A Handful of Hyperscale App Providers Drive Global Bandwidth

There is a simple reason why hyperscale app providers now drive global data traffic and build and own their own networks: a handful of firms generate most of the demand. In 2021 just six firms generated 57 percent of global traffic. 

source: Sandvine, IN Forum

These days, voice demand is paltry in relation to content bandwidth--largely video--that flows between hyperscale application provider data centers and internet points of presence where local internet service provider traffic pours onto the backbones. 

In other words, if you know the locations of the hyperscale app provider data centers, you also know the locations between which most data flows over WANs. 

source: Telegeography 

source: Telegeography 

Global Internet Bandwidth Grows 29% in 2021

Global internet bandwidth usage spiked 34 percent from 2019 to 2020 and a further 29 percent in 2021 to 786 Terabits per Second (Tbps), says Sandvine. 

“We expect the average household to use as much as 650  gigabytes to 750 gigabytes per month by the end of 2021,” Sandvine notes. Video and game content is fueling power user behavior. 

“In fixed network data, we are seeing an increase in 1 terabyte per month power users,” Sandvine says. “The heaviest users spent time on XBOX Live Video, Microsoft Outlook 365, Netflix, PS4 Games downloads, Discord, Twitch, and BitTorrent.”

And the “big six” share of traffic is growing.

source: Sandvine  

What You Do When it Was -7 Degrees Fahrenheit Last Night

This is one of the cats who regularly visits my backyard. I'm not sure whether it is a stray or abandoned cat, but am certain he is not feral (he has interacted with humans in the past; feral cats grew up "wild" with no close human interaction). He could be owned by a neighbor, but allowed to roam outdoors at night, or when it is quite cold, something I would not do. 

I always keep water and food out there, just in case, plus a cat shelter. I've done so for decades, everywhere I've lived, just in case. He's huddling up in the early morning sunshine trying to get warm. 

 

Analysts Signal Belief that Home Broadband is a Material Revenue Driver for T-Mobile

There were lots of questions from equity analysts about fixed wireless on T-Mobile's fourth quarter 2021 earnings call. That is evidence that home broadband is expected by analysts to become a material contributor to T-Mobile's revenue and earnings. 

Perhaps most surprising is the revelation that “the majority of our customers come from suburban and urban markets,” according to Dow Draper, T-Mobile EVP. T-Mobile--and most observers--might have guessed adoption would be driven by rural account additions.

Will FTTH Payback Always be Led by Internet Access Revenues?

Retail mobile and fixed network connectivity providers who sell directly to consumers arguably face some issues related to average revenue per account and cost per account. Whether in the mobile or fixed realms, mobile revenue per account seems to range from a few dollars a month up to $41 per month. 

source: S&P Global Market Intelligence 

Against that must be balanced the cost of infrastructure, operating and marketing costs plus all other overhead, ranging from personnel benefits to debt service and taxes. While not minor, network infrastructure costs are only part of the cost model. 

Some have claimed 5G can reach break even in a five years or less. But that likely rests on excluding all other business costs except the network infrastructure. 5G capex per subscriber might range between $100 and $450 per year, during the network build period. 

Even assuming a 20-percent profit margin, that still means 80 percent of revenue is consumed by operating costs, marketing, amortization of debt and other overhead, including personnel costs, retirement fund payments, dividend payments, taxes and so forth. 

Looking at internet access prices using the purchasing power parity method, developed nation prices are around $35 to $40 a month. In absolute terms, developed nation prices are less than $30 a month. 

That PPP normalization technique compares prices to gross national income per person. There are methodological issues when doing so, one can argue. 

Gross national income is not household income, and per-capita measures might not always be the best way to compare prices, income or other metrics. But at a high level, measuring prices as a percentage of income provides some relative measure of affordability. 

