How Does Starlink Stack up Against Fixed Wireless and Geosynchronous Satellite Service?

Low earth orbit satellite constellations such as Starlink promise higher bandwidth, lower latency internet access that eventually might cover every part of the earth’s surface. With Starlink’s early commercial launch, how does the service stack up against other alternatives--fixed and wireless--in terms of value?

Starlink users in the United States and Canada have seen 50.5 Mbps download speed, 14 Mbps upload speed, and 52.5 ms latency, reports Speedcheck. Latency performance is better by an order of magnitude, according to Speedcheck. 

Compared to existing geosynchronous satellite offers, Starlink beats HughesNet for maximum downlink speed and trails Viasat. But value is a combination of performance and price. 

The Starlink price is arguably lower than HughesNet at 25 Mbps and Viasat at its top speed as well. But Starlink does not cap usage, as do Viasat and HughesNet. 

Company-	Starlink	Viasat	HughesNet
Price	Dish $499 + S&H, $99 per month	$299.99 (life time lease), $30-$150 (for the first 3 months)	$499.99, $59.99-149.99 (requires 24 months agreement)
Data Cap	No data cap- Unlimited	Yes, (12-300 GB/month)	Yes, (10-50 GB/month)
Speed	15 - 107 Mbps DL (Measured), *22.99 Mbps UL	12 - 100 Mbps DL (Advertised)	Up to 25 Mbps DL (Measured), 3 Mbps
Latency	* 33 ms, (Measured)	600-800 ms, (Advertised)	600-800 ms, (Advertised)
Accessibility	Limited-Only some cities of US and 11 other countries	All 50 states	Mostly Southern Sky. US, Brazil, Columbia, Peru, Ecuador, Alaska, India, parts of Europe

source: Speedcheck

Starlink also might be considered a reasonable value choice when compared to 4G fixed wireless access supplied by Verizon, but a bit less reasonable where Starlink competes against 5G fixed wireless. 

Both fixed wireless versions cost less than Starlink and do not require a satellite dish. But both fixed wireless options by Verizon also supply less bandwidth. 

All the services offer unlimited usage. But Starlink offers more downlink and uplink bandwidth. Latency performance is comparable to 4G but less than 5G. At the moment, Starlink and 5G fixed wireless do not overlap much in terms of coverage, while Starlink’s footprint covers southern Canada and the northern United States. 

Verizon 5G fixed wireless is only available in large urban areas of the U.S. Northeast. 

Company	Starlink	4G (Verizon)	5G (Verizon)
Price	Dish $499+S&H, and $99 per month	$80/month	$80/month
Data	Unlimited	Unlimited	Unlimited
Speed	50 Mbps DL, 13 Mbps UL	17 Mbps DL, 3.84 Mbps UL	40 Mbps DL, 9.74 Mbps UL
Latency	57 ms	58 ms	39 ms
Accessibility	Limited to parts of North America	Nationwide coverage	Only in big cities

source: Speedcheck

As is typically the case, Starlink will not compete well with fiber to home or cable modem services running at speeds up to a gigabit per second. 

Satellite remains a niche. “Because of the higher relative cost of bandwidth transmitted via satellite versus terrestrial technologies, satellite is currently primarily used in situations where fiber optic cables and other high-capacity technologies are not financially viable due to low population densities and large distances between high-capacity networks and last-mile networks,” the Asian Development Bank rightly notes. 

source: Asian Development Bank 

Low earth orbit satellite constellations such as Spacelink will allow satellite broadband provides to serve 3.5 million subscribers in 2021, growing at an eight-percent compound annual growth rate to reach 5.2 million users in 2026, according to ABI Research.

To keep that in perspective, in 2021 there are about 4.93 billion regular internet users, using 1.2 billion fixed connections and upwards of seven billion mobile internet subscriptions, supporting mobile phone users, PCs and internet of things devices.

The point is that, as important as LEO constellations might be, they will remain a niche supplier of internet access services. By 2026, says ABI Research, LEO service revenue might reach US$4.1 billion. 

In 2020 fixed network internet access in the United States alone generated more than $100 billion in annual revenue in 2013, by some accounts. 

5G Will Run on Wholesale Model in Malaysia and Brunei

Reliance on wholesale market structure--one physical network and many retail competitors--has been a feature of fixed network policy for some time. It also is becoming the foundation of 5G platforms in some countries as well. 

