Category Archives: The Industry

Are Broadband Prices Dropping?

The FCC recently asked for comments in Docket 26-78, which is the latest iteration of its biennial report to Congress that looks at the State of Competition in the Communications Marketplace. Various industry players provided input to the FCC on issues related to competition and pricing for broadband and cellular service, with fewer caring about voice and cable service.

One of the issues widely discussed in this year’s filing is broadband prices. Some of the big ISPs continue to assert that broadband prices are dropping. For example, USTelecom refers to a report it generated that asserts that Internet prices have fallen for the eleventh straight year. I’ve written about the annual USTelecom reports before, and a big part of their assertion comes from looking at the price over time of the cost per megabit of speed being sold. On that basis, prices are dropping, mostly because ISPs have been increasing the speeds being delivered at a faster pace than prices.

One set of comments came from the Benton Institute, which described the issue perfectly. They cite the example that the price for 200 Mbps was around $50 in 2021. Many ISPs have unilaterally increased speeds without increasing price, and the average price for 400 Mbps in 2025 was also around $50. While the cost per megabit cut in half, customers are still paying $50.

Of course, ISPs don’t sell, and consumers don’t buy broadband by the megabit. Benton made a humorous observation on the big ISP’s focus on cost per megabit. Benton cites a USTelecom comment that the price per megabit for gigabit service is around 7 cents per megabit, or $70 per month. If USTelecom members are happy with that price, then why aren’t they applying that price to slower products so that 200 Mbps would cost $14 per month?

Perhaps the best discussion of prices in the docket comes from a study by John P. Horrigan, PhD, which is attached to the Benton comments. Horrigan takes a neutral look at prices and found that the weighted average for all broadband products increased by 4.8% from 2024 to 2025. Horrigan found that broadband prices for products slower than gigabit declined 8.5% from 2024 to 2025, with prices increasing for faster products.

Horrigan found that low-price options are disappearing from the market. When the ACP plan was operating, 9% pf broadband being sold was priced at $30 or less. He says this fell to just 3% of the market in 2025. This also holds true for plans with slower speeds. In 2022, 57% of consumers were buying Internet at a speed of 100 Mbps or less. In 2025, that has dropped to 32% of the market.

While the Benton Institute comments hint at it, I think most other comments in the docket are missing the bigger picture. Customers are choosing to migrate to lower-cost broadband options. One doesn’t have to look any further than the phenomenal success of FWA cellular. Since 2022, 16.5 million customers have subscribed to FWA cellular. While some of these customers live in rural areas where FWA is the only fast broadband option, I think a vast majority of these folks choose FWA to save money. The list prices for FWA home broadband are in the $50-$60 dollar range. However, there are big discounts for bundling with cellular service and for using autopay, and it’s possible to buy FWA home broadband for as little as $20-$30 per month.

Any analysis that just looks at prices for specific speeds over time will account for folks willing to take less speed for a lower bill. The big ISPs don’t want to talk about this, but there is no other way to discuss the huge success of FWA without talking about customers self-selecting lower prices.

How Good is Rural Cellular Coverage – Part II

Yesterday’s blog looked at AT&T cellular coverage in a typical rural county in Illinois and included the following map. The map shows where AT&T can provide 5G coverage in a moving vehicle in the dark areas, and where somebody standing stationary outdoors could get a 5G signal in the lighter colored areas.

Let’s look at the maps for the other two major carriers in the same areas. The first map below is T-Mobile, and the second is Verizon.

These maps show typical coverage. The two carriers support 5G in moving vehicles in and close to towns and cities. The light colored areas are where somebody standing outdoors can likely get a 5G signal. An indoor cellular coverage map would likely not be a lot larger than the dark areas.

Taken altogether, these maps show a typical rural story of cellular coverage. Cell carriers rarely share towers, and each carrier is on different towers and has different coverage. All three carriers have areas where they have no 5G coverage, and somebody subscribed to any one carrier in this county would find a lot of dead zones. All three carriers have little or no coverage in the northwest sector. These maps show something that every rural delivery driver knows – to work in rural America means carrying multiple cellphones subscribed to different carriers.