Generally speaking, broadband prices are dropping in developing countries, where the product is most expensive, and primarily because mobile internet access prices are dropping. 

source: ITU 

Looking at mobile voice and data prices, as a percentage of gross national income per person, one easily can see that very-high prices in lesser-developed countries skew global indices. In some developed markets, prices are less than one percent of GNI (without adjusting for purchasing power parity). 

source: ITU 

The unadjusted 2019 average price of a broadband internet access connection--globally--was $72..92, down $0.12 from 2017 levels, according to comparison site Cable. Other comparisons say the average global price for a fixed connection is $67 a month. 

Looking at 95 countries globally with internet access speeds of at least 60 Mbps, U.S. prices were $62.74 a month, with the highest price being $100.42 in the United Arab Emirates and the lowest price being $4.88 in the Ukraine. 

Another study by Deutsche Bank, looking at cities in a number of countries, with a modest 8 Mbps rate, found prices ranging between $50 to $52 a month. 

The point is that network infrastructure investment now seems to hinge on revenue--depending on how we count it--that could range from a few dollars a month up to perhaps $72 a month. 

Most of the market--with prices adjusted for currency and living standards--seems to be $40 a month or less. 

That is challenging for fixed network operators deploying fiber to the home, if less a challenge for mobile operators, whose networks cost less, per customer or passing. 

Among countries that are members of the Organization for Economic Cooperation and development, prices prices in 2016 seemed to cluster around $40 per month. 

A more recent study confirmed those figures. 

The take away is that FTTH payback--in many markets--cannot rest solely on home broadband revenues. Other revenue drivers likely must also contribute. Right now, that includes backhaul for 5G and future mobile networks, internet of things, edge computing, internet of things and applications that are owned by the connectivity providers. 

Over time, those other sources might be even more important. Payback models of several decades ago assumed significant contributions from sources such as voice and video entertainment that have declined steadily. 

The net change is a revenue-per-account ranging from $130 per month to $200 a month to the present $40 to $50 a month level. FTTH costs per passing have gotten better, but probably not enough to support revenue as low as $40 a month or even $50 a month. 

Other value must be involved, or the payback model is not there. 

Are Mobile-Plus-Fixed Bundles on the Verge of Marketing Push?

If a projection by the European Telecom Network Operators’ Association proves to be accurate, and if it is mirrored in some other markets, a growing portion of the consumer market will be amenable to buying both mobile and fixed network services as a bundle. 

Basically, a growing percentage of European customers are doing so. 

source: ETNO

How Long is the FTTH Payback Cycle?

As fixed network revenue sources increasingly rely on internet access as the foundation service, with declining take rates for voice or entertainment video services owned by the telcos, the payback model for fiber-to-home upgrades gets more challenging, above and beyond price declines.  

Without adjusting for currency differences and costs of living, internet access costs between $60 and $70 a month, some studies find. 

source: Analysys Mason 

Think about the payback implications. In the U.S. market, for example, the cost of passing a home location using FTTH might be in the $600 range. The additional cost to connect a customer might be in the $700 range. 

At 50-percent take rates (high, by telco standards), that means half the assets are stranded. The network infrastructure cost for one paying customer is $1200 plus $700, or about $1900. The reason is that the payback for one customer means passing two locations. 

Annual revenue for one FTTH home broadband location, at perhaps $600, means a rather-lengthy payback, as amortized overhead, debt service and principal repayment, inflation, operating costs and marketing and retention costs must also be included. 

Even assuming a 20-percent profit margin, that still means 80 percent of revenue is consumed by operating costs, marketing, amortization of debt and other overhead, including personnel costs, retirement fund payments, dividend payments, taxes and so forth. 

So, roughly speaking, assume that 80 percent of $600 in annual revenue is needed to cover mostly-fixed costs. That leaves about $120 annually in free cash flow. That implies a long payback cycle.

In Honor of Freedom Fighters Everywhere

Did Not See This Coming

This I did not see coming.  AT&T will spin off AT&T’s interest in WarnerMedia  to AT&T shareholders. I had assumed AT&T would retain its 71 percent interest in WarnerMedia. 

In doing so, AT&T will shrink--in terms of equity value--to the fourth-biggest connectivity firm, behind Verizon, Comcast and T-Mobile. 