Malaysia and Brunei, for example, will build a single national 5G physical network, which will support all potential retail providers of 5G service in those countries. Both of those networks will be state-owned. 

It is not a return to a monopoly framework at the retail level, but is a return to monopoly and government ownership at the facilities level. 

In the 4G era, only a few networks were built using the single wholesale provider model, notably in Mexico, Belarus and Rwanda. A few others were cancelled, including networks in Russia, Kenya and South Africa. 

As you might expect, such approaches are considered unwise by GSMA and other private firms. The dominant mobile framework globally continues to be private ownership of multiple physical facilities. 

Still, there has been concern about the cost of communications infrastructure for some time. But cost is but one of several drivers of national communications policy. 

source: Citi Research

The realm of possibility is directly affected by the costs of future networks. If the application of new architectures and technologies plus infrastructure sharing or offload contains the cost of physical networks, especially wireless and mobile networks, there will be less pressure on the business model and therefore less pressure to shift to wholesale-only infrastructure models.

Cloud Native is Not Simply a Change of Computing Architecture

In a move that tells us much about how telecom network have changed, Telefónica will build its cloud-native 5G core network using IBM Global Business Services as the system integrator while Red Hat and Juniper supply the networking platforms. 

In prior generations of mobile networks, the network would have been purchased almost turnkey from a legacy provider such as Ericsson or Nokia (or Alcatel or Lucent). It was monolithic. 

Now the network is integrated by a major system integrator using open source technology and a distributed, cloud-based and cloud native computing platform. It is loosely coupled. 

In other words, one builds a telecom core network as one would build any other computing network intended to scale. In other words, apps and functions are built how apps are created and deployed, not where they are created or deployed.  

source: Medium

The former emphasis would have been on proprietary hardware; the new approach makes heavy use of commodity hardware. The older approach relied on closed, vendor-specific software. The new approach relies on open source approaches. 

The older approach would have been voice centric. The 5G core network is essentially a computing network that runs voice and other applications, but is designed to support any app. 

Some will view this as a “mere” change of computing architecture. Others will argue it also changes the range of business models a telco can consider. In a monolithic environment, the network owner controls all apps that are lawful on the network as well as which devices can be attached. 

The internet and all cloud architectures do not work that way. All lawful devices and apps may use the network. That produces the product opportunity called “broadband access” for internet service providers. 

That same change creates the whole regime of “over the top” and loosely-coupled app creation that is “permissionless.” Any lawful and technology-compliant app can use the network. Any user or customer with the proper credentials can use such apps. 

To the extent that roles are disaggregated, so are business models. Ownership and operation of the access network does not limit third party rights to create apps that use the networks. But neither are there barriers to telco owners creating their own “third party” apps that similarly use any lawful access. 

The change is that where such apps would mostly have been designed to run “on my network,” they preferably now would be designed to run “on any lawful  network.” The difference is potential customers “only on my network, in my region or country” versus potential customers anywhere globally. 

All of that places new importance on creating value in lots of potentially new ways, most of those ways centered on becoming a supplier of end user apps or platforms to do so. Some of those apps will be “owned” by the telco and designed to run “on my network.”. Network slicing is an example. 

But many others potentially will be designed to run as any other internet app, on any network and in any country where the apps are lawful. Yet others will be targeted geographically, for reasons of language preference, business strategy, capital or human resources availability. 

The point is that the technology change to a cloud native, loosely-coupled, distributed and essentially open architecture changes the realm of business possibility. More new things are possible, but many existing things could be threatened.

How Travel Industry Thinking on Digital Transformation Applies to Telecom

As a practical matter, it sometimes is easier to understand fuzzy concepts such as digital transformation by looking at how other industries see it. In the travel industry, a digital transformation report produced by Skift and Amazon Web Services notes that   

digital transformation offers:

• Customer experiences related to communicating, shopping, and buying

• Pricing, price, promotion,  distribution, and purchasing

• Agility when demand or competitive environment changes

• Analysis and forecasting 

• Revenue, cost, efficiency improvements

• Remain competitive

• Adapt to changing industry circumstances

Note that none of those benefits are quantifiable. They are qualitative changes. So one lesson is that digital transformation is not directly measurable. What can be measured are other business metrics related to the business model, customer satisfaction, acquisition and churn rates or operating costs, for example. 

sources: Skift, AWS

In other words, the emphasis is “transformation,” not simply “digital.” 