When Chairman Carr says that 96.8% of households have 5G coverage, we have to put that into perspective. Over 80% of Americans live in cities and suburbs and likely have good cell coverage. Another substantial percentage live in smaller towns that happen to have at least one cell tower. In this particular county, 60% of people live in incorporated towns and villages, meaning there are a lot of rural residents.

What’s the point of these two blogs? The FCC considers this County to have good 5G coverage. That assumption comes largely from looking at the combined coverage of the three carriers shown for somebody standing stationary outdoors. The light colored areas of the three maps combined cover most of the county.

If the FCC ever decides to finally launch the 5G Fund for Rural America, this county will likely not be a candidate for a grant to build new cell towers. That’s unfortunate, because I estimate that 30% of the residents of this county would say they have poor cellular coverage. They will say that they don’t have good coverage indoors, and no matter which carrier they subscribe to, they hit dead spots when they drive around the county. The FCC’s assertion that 96.8% of homes have good 5G coverage can be supported by the FCC maps – but those maps don’t show the reality of the way that people judge cellular coverage.

How Good is Rural Cellular Coverage – Part I

The FCC has opened a docket that periodically looks at ways to improve the FCC’s broadband and cellular maps. As part of that docket, Chairman Brendan Carr issued the following statement: On the mobile side, 96.8 percent of locations have access to mobile 5G services of at least 7/1 Mbps.

To put that into perspective, there are roughly 116.7 million total passings counted in the FCC maps, and the Chairman is saying that all except 3.7 million have good access to 5G. The Chairman’s statement can be supported by the FCC cellular maps, but I think the reality in rural areas is far different than what is shown on the maps. I’m not saying that the FCC maps are a lie – because I think it’s likely that the maps represent what the FCC asked carriers to report. But I think the maps tell a different story than what Chairman Carr is pushing, and I don’t think anywhere near 96.8% of folks in the country would say they have good cellular coverage.

Let’s look at the FCC maps for 5G coverage in an actual county in Illinois. I didn’t pick this county because it doesn’t have good cell coverage. The coverage in the counties around it would all tell the same story. One thing to note about this county is that there are homes located in all parts of the county – the areas with no coverage on these maps are not parklands or forests.

The following map shows AT&T 5G coverage from the FCC cellular maps. The FCC asks carriers to show coverage in two ways. The darker orange areas are where AT&T claims that 5G coverage will work in a moving vehicle. The lighter areas are where AT&T says that a customer can receive 5G when standing stationary outdoors. AT&T is claiming no 5G coverage in the gray areas.

This AT&T map is typical of rural cell coverage. Cell towers are located roughly in the center of the dark-colored areas, and those areas mostly covering towns and cities. Anybody who understands cellular technology understands that speeds drop quickly with distance from a cell site. The cellular download data speeds at the center of the dark areas could easily be as fast as 300 Mbps. But within two miles of a tower, speeds drop to around 25 Mbps. 5G speeds and coverage in the light-colored areas are a lot slower and spottier, and as you get to the outer parts of the light-colored areas, farthest from the towers, it’s likely that somebody would have to move around in their yard to find the sweet spot where they could make a call.

What this map doesn’t measure, and the FCC doesn’t ask about, is indoor cellular coverage. It’s a general rule of thumb that indoor speeds are roughly half of outdoor speeds. You can easily test this by taking an outdoor cellular speed test and then an indoor test away from a window (turn off your WiFi). If the carriers were to map expected indoor cellular coverage, the areas with indoor coverage would be a lot smaller than the light-colored areas shown for outdoor coverage.

When you ask a rural resident what good cell coverage means, they will define it as working in their home and working in their car. With that definition, AT&T doesn’t have great 5G coverage in the county for people who live or drive outside the dark circles.

Tomorrow’s blog will compare AT&T’s coverage to T-Mobile and Verizon to show the overall picture of cell coverage in this county.

Global Broadband Prices

It’s interesting once in a while to look at how the U.S. compares to the rest of the world in terms of broadband prices. The prices used in this blog come from Broadband Genie, which is a firm that compares local ISP prices in the U.K. For the worldwide price study, it gathered the average broadband prices from 214 countries. Prices were gathered in January and February of this year and are collected from public ISP websites or other trusted broadband comparison websites.