I was wrong about AT&T’s “exit” from content. The characterization of AT&T’s merging of WarnerMedia assets with Discovery has been called a strategy shift that gets AT&T out of the content business. I had been characterizing it as AT&T monetizing part of the asset, since it still owned 71 percent. 

But by spinning out the interest in a tax-free transaction--along with the shifting of DirecTV to a private equity joint venture--AT&T really is getting out of the content business. 

Should AT&T also decide to spin out its interest in DirecTV, it will shrink even further, in terms of equity value. 

It is at least the second big strategic shift we have seen AT&T make over the past three decades. Each time, there was a retreat, but arguably because the debt load associated with the strategic moves was burdensome. 

Consider AT&T’s big move into cable TV in the mid-1990s, a time when the long distance provider was seeking a way to reenter the local access business with its own facilities. The thinking at that time was that a largely one-way cable TV plant could be upgraded to become full communications facilities, supporting home broadband and voice. 

Given that development by virtually all cable TV companies in North America and Europe, the thinking was sound. 

AT&T also made its first investment in  DirecTV in 1996, owned and spun off Liberty Media. 

Beside TCI, at that point the largest U.S. cable company, AT&T also bought  Teleport Communications Group, a $500-million-a-year local business phone company, for $13.3 billion; MetroNet, a Canadian phone system, for $7 billion; and the IBM Global Network, which carries data traffic, for $5 billion. He also signed a joint venture with Time Warner ( to carry phone calls over the entertainment conglomerate's cable TV systems, and with British Telecom to serve multinationals overseas. 

But the debt burden was too high and AT&T reversed course in 2004 and sold most of those assets. AT&T Broadband (the former TCI and US West Broadband assets) were sold to Comcast, making that firm the biggest U.S. cable TV company. 

By 2005 AT&T itself was acquired by SBC Communications, which promptly rebranded itself AT&T. 

AT&T's move into content with the full acquisition of DirecTV and Time Warner content assets was the second big diversification move AT&T has attempted since the mid-1990s. 

AT&T spent about $170 billion since 2015, including taking on new debt, to transform itself into a media conglomerate. 

By spinning out WarnerMedia, AT&T will shrink in equity value to about $130 billion. That is shocking in some ways. Verizon is valued at about  $232 billion, Comcast at $262 billion and T-Mobile at $175 billion.

Beyond all that, one wonders what big tier-one carriers are going to do to keep revenue growing, given the pressures on their core businesses, and especially if they decide to retrench on their core businesses. 

Growth rates are low and average revenue per user or account is flat to dropping. 

source: Statista 

Especially important is mobile data ARPU, as mobility drives global revenues, while mobile internet access drives revenue growth. 

source: Strategy Analytics

The declining ARPU poses revenue growth constraints in any market where subscriber penetration is close to saturation (every customer who wants to have a mobile account already has one). 

 

source: Telefonica 

And falling ARPU is a trend in both mobile and fixed network domains, though some hope the declines in mobile ARPU can be arrested. 

source: Researchgate 

Still, some think it is possible that revenue in the mobile industry could have peaked in 2021. That might be unduly pessimistic. Still, many expect flattish global revenue, going forward. 

source: Statista

The strategy implications of another AT&T retreat from diversification are basically the fundamental problems the industry has been dealing with for some time. Where can growth be found, especially if one assumes service providers will largely stick to connectivity services?

Perhaps the better way to characterize the issue is to ask “where can growth be found within the connectivity realm at a level high enough to stay ahead of inflation?” Perhaps we ought to simply acknowledge that the public communications business remains a slow-growth industry that is challenged by disruptors of many sorts. 

Perhaps “growth” needs to be framed in a modest way: enough revenue growth to replace lost revenues from declining product segments while staying ahead of inflation.

Virgin O2 Reportedly Seeking Co-Investment for FTTH Push

In yet one more sign of changing fixed network infrastructure costs, Virgin O2 is reportedly in talks with infrastructure funds to create a facilities-based U.K. alternative to Openreach.  The goal is to create new fiber-to-home coverage of seven million homes. 

