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.

NTIA Trying to Regulate Through BEAD

NTIA has circulated guidance to BEAD winners titled BEAD Subgrantees: Protect Your Rights. Most of the two-page document is fairly routine stuff, but it also includes a bizarre section discussing permitting. The overall tenor of the document is odd in that it invites an ISP to directly contact NTIA if it thinks the contract offered by a State Broadband Office contradicts NTIA policy.

The document starts with a reminder that States can’t engage in ratemaking and demand specific rates in a BEAD contract. NTIA’s position on the issue is not controversial since rate regulation was prohibited in the original IIJA legislation that created BEAD. What is unusual is to see NTIA making a big deal out of this topic. My guess is that the NTIA guidance is mostly aimed at New York, where large ISPs that won BEAD are also subject to a state law that mandates a cap of $15 per month for qualifying low-income subscribers. Perhaps NTIA is hoping to goad one of the large ISPs in New York to use BEAD as a chance to challenge the state law, although the Supreme Court has twice refused to accept challenges to the legislation.

The guidance also alerts ISPs that each state is required to create a permitting roundtable where ISPs that encounter delays in permitting can discuss delays and fees. NTIA also seems to be reminding ISPs that States are required to document any problems encountered in implementing BEAD projects in semi-annual reports.

Where I think NTIA went off the rails is a set of requirements related to permitting:

  • NTIA wants a 90-day shot clock for the approval or rejection of permitting requests.
  • Grant winners can demand a single, dedicated point of contact for broadband-related permits.
  • Permits must allow the construction techniques chosen by the grant winner.
  • Batch processing of permit requests must be allowed.
  • Grant winners must not be subjected to unnecessarily duplicative or burdensome permitting requirements.

I find these requirements to be odd since NTIA doesn’t have the regulatory authority to specify permitting rules. For the most part, States also don’t control permitting rules and processes, which are left up to local jurisdictions. It’s highly questionable in most States if the Broadband Office can even assert any real influence over permitting practices for State highways.

NTIA has no authority to demand a permitting shot clock. NTIA can’t mandate that localities accept construction plans from grant winners. For example, what if a grant winner wants to bury fiber one foot deep instead of the locally-demanded three-foot depth? There are plenty of localities that won’t allow large-scale construction using trenching with a backhoe to bury fiber. Many local jurisdictions might be skeptical of microtrenching. Most localities will expect BEAD winners to abide by the same rules that apply to other telcos and utilities.

The last bullet point might be the most troubling since nobody knows what a ‘burdensome’ permitting requirement is. Is NTIA planning to intervene in disputes over local permitting rules that a grant winner doesn’t like?

NTIA also wants States to agree that permitting fees must be set at an approximation of actual cost. This is something that Congress could tackle, but any federal law demanding this would be heavily challenged in court.

The requirement that will get the most pushback is the requirement that a grant winner that also owns pole becomes subject to state or FCC pole attachment regulation by accepting the BEAD grant. As a reminder, cooperatives and municipalities are not subject to most pole attachment rules. NRECA, an association of electric cooperatives, wrote this letter to Commerce Secretary Lutnick, warning that many cooperatives will walk away from BEAD awards rather than let themselves be subject to pole attachment regulations.

I have to wonder if any BEAD grant winner will actually complain to NTIA to try to get a State to enforce permitting requirements or fees. I have to think that a State’s reaction to such a complaint would be to put that grant project on hold until the issue is resolved, which could take years and could even run out the five-year BEAD timeline.

The funniest part about the drama related to permitting is that very few, if any, local rural jurisdictions will make it hard for BEAD winners to get permits. Rural counties want better broadband infrastructure, and most counties I know will bend over backward to speed up the process. These odd NTIA rules don’t address the real source of the real permitting problems, which are railroad crossings, bridges, and state and federal lands. I’m honestly scratching my head, wondering why NTIA wrote this guidance. But BEAD has been odd since the beginning, so I guess there is no reason to stop the oddness now.

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.

Signs of Trouble for ISPs?

It’s always difficult for ISPs to fully understand how changes in the economy might impact them. Folks in the industry see the usual statistics on unemployment and inflation, but those don’t really tell much about the future as it relates to broadband adoption. I’m not an economist, and this blog is not a prediction, but in the last few weeks, I’ve heard a number of unrelated economic statistics that I find troublesome when taken as a whole.