This is far from being a scientific study. Just imagine the difficulty of determining the average broadband price in the U.S. However, the study calculated the average U.S. broadband price of $80, which feels a little high to me, but not by much. You can judge the other prices accordingly. The price comparison doesn’t discuss the availability of broadband in each country, which varies widely across the world. All of the prices quoted below are in U.S. dollars, as calculated by Broadband Genie at the conversion rates at the time of the article. The article includes a database that provides the average cost in the local currency in each country.

There are ten countries with average prices below $10 U.S.. The countries with the lowest-cost broadband are:

Iran                  $2.61

Ukraine           $5.35

Ethiopia           $6.46

Bangladesh     $7.38

Mongolia         $7.41

The most expensive prices are as follows. As might be expected, a lot of the highest broadband prices come from islands, and I imagine the cost of international underwater backhaul has a lot to do with the higher prices.

Eswatini                        $193.21

Saint Barthelemy        $207.26

Turks and Caicos        $252.00

Turkmenistan              $286.24

Wallis and Futana       $373.88

The report also calculates a weighted average price in 22 regions of the world. The lowest prices by region are in Eastern Europe ($15.76) and Northern Africa ($20.83). The highest is Polynesia ($118.48).

Finally, the prices for some of the largest countries include:

India                $8.82

Russia              $9.71

Indonesia        $10.66

China               $14.30

Brazil               $23.08

South Korea    $26.92

France             $29.77

U.K.                  $31.43

Japan               $32.62

Germany         $47.59

South Africa    $50.20

U.S.                  $80.00

25-Gigabit Home Broadband

I recently read an article that touted residential 25-gigabit fiber in Switzerland. The article made it sound like the product was available everywhere, and the reality is different, but it is still a great story.

The fiber network being discussed is funded and owned by Swisscom. The company is owned 51% by the Swiss government and 49% by investors. During nationwide discussions in 2008 about the best future path for building fiber in the country, Swisscom pushed for the idea of building multiple fibers whenever new fiber is built. The decision was made to adopt what was called a four-fiber point-to-point model, which means four separate fiber paths built between the network core and each home and business. This would allow for an open-access network with four separate ISPs getting direct layer 1 fiber access to each customer.

In 2020, Swisscom changed its mind and decided to pivot to a point-to-multipoint using splitters where multiple ISPs share bandwidth on the same fiber, which is the traditional open-access architecture that has been used in many communities in this country. This is not entirely surprising since Swisscom doesn’t just own the fiber network, but is also the incumbent telecommunications company that still sells ISP services.

Init7, a small ISP, challenged the change in direction and filed a complaint with COMCO, the regulator in Switzerland. COMCO issued a preliminary decision at the end of 2020 that Swisscom needed to return to the four-fiber model. Swisscom appealed to a federal court, which sided with Init7. In April 2024, COMCO fined Swisscom 18 million francs for being anticompetitive and finalized the ruling that mandated a return to the four-fiber open-access model. Swisscom has now completed fiber construction to over half of the passings in the county and has a goal to pass 75% to 80% of passings by 2030.

The primary ISP offering 25-gigabit service on the Swisscom network is Init7, which uses PON electronics from Zyxel. The company markets under the brand name Fiber 7. The 25-gigabit product is priced at 65.75 francs ($82.73 per month) or 777 francs ($992.70 per year).

There are only a handful of ISPs around the world that widely deploy 25-gigabit broadband technology for residential service. In the U.S., there are now multiple fiber ISPs with products as fast as 5 to 8 Gbps. I have to wonder if there is any practical noticeable difference between these products and 25-gigabit broadband.

Telcom Liechtenstein announced the launch of a nationwide 25-gigabit symmetrical residential product in December 2025. The ISP is using Nokia’s 25G platform. Liechtenstein is unique in that it is one of the few countries that has near-universal fiber coverage. The underlying fiber network is owned by Liechtenstein Kraftwerke, the national power utility, which offers open-access to multiple ISPs.

NURO Hikari in Japan announced the launch of 25-gigabit residential service in parts of Tokyo in March. This network is utilizing electronics provided by So-net, a subsidiary of the Sony Group. The 25-Gbps product is priced at 6,480 yen ($41.08) per month.

