A New Voice Feature

T-Mobile is currently beta testing a real-time translation service for T-Mobile cellular customers. The service will offer translations between fifty languages. The company is touting this as the first real-time agentic AI platform used on a wireless network.

There are already a lot of translator services available today like Google Translate, JotMe, Wordly.ai, Maestro AI, and others. The advantage of the T-Mobile offering is that it would a built-in feature that comes embedded with cellular service – a device that billions of people carry around all day.

It will be interesting to see how the beta test goes, because the biggest challenge of any translation service is to be able to translate quickly enough not to introduce big pauses into a conversation. Failure to do that makes a conversation feel robotic. Meeting that kind of real-time requirement will require low latency on the network as well as software that can translate quickly somewhere in the backend.

This is the first significant new voice feature I can remember that has been introduced since talk-to-text was introduced by Apple Siri in 2011. This is an amazing use of AI. For Star Trek nerds like me, this is the first baby step towards a universal translator. This feature, if it works as promised, will make it lot easier for people around the world to communicate.

One of the best parts of this feature is that it’s not tied to having a T-Mobile smartphone that requires specific software. The translations are done in the cloud, and T-Mobile says this can be made to work on any phone used by a T-Mobile customer, including flip-phones.

I keep hearing that the telecom companies are integrating AI into their businesses. It’s easy to see the AI agents that are popping up on customer service screens. Most of the articles and reviews I read say that a lot of people are unwilling to interact with AI agents, and it’s going to be interesting to see how big companies react if their customers won’t use the AI tools the companies prefer.

Much of the AI being introduced by telecom companies is being done out of sight. Industry technical news keeps describing initiatives for network owners to use AI to better manage networks. I’ve written a few blogs about this topic, and I suspect that reliance on AI instead of experienced technicians is a contributing factor to the big national network and service outages and contributes to it taking longer than suspected to diagnose and clear problems.

If AI is going to win over a lot of people if it can be used for features that people want to use. In today’s world, a lot of people know people who aren’t conversant in English. An easy real-time translator service would quickly broaden the horizon for a lot of us.

It’s certainly a marketing coup for T-Mobile if this works and if it takes others a while to offer a competitive alternative.The most interesting question for me is what’s next – what other AI features are on the way?

The Rapid Evolution of Transport Lasers

The Internet in the U.S. relies on long-haul and middle-mile fiber routes that are used to connect every part of the country to the core internet hubs located in Virginia, Dallas, Chicago, Atlanta, Los Angeles. New York, and Denver. In more recent times, the growth of data centers has created additional major Internet hubs in places like Phoenix, Silicon Valley, Portland, and Seattle.

Like every other part of the industry, there has been a constant evolution in the lasers that were used to power the long-haul fiber routes. When I first got involved in working with companies providing transport in the early 2000s, the transport electronics delivered 1 GB (gigabit) speeds – something that everybody at the time thought was blazingly fast. Today, millions of homes are buying 1 GB broadband.

The Internet was exploding during the 2000s as millions of people started to buy broadband provided by DSL and cable modems. Gigabit transport routes became full, and carriers knew they had to upgrade. The IEEE standard for 10 GB transport was adopted in 2002, and over the next decade, it became the standard for transport fiber routes.

Of course, 10-gigabit transport routes grew full as growth continued, and carriers were looking for more speed. The IEEE standard for 40 GB transport was adopted in 2010, although a few vendors, like Nortel, had started to market 40 GB products as early as 2008. The biggest technical breakthrough for 40 GB lasers was the introduction of Digital Signal Processing, which better handled light dispersion across long-haul fiber routes. The higher speed became the industry standard for transport by 2012.

Next in the evolution were 100 GB lasers. This standard was also adopted by IEEE in 2010. This faster technology was slower to be adopted because of the relatively high cost of the lasers. By 2014, there were only about 600 deployments of this technology worldwide. But over time, 100 GB lasers became standard for anybody building transport fiber routes.

The next step in progressively faster lasers was 400 GB, with the IEEE standard adopted in 2017. Network owners started to introduce these faster lasers into networks in 2020, and by 2022, 400 GB lasers became the new standard for long-haul transport.

The general continuous growth of Internet traffic, and the new demand from AI, is pushing transport fiber owners to seek even faster lasers. A few vendors introduced 800 GB lasers as early as 2019. Ciena announced the 800 GB WaveLogic 5 laser in 2019, and Infinera and Windstream successfully tested a 800 GB long-haul route in 2020. While 400 GB lasers are still the most affordable option, Nokia and Ribbon say that they are now seeing a lot of demand for 800 GB lasers.

Ciena says it is seeing demand for even faster lasers and has installed a few fiber routes with 1.6 TB lasers for Lumen in the U.S., e& in the USA, and Cirion in Latin America.

The faster speeds are also moving down market into last-mile uses. Nokia is selling a lot of 800 GB pluggable fiber electronics for inside data centers.

This has been an amazingly fast evolution. As recently as 2019, almost everybody in the industry was still buying 100 GB lasers for transport, and in the few years since then, we’ve seen increases to 400 GB, then 800 GB, and now the beginnings of 1.6 TB. I remember seeing a PowerPoint at a trade show twenty or so years ago where a vendor claimed that within twenty years we’d be seeing terabit lasers. It was a bold prediction at a time when 10 GB lasers were cutting-edge technology, but it turned out to be a good prediction. I’m not even going to try to predict the speeds we’ll be seeing twenty years from now.

Technology Shorts March 2026

Today’s blogs looks at some of the recent breakthroughs coming out of labs and research facilities that could have practical applications that could eventually benefit the broadband industry.

Rainbow Chip. Researchers at the Columbia University School of Engineering and Applied Science have created a chip that turns a single laser beam into a “frequency comb” that produces dozens of light channels at once. As often happens in science, the breakthrough was discovered by accident when the team was working on a project related to Lidar.