  • MVPDs like Hulu, Sling TV, and Fubo are all reporting a significant loss of subscriptions. At the same time, the free ad-supported video services like Pluto TV and Tubi are seeing big customer gains. The press that covers these companies believe that subscription losses are mostly due to households cutting back on spending.
  • A Washington Post article reported that the number of households being cut off for nonpayment of electric and gas utility bills is climbing.
  • The delinquency rate for mortgages has been increasing. In the fourth quarter of 2025, the national rate was 4.26%, up 28 basis points from a year earlier.
  • Auto loan delinquencies are at the highest level in thirty years, and in early 2026, 6.9% of loans were 60 days overdue. More than 3 million cars were repossessed in 2025, a huge increase over prior years.
  • The U.S. consumer sentiment survey taken in April 2026 showed an historically low level of consumer confidence. This survey has been conducted monthly for the last 74 years by the University of Michigan. 22% of the respondents to the April survey reported deteriorating personal finances.
  • It seems like I’m seeing a new announcement almost daily from companies that are laying off thousands of employees or shutting down outlets or factories. Meta, Microsoft, and Oracle all recently announced huge layoffs. While there are always announcements of this sort, even in a robust economy, I can’t remember the last time I saw this volume of announcements.

None of this sounds like good news for ISPs. There has always been a general consensus in the industry that broadband is somewhat recession-proof. But is it really? The question of whether broadband is recession-proof is really asking if people will willingly give up the many things that they do online. Is there a point in people’s lives where broadband becomes a necessity that they will fight to keep when times get tough?

We already have strong evidence that broadband is related to household income. A recent Pew Survey showed that most households have smartphones. It also showed a strong correlation between household income and broadband adoption. The survey showed that 14% of homes with household incomes under $30,000 don’t have home broadband, while only 4% of homes with household incomes over $100,000 don’t have broadband. This leads to the fairly obvious conclusion that households will give up a broadband connection in favor of a smartphone subscription if money gets really tight.

Another thing to consider is that even if broadband is recession-proof, it doesn’t mean that people will continue to pay high prices for broadband. Consider the FWA cellular broadband sales from AT&T, T-Mobile, and Verizon. The three carriers have consistently been adding around one million new subscribers each quarter. The main attraction of FWA is broadband priced between $20 and $30 per month for customers who will bundle broadband with a cellular plan.

Of course, a recession is not inevitable and may not happen this year or next. But the statistics cited at the beginning of this blog tell the story that a lot of homes are in financial distress, even if the overall economy might not be in a recession. It’s possible that the traditional paradigms of what defines a recession no longer apply. Perhaps we’re seeing the economy collapse for the bottom earners in a way we haven’t seen before.

I suspect most of the people who read this blog think that broadband is essential for daily life. But the big question that will have to be answered is how many customers find home broadband to be indispensable. It’s easy for those of us live and breathe broadband to suppose that people think of broadband as a necessity – but is that really true for homes that can’t afford electricity or who can’t make car or mortgage payments?

Life Left in HFC Networks

There was a time when it seemed certain that cable companies would have to bite the bullet and spend the money to upgrade to fiber. While there have been some upgrades by cable companies like Cox and Altice, most cable companies seem to be deciding that there is still good life left in DOCSIS cable networks. As you might expect, CableLabs has been quietly working behind the scenes to improve existing HFC technology.

DOCSIS 3.1 networks have become standard across the industry, and it’s now rare to see older technologies except for some small cable companies. Cable companies have been using DOCSIS 3.1 networks to deliver gigabit or faster download speeds, with the top speed depending on the overall size of the bandwidth being utilized in the internal radio network that controls the signal.

The whole cable industry got a shock during the pandemic when it became obvious to many of the millions of students and employees who began working out of homes that the slow upload speeds on DOCSIS 3.1 were a bottleneck. I think the inadequacies of this technology and slow upload speeds are what gave a big jumpstart to the public perception that fiber is far superior to cable technology.

CableLabs and vendors responded to the upload speed bottleneck by introducing two solutions that can add to the upstream portion of the cable network. Labeled as midsplit or highsplit, both solutions require some upgrades in the outside plant electronics, along with upgraded cable modems in homes. The midsplit upgrade is accomplished by increasing the frequencies used to support upload from 5-42 MHz to 5-85 MHz. The highsplit upgrade allocated even more frequency to uploading, as much as 204 MHz. Both of these upgrades increased upload speeds to 100 Mbps or faster, which eliminated the bottleneck for the average customer. In 2025, CableLabs offered an even better version of the midsplit upgrade by offering a new cable modem that can handle up to four more channels of bandwidth.

CableLabs is also improving DOCSIS 4.0 technology. This is an upgrade that became available for cable companies in early 2024 that can provide symmetrical broadband speeds. While the upgrade can deliver speeds up to 5 Gbps, most cable companies are using it to offer symmetrical 2 Gbps broadband. This upgrade makes it practical for a cable company to say it can match fiber speeds – or at least it did in 2024. There are now fiber ISPs offering residential broadband at speeds up to 10 Gbps.