Increasing Pushback Against Data Centers

It seems like I’m seeing articles almost every day about local pushback to the creation of new data centers. This sudden surge of antagonism seems to have caught the people who build data centers by surprise.

The following are just a few of the dozens of examples of communities that are skeptical or that don’t want new data centers:

  • After public feedback, local elected officials in Peculiar, Missouri, passed an ordinance to block a $1.5 billion data center proposed by Diode Ventures.
  • A $1.3 billion data center project was withdrawn from consideration in Chesterton, Indiana, following massive community pushback over environmental concerns.
  • In Fauquier County, Virginia, residents successfully pressured Headwaters Site Development to withdraw a $400 million data center project.
  • Residents of Prince George’s County, Maryland, persuaded elected officials to enact a six-month moratorium on data center construction in late 2025.
  • The legislature of Maine passed a new law creating a moratorium on new data centers. While that was vetoed by the governor, the state push was not unique, and similar moratoriums have been discussed by the legislatures in Georgia, Oklahoma, and Vermont. Other legislatures, like Illinois and South Dakota, have scaled back tax incentives that were aimed at attracting new data centers.

It’s an interesting public debate. There are clearly some significant benefits from bringing a data center to a community.

  • Job Creation. There is a big burst of economic benefit while a data center is being constructed. While most data centers bring fewer jobs than suggested by the data center owners, the jobs they bring have high salaries.
  • Tax Base. Assuming that a community assesses them properly, a data center should bring a nice boost to property and other local taxes. The extreme example of this is in Loudoun County, Virginia, which is home to a huge number of data centers. In 2025, the data centers contributed $895 million in local taxes, which represented 95% of the entire County budget.
  • Infrastructure Improvements. The infrastructure needed to support data centers can benefit the wider community if done right. Bringing a data center means new roads, an improved electric grid, modernized water infrastructure, and fiber optics.

The big public pushback comes because there are also downsides to data centers.

  • Huge Users of Electricity. A traditional data center used for cloud services might use as much power as 25,000 homes. An AI data center of the same size might require enough electricity to power 100,000 homes. The new giant data center being built in Louisiana is expected to draw twice as much power as the entire City of New Orleans. Communities worry about higher electric prices, brownouts, and electric shortages.
  • Huge Users of Water. Data centers generate a lot of heat. A chip used for AI consumes 700 to 1,000 watts each, compared to 150 – 300 watts used by a traditional chip used for cloud services. The huge use of power generates heat, so data centers must be cooled. An average AI data center might need up to five million gallons of water per day for cooling. Communities worry about the strain on water systems, particularly in parts of the country that see periodic droughts.
  • Data Centers are Noisy. Data centers generate significant, continuous noise that typically ranges from 55 to 85 decibels. This is generally experienced by neighbors as constant, low-frequency humming, similar in volume to a vacuum cleaner. When backup diesel generators are used or tested, noise can grow to 110 decibels, which is equivalent to the noise generated by a rock concert.
  • Air Pollution. Many data centers are generating their own electricity by constructing a power plant fueled by natural gas or other fossil fuels. Almost all data centers use diesel to power backup generators, and it’s not unusual for a hyperscale data center to have several hundred huge diesel generators. Neighbors say that living close to a data center is like living close to a traditional electric power plant in terms of air pollution.
  • Electronics Waste. The high heat and constant usage for AI can burn out cards in less than two years. This means data centers generate a lot of electronic waste that includes significant amounts of toxic materials and heavy metals. Most local landfills are not prepared for a large quantity of this kind of waste.
  • Require Large Plots of Land. Hyperscale data centers occupy a large plot of land that might otherwise be used for agriculture, residential housing, or industrial expansion. Since nobody wants to be a neighbor to a data center, there is also a larger circle around new data centers where others don’t want to build.
  • Not Transparent. A surprising number of new data center developers are doing things like requiring local officials to sign NDAs before showing their full plans.

The growing distrust of new data centers is not universal, and many communities are actively seeking new data centers because of the benefits. But a growing number of communities are deciding that the downsides outweigh the benefits.