Virgin O2 already has said it would upgrade all current 15.5 million gigabit connections to FTTH by 2028. 

Though the big story is the creation of a nationwide facilities alternative to Openreach. The new network would have Virgin O2 as an anchor tenant, but the network would also offer wholesale access to third parties. 

To be sure, since the U.K. market essentially has had two tier-one firms competing in the fixed access market--cable TV and telco--the shift is not new in principle, but new in coverage and ubiquity. 

The proposed additional upgrades of the full network, plus addition of seven million new locations, would create a second nationwide FTTH provider with wholesale access. 

Secondarily, the move ends the historic cable reliance on hybrid fiber coax--in the United Kingdom--forever. 

Also, the financing and business model for FTTH also changes. Where cable companies historically have financed all of their own access network, the new move creates a new entity to own the access infrastructure.

Setting up the new entity changes the payback model for FTTH deployment, even as it shares the revenue and profit upside with new investors. 

But infrastructure has taken on greater importance in private equity and institutional investor portfolios in recent years. So the desire by Virgin O2 to invest is matched with matching desire by investors to fund and own such infrastructure. 

All of our decades-long assumptions about FTTH payback models are thereby upended. In principle, co-investment is one solution for revenue assumptions that have drifted downwards from perhaps $130 a month to $170 per month in revenue to a more-dependable $50 per month to $70 per month revenue per household. 

Revised payback models therefore must be revised accordingly.

Blockchain Value in the Connectivity Business Could Center on Further Disintermediation

Blockchain is believed by many to have application in the connectivity business, such as creating mechanisms to verify buyer identities, seller assets and liquify the process of settlements, perhaps especially across national borders. 

Some believe blockchain can ease the chores of number portability as well, since verifying identities is made easier. Others believe blockchain can reduce fraud and waste, for such reasons. Blockchain is viewed as a way to prevent vaccine fraud, for example, another similar use case in the health industry. 

Verifying actual performance might be quite valuable for eliminating or limiting disputes over service level agreement performance. “One version of the truth” should reduce instances of uncertainty about actual performance. 

Blockchain could help reduce phone theft, by making stolen devices unusable because the lawful ownership history is clear. 

Possible uses of blockchain in other industries might eventually suggest additional connectivity business uses. 

Blockchain now is seen by some as a way to disrupt and decentralize movie financing, for example. 

And while blockchain might not have much incremental value for tier-one connectivity providers, who have means to acquire capital, blockchain could well be important for smaller, upstart providers as a means of raising capital. 

Strategically, blockchain also is seen as a new form of disintermediation beyond the use of internet mechanisms to displace distributors. 

If you think back, the creation of huge e-marketplaces displaced distributors of all sorts in virtually all value chains. Blockchain could take that process a step further. 

Think of a potential global ability to buy and sell assets--connectivity (access or transport), compute cycles, interconnection, application use, storage or radio use--by means of an online portal that is  blockchain-enabled. 

That would be a sort of ultimate fulfillment of the 50-year drive for on-demand provisioning that the connectivity industry has sought.

Why Fixed Wireless Will be a Bigger Use Case for 6G

Even if support for ever-faster mobile internet is the objective of each mobile next-generation network--including 5G and succeeding platforms--there is reason to believe that fixed wireless might be a growing use case. 

The reason has to do with radio frequency signal propagation in the millimeter regions that will become necessary. 

“Wireless channels in this frequency range experience large propagation and reflection loss, sporadic availability of line-of-sight links due to blockage, and molecular absorption,” notes MediaTek. “These phenomena result in a link performance with shorter range and an intermittent on/off behavior.”

Millimeter wave radio signals will use beamforming to overcome signal loss. But that also means near line-of-sight performance. 

For that reason, sub-THz spectrum will be suited for fixed backhaul or nomadic devices communicating with a hub or as part of a network mesh, says MediaTek. The mesh capability is important as mobile users might move often among radio sites, requiring ability to shift between different radio sites on a dynamic basis. 