CableLabs has demonstrated an upgrade to DOCSIS 4.0 that can mimic the faster advertised speeds of fiber providers. CableLabs recently released a new standard it is calling DOCSIS 4.0 Optional Annex. This standard works by increasing the network bandwidth inside the coaxial cable to 3 GHz. Cable networks operate by using radio frequencies inside the coaxial wires. Most DOCSIS 3.1 networks use 1.0 to 1.2 of total frequency. Some companies have upgraded to 1.6 GHz for DOCSIS 4.0. This new optional Annex, double that bandwidth and will supposedly support speeds up to 25 Gbps. CableLabs is also looking at a version of the new technology that would increase total network bandwidth to 6 GHz, which might support broadband speeds up to 50 Gbps.

These new options will give pause to any cable company thinking about upgrading to fiber. These new technologies provide a realistic alternative to fiber with DOCSIS 4.0.

Restart on Digital Discrimination Rules

On May 6, the U.S. Court of Appeals for the Eighth Circuit vacated the FCC’s digital discrimination rules. The discrimination rules were required by the Infrastructure Investment and Jobs Act (IIJA). The FCC took an interesting approach to the issue and defined discrimination in two ways. The FCC prohibited intentional discrimination – meaning that ISPs can’t have policies and practices that are clearly intended to discriminate against any portion of the public. The FCC also prohibited disparate discrimination, which measures discrimination by looking at the results of ISP practices in the market rather than trying to judge the intentions of ISPs.

As was expected for almost all FCC orders these days, ISPs quickly banded together and sued to stop the FCC order. The U.S. Chamber of Commerce brought the first suit in the Fifth Circuit Court of Appeals in January on behalf of big ISPs like AT&T, Charter, Comcast, T-Mobile, and Verizon. Twenty industry groups like NCTA, WISPA, ACA Connects, and US Telecom entered the fray, and the suits were eventually joined into one case in the Eighth Circuit.

The ISP industry threw a number of arguments against the wall, hoping one would stick. The primary complaint was that Congress didn’t intend to impose a disparate discrimination test, and that disparate discrimination is rarely applied anywhere in the regulatory world. The ISPs also argued that the FCC violated the major questions doctrine with its ruling. This concept was based on recent Supreme Court rulings that prohibit federal agencies from adopting regulations that have “vast economic and political significance” without clear authorization from Congress. ISPs argued that the FCC went further in its discrimination rules than was specifically authorized by Congress. Finally, ISPs said the ruling was too widely applied, and should only have been applied to last-mile ISPs, while the FCC rules applied to a wider market, such as MDU owners who provide broadband in their buildings.

The Court made a number of rulings. It said the FCC had overreached its authority since Congress had not explicitly allowed a disparate discrimination test. The Court also ruled that the FCC had exceeded its authority by applying the discrimination rules to entities other than last-mile ISPs.

The Court completely vacated the FCC’s 2023 discrimination order, which means it is the same as if it didn’t exist. The FCC is still obligated by the IIJA to implement discrimination rules, so we can expect the FCC to restart the process. Any new FCC proposed rules will undoubtedly acknowledge the Court rulings.

The natural question to ask is what the court order means in the market. There should be little immediate impact since the FCC’s 2023 discrimination rules never went into effect when they were immediately challenged in court.

There will be repercussions since the Court considerably weakened the 2023 FCC order by only allowing the intentional discrimination test. It seems likely that it will be nearly impossible to prove that an ISP intentionally discriminated against some subset of customers. The proof would have to be some sort of written documentation or public statement that proves the ISP’s intention to discriminate. The Court eliminated the disparate discrimination test, which is basically an “if it quacks like a duck, it is a duck” test. That’s the kind of test that has routinely been applied to housing discrimination complaints, as was ordered by the Fair Housing Act. The revised rules will also let landlords off the hook since they will not be subject to broadband discrimination complaints.

The Explosion of Machine-to-Machine Traffic

Fierce Networks recently reported that on the earnings call for the first quarter of this year, CEO Justin Hotard of Nokia said that machine-to-machine (M2M) Internet traffic will explode, and in a few years will become the largest source of data transmitted across the Internet. He said that about 20% of all network traffic today, about 80 exabytes, comes from machine-to-machine traffic, and that alone is big news. Nokia is betting its future growth will come from meeting this growing demand.

I went back and looked at older blogs, and in 2020, there were estimates that between 3% and 5% of Internet traffic came from machine-to-machine traffic. In 2020, it was estimated that video streaming accounted for 80% of the traffic on the Internet. I remember a lot of discussion around the industry at that time asking why we were spending so much money to expand broadband networks to support entertainment.