Though business process changes believed to be associated with digitalization can be measured, there are no direct quantitative measures of “transformation.” What can be measured is process performance.

sources: Skift, AWS 

What can be measured are the results of changes in business processes “digital strategy” was meant to address. That includes marketing, advertising, customer experience, analytics, automation, resiliency and so forth. More of this; less of that, in other words. 

Digital transformation is qualitative and non-quantifiable. Digitalization is quantifiable, but only as it relates to process outcomes. 

As all that might apply to connectivity businesses, the obvious answer is to focus less on the qualitative benefits (agility, speed, adaptability, revenue growth, cost containment, experience advantages) and more on the practical business outcomes and how digitalizing can help produce the desired outcomes.

Telco Disaggregation is Coming

As it is conceptually possible to separate simple replacement of existing processes with digital versions--keeping the processes intact--and the use of digital technologies to reshape and recreate processes (digital transformation), so it might now be possible to suggest future lines of development for “telcos and retail connectivity providers.”

In principle, as telecom and connectivity networks are virtualized, there are opportunities--or risks--related to reconceptualizing “who does what” and “who and what creates the value?” 

The risk is disintermediation: removing the distributor or middle man from the value chain. The potential upside is the ability to create higher-value services and apps faster, more affordably and more flexibly. 

We might then ask the question: “what is the value of the telco and where is that value located?” 

source: STL Partners 

In part, this is a shift from a vertically-integrated to horizontal model. That can have huge business model implications. The vertical model is closed; the horizontal model is open. The vertical model tends to come with higher costs; the horizontal model with lower costs. 

The vertical model tends to feature lower rates of innovation, while the horizontal model tends to increase rates of innovation. 

The vertical model entails more control of products and business model, while the horizontal model tends to lead to  less control by the telco. Think of the internet as a horizontal model, where functions are disaggregated: access networks are distinct from platforms. Platforms are distinct from applications running on those platforms. 

Another way of looking at the difference is to say that a traditional telco was at the center of its own business. In the internet era, telcos are part of the transport and access function, but not necessarily part of the applications or platform functions. 

In other words, think of telecom as becoming part of the computing business. In that business communications functions are demarcated from applications, hardware and platforms. 

Communications are required, but logically separated from the other higher-order processes. Local area networks require spectrum and access points, but the actual applications, platforms, hardware and functions are logically separated. 

Ethernet cables might be necessary for a local area network, but nobody would mistake connectors and cables for the value of devices, software and functions those cables support.

Disaggregation might lead to opportunities related to creation of vertically-oriented services and apps, but it might also limit the ability to benefit from scale, to the extent that extensive domain knowledge has to be mobilized, and that knowledge will not translate directly to other domains. 

Still, the move to disaggregated processes is subtle. Telcos might gain directly agility, but also face new forms of competition. Telcos might be able to rework cost structures and create revenue growth opportunities, but also face some additional commoditization of contributed value. 

There will be a new balance of “what we do” and “what others do for us.” It is a complicated, multi-dimensional set of changes. 

An analogy is disintermediation and something we might call “re-intermediation.” The internet makes it possible to remove retailers from value chains. That gives us online commerce. But online commerce also allows the creation of a platform and marketplace (Amazon, for example) that recreates a role for the distributor. 

To the extent that telcos are distributors, they face the danger of disintermediation. Look at the move by Google, Facebook and others to own and build their own wide area networks. That removes customers and demand from the retail capacity business.

At the time, there is at least the possibility that exchanges (for bandwidth, cloud computing, edge computing, security, private networks) could be created. Perhaps the form would be a federation of exchanges globally, as it seems unlikely any single entity could emerge on the Amazon model, in terms of scale. But regional exchanges seem possible, as Alibaba is dominant in some markets, Amazon in others, third parties functioning in yet other markets. 

In any event, there are possible positive and negative surprises. Disaggregation does raise, ast least indirectly, a key question: what is the unique core competence of any telco? And how does that get leveraged in a disaggregated network and business model?

Altice Loses Broadband Subs, Comcast Sees Slowdown: Early Signs of a New Trend?

Altice USA reported a lost of broadband accounts in the third quarter, something that has been virtually unseen over the past couple of decades. Comcast, for its part, reported a slowdown in net acquisitions in its third quarter.  