In principle, this mesh approach means a mobile user might be serially (or simultaneously) connecting to different mostly line-of-sight radios. Though the physical paths might be fixed, the mobile user will take advantage of any number of fixed paths to sustain a mobile connection to the “capacity” millimeter wave spectrum, while defaulting to coverage spectrum when necessary. 

Faster speeds are coming. The issue is how the radio access network is re-architected to take advantage of frequencies that largely are line of sight. 

MediaTek believes 6G will bring speeds 10 times to 100 times faster than 5G. That is not an unusual prediction. Indeed, every mobile digital generation has increased bandwidth by 10 times to 100 times, and reduced latency about 10 times, each generation.

source: MediaTek

Devices might also have the ability to sidelink, much as current devices can use Bluetooth for short-range device-to-device communications. 

 source: MediaTek

Other changes are possible. “A distributed MIMO deployment, where Tx-Rx signal pairs are not just bound to one node/site but distributed across multiple sites and nodes, has the ability to improve spectral efficiency and user experience across an area,” MediaTek says. That would not eliminate the use of transmitting “cells,” as the architecture is described as “cell free,” but rather mean user devices could communicate with more than one radio site at a time. 

There would still be transmission cells, but user devices would be free to connect with any adjacent cells and not be restricted to a single tower or radio. The practical advantage for end users is that the edges of a cell would no longer be a transmission quality issue. 

With a traditional one-cell connection design, lower signal strength at cell extremities has always meant reduced signal quality. In the cell-free design, signal quality close to the radio would be the same as quality at the edge of any single  radio’s coverage. That includes the degree of packet loss. 

The ability to incorporate non-terrestrial (satellite) connections with mobile connections also is envisioned. Artificial intelligence should be a native capability. 

In the spectrum area, frequencies in the 7 GHz to 24 GHz bands will probably be parts of the 6G standard, as will spectrum sharing, MediaTek says.

What is Included in "Digital Infrastructure?"

What is included in the more-popular term “digital infrastructure” (infra) these days. Equinix, which most of us would say is in the data center business, says it is in the “digital infrastructure” business. But note: Simplicable says DX includes:

• Cloud computing

• Platforms

• Systems

• Applications

• APIs and integration

• User devices

• Internet of Things

The point to note is that although some think of infra as limited to connectivity networks, practitioners refer to the ecosystem of applications, devices, platforms, hardware and software, computing, networks and systems that support digital work and life. 

Brookings in a study also includes open source as infra. IDC emphasizes cloud services

“for enhancing customer experiences…and…business operations.” Note the explicit link between infra and customer experience and business operations. 

According to Base22, digital infrastructure includes:

• Internet backbone, broadband

• Mobile telecom and digital communication suites, including apps

• Data centers and networks

• Enterprise portals, platforms, systems, and software

• Cloud services and software

• Operational security, user identity and data encryption

• APIs and integrations

The Ford Foundation says “digital infrastructure is the code, policies and standards powering the technology that permeates every aspect of life,” Ford does not mean “code” in a software sense, but in technology values such as equity.

Likewise, the Harvard Business School Digital Initiative looks at infra through a social impact lens as well. 

HCL says “digital Infrastructure is defined by a focus on enabling business agility and powering user experiences that drive customer engagement and loyalty.” 

The World Economic Forum, sees infra through the lens of digital economy, notes that include communications service providers, or CSPs (fixed line and wireless telecommunications companies, cable companies, and bandwidth providers), digital service and content providers (content, media and IT service companies), and hardware and software manufacturers (infrastructure equipment, device, software and component manufacturers) are all parts of the digital infrastructure. 

source: World Economic Forum

The point is that digital infrastructure includes much more than data centers and communication networks. Infra includes  chips, devices, end user software, cloud computing, data centers and connectivity networks, platforms and applications. 

Some would even include social or business impact. A few might include user digital literacy. The point is, it might be better to think of digital infrastructure as part of an ecosystem that overlaps with economic and social goals.