Abandoned Rural Calls

I’m hearing an increasing number of stories from rural ISPs and telcos about voice calls that are not completing to their customers. People place a call to customers on a rural network and give up when they don’t hear the phone ringing at the receiving end of the call in a reasonable amount of time. The industry term for this phenomenon is an abandoned call, which generally occurs when the caller assumes the call didn’t work.

You might assume that this means that something is wrong with the PSTN (public switched telephone network) that is stopping calls from being completed. That would be a huge problem, and one that would also affect calls made to urban areas. From what I’m hearing, this is strictly a rural problem. The telephone environment has changed a lot over the years. Telephone calls today originate from a dizzying array of different sources. While people can still make phone calls from landline telephones and cellphones, they can also place calls from numerous online platforms, applications, and devices.

I think it is far more likely that this is happening for financial reasons and is related to the fees charged to terminate long-distance calls. Rural carriers still charge a fee, called an access charge, to terminate a long-distance call made into their local network. Access charges were created in 1983 when the FCC approved Part 69 rules that were put into place after the divestiture of AT&T into several regional Baby Bell telephone companies, with AT&T remaining as a long-distance company. Access charges were the mechanism by which long-distance companies compensated the telcos that owned the local infrastructure needed to reach customers and complete long-distance calls.

Access charges were originally fairly expensive, and I recall access charges in 1984 being around five cents per minute, even in some of the Bell companies. That may sound high, but at that time, most long-distance rates ranged between twelve and fifty cents per minute. Over time, The FCC forced a series of drastic reductions in access charge rates, and today the rate to terminate a call in urban areas is at, or just barely above, zero. The cost to terminate a call in most rural areas has been reduced to a small fraction of a penny per minute. Most people probably think that long-distance call are a thing of the past since they no longer pay by the minute to call, but long-distance is still very much real, and companies like cellular carriers charge customers a flat rate to cover the cost of the calls.

I think the resurgence of abandoned calls is due to least-cost routing. Anybody company with customers who originate calls, be that a telco, cable company, VoIP provider, or some online app, must pay to have that call terminated at the other end. This has historically been done by using long-distance carriers that carry the call between the call originator and the called party. However, there is an industry segment that few people know about. There are a lot of companies generically referred to as intermediate carriers that provide the function of carrying calls between carriers.

That’s where least-cost routing comes in. Long-distance companies use real-time software to determine the lowest cost to get a call completed. The long-distance carrier might deliver many of the calls using its own network. But it will hand calls off to an intermediate carrier that charges less than its own cost to complete the call. I think the dropped calls are happening because intermediate carriers also have least-cost tables, and they also hand off some calls to another intermediate carrier if that saves them money. This process is automated, and it’s possible for a call to be handed off multiple times to different intermediate carriers. Each transfer between carriers takes time, and the customer making the call abandons the call when nothing is happening.

The phenomenon of abandoned calls to rural areas is not new. This was an issue in 2017, and the FCC implemented rules from the Improving Call Quality and Reliability Act of 2017 (RCC). Those rules did not forbid using multiple carriers to route a call, but established regulations to ensure reliability and accountability, particularly to prevent rural call completion issues. In those rules, the originating carriers were held responsible for making sure that calls are completed. The rules required intermediate carriers that touch calls to be registered with the FCC, and it was forbidden to hand calls to an unregistered carrier.

The FCC needs to deal with this issue again, because something has broken down. There might be new, unregistered carriers in the mix. Or maybe AI is now involved and is making poor routing decisions. But it’s a problem that must be fixed. If not, rural residents won’t be able to receive calls, and rural businesses will be at a huge disadvantage.

Broadband for Precision Agriculture

The Fiber Broadband Association recently published an interesting article talking about broadband’s role in precision agriculture. For those not familiar with the term, precision agriculture is a data-driven, technology-enabled management strategy that uses satellite data and IoT sensors to optimize inputs like water, fertilizer, and pesticides to improve efficiency, profitability, and sustainability. It also includes autonomous machines like tractors, sprayers, combines, and drones to perform tasks like planting, weeding, and spraying.

