s FTTH was seen as the only viable way for telcos to take market share in the linear video subscription business from cable TV operators. So the revenue upside was subscription video and internet access speeds. To be sure, video arguably was seen as the bigger revenue driver, as late 1990s telco FTTH speeds were in the 10 Mbps range. 

Bundling (triple play or dual-play) also was seen at that time as the way to compensate for competition-induced account losses. While telcos or cable each competing across the voice, business customer, internet access and video entertainment markets might have fewer total accounts, revenue per account from triple-play services would compensate. 

But something else now seems to have changed. A decade ago, independent internet service providers began to attack the market increasingly based on one service: home broadband. To be sure, many independent ISPs tried a dual-play or triple-play approach for a time. 

But nearly all eventually settled on a home broadband-only approach. Since virtually all independent ISPs face both telco and cable TV competitors, the single-product business model makes some concessions on potential revenue that necessarily must be balanced by lower capital investment and operating costs. 

The latest developments are that such tradeoffs are seen as feasible even for incumbent telcos: in other words, the business model increasingly relies on broadband as the foundation, with some contributions from voice. Video (linear or streaming) plays a lesser or no role in revenue assumptions. 

There are other changes. Subsidies have been rising for broadband deployment, and that also changes the capex requirements. Some of the investment in optical fiber also is helped by the denser optical fiber networks necessary to support 5G networks. Essentially, the payback model is bolstered by the ability to defray some optical media costs from mobile service revenue opportunities. 

Also, 5G supports home broadband using the same transmission facilities as does mobile service, often offering a chance for mobile operators to compete in the home broadband business at relatively low incremental cost. That also helps lower the cost of fixed network FTTH as more revenue is wrung from the installed assets. To the extent that higher revenue produces incrementally higher free cash flow, more capital is available to invest in additional FTTH facilities.

The incremental cost of consumer home broadband is lower once a dense trunking network must be put into place to support small cell mobile networks. 

Also, the value of FTTH facilities has changed as rival investors (institutional investors, private equity) view consumer broadband as a legitimate alternative investment. That boosts the equity value of an FTTH network and supplies new sources of investment. 

Also, the cost of FTTH construction has improved steadily over the past few decades. Also, the expected reduction of operating costs from fiber networks, as opposed to copper networks, now is well attested. So there are opex savings. 

FTTH remains a challenging investment, nonetheless. But it is noteworthy that assumptions about the business model now have changed for incumbent and new providers as well. Where it once was thought an FTTH upgrade virtually required revenue from three services, in an increasing number of cases the investment can be justified based on home broadband alone. 

In greater numbers of cases, the primary value of home broadband is supplemented by some revenues from other sources. But where a triple-play might have produced $130 per month to $200 per month revenues, home broadband might produce $50 to $80 a month. 

That projects increasingly are feasible with a $50 monthly revenue target and adoption around 40 percent to 50 percent shows how much the capex and opex assumptions have changed.

ISP Bandwidth Planning has been Remarkably Effective and Efficient

Something we learned during the Covid pandemic was that the way internet service providers engineer their networks--adding capacity in advance of demand--does work to handle unexpected demand spikes. They have been effective at building networks that can withstand even an unexpected and sudden change in the demand curve.  

On the other hand, it always also makes good business sense to invest in additional capacity only with respect to anticipated demand increases, whatever rate you believe reflects actual demand growth. AS it turns out, ISPs and their suppliers also have been good at "efficiency" in supplying new capacity.

This forecast by Point Topic illustrates the concept. Given expected demand growth, capacity growth is planned at a rate that stays ahead of demand, but not too far ahead. 

In other words, investment  is matched to revenue. The trick always is that customer segments exist. Some customers have higher demand than others. The geographic locations of those customer segments also is mixed. Business locations are mixed in with consumer locations. Higher-demand home worker locations are mixed in with lower-demand “average consumer” locations. 

source: Point Topic 

In other words, the whole network embeds assumptions about the minimum performance that must exist to handle the peak load by the heaviest users. At the same time, it makes sense not to “over-engineer” the network, adding cost that has no corresponding revenue upside. 

So much hinges on how fast any firm believes typical demand will increase. Is itr 50 percent per year; 40 percent per year or some lower figure? Those assumptions might also fail to account for improvements in networking infrastructure efficiency or the emergence of new bandwidth-intensive applications that change demand expectations.

Did a Covid Emergency Program Work? We Don't Really Know

It often is difficult to determine whether any specific government or private program to “fix a problem” actually worked. An emergency program  for broadband service might provide a case in point. 

The Emergency Broadband Benefit (“EBB”) Program, established by the Consolidated Appropriations Act of 2021 had nearly nine million participants by the end of 2021. The stated purpose was to keep low-income households connected at a time when Covid restrictions made it hard for people to go to work. 

So the U.S. Congress created a program providing up to a $50 monthly subsidy (more in tribal areas) for Internet connections, in addition to existing programs. 

The issue is how to interpret program success. The stated objective was to “keep people connected. 

The problem is that most of the people using the temporary program also were using the existing programs. So it is akin to trying to  “prove a negative” (proving something to be true--with certainty--in the absence of evidence).

Households on support programs did not disconnect. What we do not know is whether they would have disconnected in the absence of the emergency program. 

“My analysis suggests that in November 2021, Lifeline subscribers (households receiving discounted service) accounted for about 80 percent of EBB participation,” says George Ford, Phoenix Center chief economist. “With broadband adoption by low-income Americans being about 75 percent, it could be that only about five percent of EBB participants were not previously online.”

What we might be able to say is that the “EBB Program did not appear to be increasing broadband adoption by much, though it may be argued that was not the point,” says Ford. “The point of the EBB Program was not necessarily to expand adoption but to maintain it during the pandemic’s economic malaise, so perhaps this finding is untroubling.”

Still, we do not know what might have happened if the EBB did not exist.