Telco broadband net additions are growing. In the same quarter of 2020, telcos lost subs. Some analysts now believe telcos will continue to gain share from cable companies as telcos scale up their gigabit fiber to the home networks. 

Where cable once had 95 percent share compared to telco digital subscriber line services, and now has about a 70-percent share of broadband lines overall, some now believe telcos will eventually grab half the market.

Inflection Point for Gigabit Access?

To the extent that faster home broadband speed tiers cost more, there is a clear and compelling argument to be made that fixed network providers can increase average revenue per account as customers migrate to higher-priced tiers of service. 

Also, we might be on the cusp of a bigger opportunity for telco gigabit services, as history suggests take rates for gigabit services has reached an inflection point. 

source: Openvault

Historically, popular consumer products accelerate after adoption reaches the 10-percent of homes level, and gigabit internet access now has reached that point. 

Openvault data shows that consumers are migrating to speed tiers faster than 100 Mbps. 

source: Openvault

At the same time, markets for 4G fixed wireless are likely to be centered on rural areas, as urban customers move to services operating faster than 100 Mbps. 5G fixed wireless might be competitive for significant percentages of urban market customers.  In fact, 5G fixed wireless offering speeds between 100 Mbps and 200 Mbps might appeal to nearly half the market.

It is hard to quantify demand for symmetrical services. It seems clear that services with faster upstream speeds, such as telco fiber to home services, could be the near-term winners in that regard.

Digital Transformation is More than Data, but Builds on It

Digital transformation is much more than digitizing or relying on data, but the ability to harness data arguably is a requirement for DT. 

In surveying 4,036 data decision-makers, Forrester researchers found 55 percent of respondents struggling to meet their digital transformation goals. Some 15 percent of data strategy decision-makers say they have already realized their DT goals. 

Data decision-makers have already increased their investment in DT over the past three years by 77 percent and plan to increase their investment by a further 57 percent over the next three years, Forrester says. 

Three-quarters of data strategy decision-makers have seen the volume of data their firm generates increase over the past three years. Some 56 percent have also seen an increase in the amount of data they collect. 

source: Dell Technologies 

User Experienced Speeds are Slower than Delivered Speeds

Data from Speedtest Intelligence shows that 17 percent of U.S. counties with sufficient samples did not meet the minimum median speeds for the current FCC definition of broadband (25 Mbps download, 3 Mbps upload) in the second quarter of 2021, Ookla says. As often is the case, the data must be interpreted.

Those figures are based on end user speed tests, in most cases conducted on devices that are Wi-Fi connected. 

So one caveat is that the samples are not random. How many of us routinely test our access speeds when they are not a problem? Almost by definition, speed tests are conducted when there is a perceived problem. Also, most of those tests happen on Wi-Fi connections that are far slower than the actual speeds delivered to the router. 

Wi-Fi speeds can often be an order of magnitude slower than the wireline delivered speed, for all sorts of reasons. 

On this map, areas shown in dark blue do not meet the downstream minimum of 25 Mbps. The green areas show issues with upload speeds. Those areas are mostly rural. 

Aside from the “Wi-Fi speed, not delivered speed” issue, keep in mind that 86 percent of the U.S. land surface has home density less than 15 homes per plant mile.  

source: Ookla 

Most people in the United States live on just six percent of the U.S. land surface, according to the USDA. About 94 percent is unsettled or lightly populated, including mountains, rangeland, cropland and forests. 

That means people or locations unable to “buy broadband” are fewer than geographic coverage would seem to indicate. Networks serving most of the people can be built on a single-digit percentage of areas. 

But that still leaves huge amounts of space where networks are expensive. 

Some 92 percent of counties with sufficient samples were not operating at a minimum 100 Mbps standard in the second quarter of 2021, Ookla also says. 

What cannot be ascertained from the Ookla data is the actual speed delivered by internet access providers to the router. What we appear to be measuring is device experienced speed using Wi-Fi. Distance from the router; in-home obstructions; interfering devices operating; age of the router or devices all can reduce experienced speeds.

Has FTTH Business Case Changed for AT&T?

As difficult as the fiber to home business model has been for firms such as AT&T, some positive changes to the business model have occurred. First, take rates seem to have increased. 