The article argues that the FCC’s definition of bandwidth, at 100/20 Mbps, is not fast enough to support a rigorous precision agriculture application. The FBA’s Agriculture Working Group recommends that a speed of at least 100/100 Mbps is needed for precision agriculture. The faster speed is due to the real-time feedback needed by sensors and self-driving equipment. The article rightfully recognizes that the only two technologies that can support those speeds are fiber and HFC networks that have upgraded upload speeds.

I think the article mostly skips over the most important aspect of precision agriculture, which is getting  broadband to the fields. The article does acknowledge that the most resilient network would have wireless at the edge and fiber at the core. But fiber at the core means at the farmhouse. The challenge is getting the bandwidth from fiber at the farmhouse out to the fields.

According to 2023 data from the USDA, 27% of all farms used some form of precision agriculture, and it seems likely this has increased since then. But that statistic is heavily skewed towards large farms, and around 70% of large farms used some form of precision agriculture, while adoption is much lower in smaller farms with gross incomes under $350,000.

How are farms making this work today? If a farm is lucky enough to be covered by decent cellular coverage, then 4G or 5G can be used for smart devices. But a large portion of rural America has poor or no cell coverage. Even where this works, it can be expensive to buy separate cellular subscriptions per device.

A lot of farms are now using Starlink. This became easier when John Deere and other equipment makers started to build Starlink capability into new gear. This also requires a subscription, mostly through farm groups or the farm equipment manufacturer.

The most complicated solution, but one that will work everywhere, is for a farmer to construct a private wireless network using licensed or unlicensed spectrum. This can work as long as the farmhouse has a strong broadband signal at the core. This has several big drawbacks. It means a sizable upfront outlay to build the wireless network. There aren’t any easy off-the-shelf options, and the farmer would also have to master the technology. There is also an ongoing effort to babysit the network. I’ve had several farmers tell me they are feeling more like IT guys some days than like farmers.

According to FBA, Starlink is an interim solution that doesn’t have enough speed and bandwidth to keep up with the future data demands of precision agriculture. But Starlink has a huge advantage in being available now. There are two possible solutions for meeting faster future broadband needs. First, Starlink’s new generation of satellites might provide the needed speed to make it more than an interim solution. But a farmer who wants to guarantee a robust network might have to build their own solution, and that will require an off-the-shelf network solution that can be easily installed and maintained.

Broadband Usage 1Q 2026

OpenVault recently published its Broadband Insights Report for the end of the fourth quarter of 2025. One of the most useful statistics from OpenVault is the average monthly broadband usage for households and small businesses in gigabytes. Below is the trend in average monthly U.S. download and upload volumes since the first quarter of 2022.As can be seen in the table, upload usage has been growing at a faster pace than download usage. In this report, OpenVault credits most of the growth in upload usage to computers syncing with the cloud. I expect that the average household would be surprised by the volume of data they are uploading each month and probably wonder what data their computer is uploading.

This quarterly report also highlights two other broadband topics. First, OpenVault compared usage for customers on fiber networks compared to those using cable company HFC networks. Had I been asked to guess at the results of this comparison, I would have guessed the usage would be similar for customers using the two technologies. I was surprised to see the results shown in the following table.OpenVault shows that the average fiber customer downloads 26% more data and uploads 88% more data in a month than a customer on a cable company network. I’ve been thinking about these differences for several days, and I can’t think of any obvious technology reason that would drive the difference between the two groups of customers.

Both groups of customers are mostly urban and suburban. I would guess this isn’t related to price; in most markets, the list price of fiber is lower, although cable companies typically will match or beat a fiber price when asked. My intuition says that this difference is likely due to demographics and not a difference due to the technology. My guess is that fiber attracts younger users and those with families, while older households are sticking with the cable company. For example, I know families with serious gamers who are only interested in fiber. Conversely, there are still over 58 million households with traditional linear cable from a cable company, and I expect they are collectively older than average. It’s an interesting topic for a researcher to tackle. OpenVault is good at analyzing raw usage data, but it has no way to add color to the data based on demographic factors like age and income.

My first reaction when reading this part of the report is that I now expect to see articles that cite this OpenVault report to claim that cable company technology is somehow restricting customer usage. The statistics in this report don’t support such a premise. In fact, the report shows that cable has faster average download speeds.