How did M2M traffic grow to be such a big part of the Internet today? There are a lot of different sources of M2M traffic, a few that should be familiar to most of us, but some that are not obvious:

  • Software Driven Usage. Compared to 2020, much of the software we use in homes and businesses has moved to the cloud. It’s not M2M traffic when you save a Word file in the cloud, but it is M2M traffic when the software you use communicates with the cloud behind the scenes.
  • Internet of Things. Consumer Affairs recently estimated that the average U.S. home now has 21 connected devices. That’s double the number of average devices from 2020. A large percentage of these devices transmit data to the cloud without the homeowner’s knowledge.
  • Industrial IoT. Businesses are a big source of M2M traffic. This includes video surveillance cameras and burglar alarms. It includes sensors of many kinds. Industrial IoT includes traffic involved with reading RFID tags, barcodes, and robotic scanners to track inventory and materials.
  • Automative. I think most people would be amazed at the large amounts of data transmitted from your vehicle if you have it connected to your home WiFi. This includes monitoring the sensors in the car as well as transmitting performance statistics every time you drive the car.
  • Healthcare Monitoring. There is a growing use of health monitors that track devices like pacemakers and glucose monitors. Patients are routinely being monitored after having medical procedures.

While the traffic from all of these sources is growing, the huge M2M growth predicted by Nokia will largely be driven by the use of AI. AI is currently driving a huge amount of new traffic. A lot of it is coming from AI datacenters that communicate with each other. There is also huge traffic growth coming on the edge from uses like AI-driven factories and other industrial uses that are using software to replace people.

Nokia sees its own growth coming from supplying the gear needed to provide faster transmission speeds on long-haul fiber and keeping up with the growth to power the growing number of long-haul fiber routes being built. Nokia also expects to see growth at the edge as the need to be able to process huge amounts of M2M traffic keeps increasing.

Interestingly, there is a big push among the companies that operate data centers to create a parallel long-haul network that segregates AI traffic from everything else. This is being done to improve latency by eliminating contention with other traffic and by eliminating intermediate traffic switching points. Data centers also want to use fiber routes with the fastest speeds and largest capacity.

Who Needs Moore’s Law?

I ran across a reference to Moore’s Law the other day. This was named after Gordon Moore, an engineer who later became one of the founders of Intel. In 1965, Moore observed that the number of transistors that could be squeezed into a given area of a circuit board was doubling every two years. He predicted this trend would last for perhaps another decade, but the microchip industry kept fulfilling his prediction for over fifty years, and the general consensus is that the Moore’s Law prediction died somewhere between 2016 and 2018. Chip density has continued to improve, but at a slower rate, and there is general consensus that we are getting close to reaching the maximum possible density of transistors, limited by the law of physics.

For a number of years, there were a lot of predictions that the end of Moore’s Law would mean the end of faster computing. For decades, our devices became obsolete every few years as the next generation of faster chips hit the electronics market. But chip makers have discovered a wide variety of ideas and technologies that continue to improve the speed of computing.

Consider some of the techniques that continue to improve computing power:

  • ASICs (Application-Specific Integrated Circuits): These are specialized chips designed for one specific task. For example, there are ASIC chips in data centers that are specifically designed to handle specific tasks related to processing masses of data.
  • GPUs (Graphics Processing Units): These chips take the opposite approach of ASICs and are designed to handle parallel tasks and tackle multiple functions at the same time. First designed for gaming, a GPU breaks a task into many independent threads and executes each thread using a different small core. GPUs keep improving as chip makers and software designers find better ways to coordinate and control the many threads.
  • 3D Stacking. Another new technique is stacking transistors and memory vertically as well as horizontally. This provides a huge boost to computing power. One of the more widely used 3D stacking technique is the use of chiplets, which are small chip components that can be stacked and fused to create a multi-layer chip. Chiplets benefit from Through-Silicon Vias (TSVs) that enables fast low-power communications between layers.
  • Memory Integration. One of the biggest bottlenecks for any chip is the process of moving data into and out of the chip core during the computing process. Memory integration creates temporary memory directly on the chip to store data that is still needed for the specific calculations being handled. Pulling needed information out of the chip’s own memory bypasses the normal data transfer issue.
  • Optical Computing. Optical computing uses light instead of electrons to transfer data around a chip. The primary benefit of light computing is that different colors of light can be used to allow for multiple streams of data transfer at the same time, instead of the single stream that comes from electrons. I recently wrote a blog that talked about a technology that can generate multiple wavelengths of light directly on a chip, which eliminates the need for bulky external lasers.
  • Optimized Algorithms: Some of the biggest improvements in chip speed come from rewriting software to be “hardware-aware”, meaning it perfectly aligns with a chip’s architecture.
  • Reconfigurable Computing. This is an architecture where portions of the chip can be programmed to change function or spatial configuration during the computing process.
  • Quantum Computing. Quantum computing increases computing power using qubits and the principles of superposition and entanglement. Qubits exist in multiple states instead of the two states of 1 and 0 for digital computing, which provides an exponential increase in calculation power.

We’ve just barely begun exploring many of these ideas, and there are likely many breakthroughs still to come. Picture what something like a reconfigurable architecture using multiple colors of chip-generated light waves in a quantum computer might mean.