Where AT&T has been deploying its new FTTH facilities, it is seeing that almost 80 percent of the net additional accounts were not AT&T customers prior to the deployment, according to John Stankey, AT&T CEO. So AT&T is capable of taking market share from cable companies. 

In past years, AT&T and Verizon, for example, might have been doing well to get to about 40 percent overall take rates, over several years of marketing, and most were likely upgrades of existing customer accounts. 

Since cable share of the installed base is about 70 percent in most markets, AT&T has room to grow if it can take share. 

It is conceivable that the value is not just gigabit speeds, but more upstream speed, compared to cable hybrid fiber coax offerings, which support gigabit downstream speeds as well, but are more limited in return bandwidth. 

But deployment efficiencies also have been reaped, and it may be possible to do more, once the neighborhood-by-neighborhood FTTH gigabit networks have been built. There are marketing economies and physical plant economies that would allow metro-wide marketing, for example. 

That changes the consumer fixed networks business into a growth opportunity that AT&T previously had not deemed so feasible. Right now AT&T believes the current business model works for about another 30 million U.S. homes. Any subsidies would grow that number. 

Right now, the new FTTH builds produce profit margins “in the mid- to upper teens,” said Stankey. While that is not the 40-percent range that once was feasible in many parts of the access business, that figure is notable because it is not a negative number. 

“I'm not going to be happy until we have a 50-50 share split in places where there's two capable broadband providers,” Stankey says. That means a split of the market with cable operators, where today cable has about 70 percent share of the installed base. 

Today, “we've driven to 40 percent penetration levels to basically be the threshold for us to get to warrant an investment,” said Stankey. 

If a new infrastructure bill passes, that could well result in subsidies for building new FTTH plant that favorably affect the business model for AT&T and others.

Mid-Band Will Change 5G User Experience

Mid-band spectrum powers 5G in most countries globally. The relative lack of deployed mid-band spectrum in the U.S. market explains the slower speeds we tend to see for 5G services in most areas. The difference so far has been millimeter wave spectrum deployed by Verizon and AT&T. 

source: Opensignal 

The shift from low-band spectrum to mid-band spectrum for 5G coverage will make a huge difference in user experience. Simply put, low-band is for coverage, not speed. Millimeter wave is best for capacity, but not coverage. Mid-band is the best blend of coverage and capacity. 

source: Opensignal

Despite Assertions, U.S. Broadband is Neither Slow Nor Expensive

One often hears it argued that U.S. broadband is expensive or slow. That might not actually be the case, as data published by the European Telecommunications Network Operators’ Association suggests. 

Simply put, the ETNO analysis suggests U.S. downstream speeds actually are higher than in South Korea, Japan, Europe or the global average. 

Comparing prices, some point to costs that are not indexed for currency values. Looking at spending as a percentage of gross domestic product or household spending over the last decade or so, U.S. prices have dropped since 2010, as have prices in South Korea, Japan and Europe. 

source: ETNO 

The universal trend in those regions--and throughout the world--is lower prices. 

source: ETNO 

There are lots of nuances. For example, “fiber to the home” does not equate to “gigabit speeds.” In South Korea, acknowledged to be a world leader in broadband access, “next generation access” is close to 100 percent. But “fiber to home” or “premises” is at about 40 percent. 

So the issue might not be “access media” but rather “capabilities.”

source: ETNO 

That is clear in the analysis of gigabit capable or “upgradeable” networks. In the U.S. market, cable operators lead the gigabit market. 

Also, not all FTTH networks actually are upgradeable to gigabit service levels without substantial rework. In South Korea and Japan, most FTTH networks are gigabit capable or upgradeable. In Europe, about a quarter of FTTH networks are gigabit capable or upgradeable. 

source: ETNO 

Also, average downstream speeds in the United States are faster than in South Korea, Japan, Europe, or the global average. 

source: ETNO 

The point is that the repeated assertion that U.S. broadband speeds are low, or that internet access is expensive, does not hold up, either internally over time, or in comparison to trends in other developed nations. Globally, internet access is getting better, fast.  

Adjusting for purchasing power, U.S. internet access was deemed “among the most affordable in the world” by the International Telecommunications Union. 

Also adjusting for purchasing power, using the purchasing power parity method, internet access prices are around $35 to $40 a month. IIn developed nations  prices are less than $30 a month.  

Internet access in the developed world--including the United States--simply is not that expensive.