The OpenVault report also compared the way that homes and businesses use broadband. The analysis showed that households use 23 times more download than upload. For households, video represents 48% of all download data use. Businesses use broadband in a very different way, and businesses use 7.3 times more download than upload. Said a different way, businesses use a lot more upload, with 20% of all business usage used to connect to the cloud. OpenVault reaches a conclusion that I think most ISPs already understand, which is that residential and business customers should be treated differently for network planning.

Fiber Matters in the Market

Roger Entner, of Recon Analytics, published an article in Light Reading that challenges the paradigm of the benefits of convergence. Convergence has most recently come to mean bundling broadband and cellular service. There is a widespread industry belief that ISPs need to have a cellular product to thrive, and cable companies have added a cellular product in the name of convergence. Entner says there is only one kind of convergence that makes a real market difference – fiber ISPs with a cellular product do far better than any other kind of convergence.

He bases his conclusions on different sets of facts. First is the results of 1.2 million surveys given by Recon Analytics between April 2023 and March 2026, asking about ISPs that offer a cellular product. Those surveys showed that, by far, the best indicator of a high cellular market share is the ownership of fiber. Other factors like brand, network quality, price, or specific details of the bundle had a far smaller impact on cellular market share.

Entner also looked at some specific examples. In markets where AT&T has no wireline network, the company has a 13.9% cellular market share. Where AT&T is the incumbent telco but doesn’t own fiber, the cellular market share is 6.2% higher at 20.1%. But where AT&T owns significant fiber, its cellular market penetration rate is 28.9%. These statistics show a bump in cellular market share for being an incumbent, but a bigger bump of an 8.8% market share for having fiber. The statistics for Verizon are similar.

Entner’s findings mean a massive financial boost for a company with both fiber and a cellular product. The statistics show a big extra financial boost for building fiber that I’ve never seen a fiber ISP talk about. This extra boost from a higher cellular market share means there is a greater benefit of building fiber than just fiber revenues, and greatly increases the value of investing in fiber.

What does this mean for cable companies? Comcast first started selling cell service in 2017, and Charter in 2018. If just owning the network was the driver of a high cellular market share, these two companies would be winning the cellular marketing battle since they have far more passings and customers than telcos. Enter concludes that the perceived quality of the ISP is what matters. He compared the Customer Net Promoter Score (cNPS) for the two industry segments. Big fiber ISPs have a cNPS around 27, while big cable is at 2.6. For those not familiar with cNPS, the number is derived by subtracting the percentage of customers that don’t like an ISP (detractors) from those that do (promoters). The difference in customer perception between fiber and cable companies, as shown by the cNPS ratings, is gigantic.

What does all of this mean in the market? First, this one simple rating explains why AT&T is building so much fiber, why Verizon bought Frontier, and why T-Mobile is buying ISPs left and right. These companies all clearly understand the extra market benefit of owning fiber.

This also explains why smaller fiber overbuilders are trying to find a way to add an affordable cellular product, since they assume a big earnings boost from the convergence. That desire comes with a word of caution for rural fiber owners. It could actually be a negative to be the cellular company in a rural market where the cellular coverage is crappy, as is true in much of rural America.

Interestingly, Entner ignored FWA cellular, at least in this article. (The full report is available for a fee.) FWA cellular providers currently have an overall cNPS rating of 40, which is significantly better than the big fiber ISPs at 27. The high cNPS for FWA means that customers are much happier with FWA broadband than with fiber or cable. This might explain why FWA cellular has been outselling all other types of broadband for the last several years.

It’s worth noting that customer sentiment and cNPS scores change over time, so what is true today will probably not be true five years from now. Cable companies got a huge black eye during the pandemic when they struggled to support people working and schooling at home. I think this is when fiber got a better reputation than cable companies. But cable companies have been making big investments to improve both download and upload speeds, and over time, the public will likely eventually feel better about them. Cable companies have also quietly been building fiber, but for now, the big ones seem to mostly be content to improve speeds on existing networks.

This is an interesting way to look at convergence, and it explains a lot about the market experience of fiber ISPs, which are growing, and cable companies, which are losing broadband customers.