Normal laser beams used in telecom are not precise and transmit a closely bunched group of similar light frequencies that scientists refer to as a messy light signal. This new chip creates multiple laser beams in a range of colors, with each beam precisely at a single light frequency. The chip output is called a comb because there is a clear gap between each different beam, so there is no interference between separate light beams. This chip could revolutionize fiber optic technology by simultaneously sending dozens of even-spaced light channels at precise frequencies through a single fiber, with no interference between colors. Scientists have created precise laser beams in the lab for research, but this chip could bring the technology into practical use.

Energy Efficient Wireless Chips. Researchers at the University of Colorado Boulder have developed a new device that could revolutionize wireless technology. The breakthrough is the creation of a surface acoustic wave (SAW) phonon laser that can create ultra-high frequency vibrations on a single chip. The new device layers silicon, piezoelectric lithium niobate, and indium gallium arsenide to amplify radio vibrations much like a diode laser amplifies light. SAW technology is already embedded in smartphones, GPS, and radar systems and is used to filter signals and reduce noise. However, today’s SAW technology  requires multiple chips and external power. The new phonon technology simplifies this to a single chip that can be powered by a battery. The new chip can also reach far higher frequencies and currently operates at about 1 gigahertz, but has a clear development path to boost this ability to tens or even hundreds of gigahertz.

Efficient Power Module. Researchers at the National Renewable Energy Laboratory unveiled a breakthrough that could squeeze more power from existing electricity supplies. They’ve created a silicon-carbide-based power module they call ULIS (Ultra-Low Inductance Smart). The ULIS device dramatically improves the way electricity is converted and delivered inside devices. Most electronic devices contain a power module, which houses the power electronics that regulate the flow of electricity inside the device. The ULIS device is smaller and lighter while bringing up to a five times improvement in power efficiency. The device would make sense in data centers, electric grids, and any devices using next-generation electronics, like in ships and aircraft. The secret to the success of the new device is that it slows parasitic inductance by seven to nine times, which is the resistance to the process of changing or converting an electric current inside a device.

Some of the benefits come from its new design. Traditional power modules stack components inside a box-like package, while ULIS has found a way to arrange components in a two-dimensional octagon. This creates a smaller light-weight device that also minimizes magnetic interference. One of the most interesting features is that the device can be controlled wirelessly, without needing to be connected to communications cables.

ULIS is expected to impact multiple sectors. Probably the most beneficial is in the electric grid. Today, the devices in the grid require electricity to be converted into a usable form before entering every smart device in the grid. The ULIS device could make this conversion more efficiently and with less power loss in the grid.

Cooling Data Centers with Hot Water. One of the biggest challenges of large data centers is having a large supply of cool water for cooling. At CES this year, Nvidia CEO Jensen Huang announced the company is using water at 45 degrees Celsius (113 degrees Fahrenheit) to cool supercomputers. This is a big breakthrough because hot water doesn’t require water chillers and the accompanying power-hungry compressors. Those devices account for about 6% of the power used at a data center. This breakthrough could be a boon for two-phase liquid cooling systems. Most liquid cooling systems today circulate water, which then must be cooled before reuse. A two-phase system extracts more heat from computers by using the heat to convert the liquid to a gas and then converting back to a liquid. This is not a new technology and has been used on a limited basis for a few years, but the NVIDIA announcement will prompt data center owners to consider hot water as the primary way to cool data centers. The announcement instantly tanked the stock prices of companies that make cool-water chillers for data centers.

Rising Costs of Broadband Construction

The Fiber Broadband Association, along with its consultant, Cartesian, published its annual report on the cost of fiber deployment. The report was compiled from online surveys and phone interviews conducted in September and October 2025.

The report includes a lot of interesting statistics and trends for anybody getting ready to build fiber.

  • FBA believes that 60% of the homes in the country are now passed with fiber. There were 11.8 million fiber passings built in 2025, including 8.1 million to new locations.
  • The median price of fiber construction that was garnered from the survey was $18.00 per foot for buried construction and $8.00 per foot for aerial construction. That was a 3% increase in cost for buried construction and a 14% increase for aerial construction.
  • Labor is still the largest component of cost. Labor represented 72% of the cost of buried construction and 64% of the cost of buried construction.
  • The method of underground construction matters. It cost 40% less to directly bury fiber compared to the cost of putting fiber into new conduit. Trenching had the highest cost, at 60% more than plowing and 6% more than directional boring.
  • 92% of survey respondents saw higher costs for fiber construction in 2025 compared to 2024.

The most relevant news from the report is that 88% of the respondents expect a cost increase for construction in 2026. 62% of respondents expect a ‘slight’ cost increase of less than 10%. 26% expect a cost increase of more than 10%. 9% expect costs to stay the same, and 3% expect costs to decrease by less than 10%.

Nearly two-thirds (62%) of respondents said they expect fiber deployment costs to rise “slightly” (less than 10%) compared to 26% who expect a “significant” increase of 10% or more. About 9% expect costs to stay the same, and 3% expect costs to decrease by 10% or less. Considering all responses, 88% of respondents expect costs to increase again in 2026.

Note that the cost increases experienced in 2025 and the expected increases expected for 2026 are national averages. Responses to the survey came from 38 different states, and there was a fairly even split between urban, suburban, and rural fiber builders. National averages blur the significant differences in the cost of labor in different parts of the country. National averages also even out the differences between the big fiber builders who build huge numbers of miles of fibers, and smaller ISPs building smaller projects. The labor costs in a national survey also include projects built with prevailing wages, projects built at union wages, and projects built at market non-union wages. The averages also blend the costs from ISPs that build with employees versus those using contractors.

I have to wonder if all of the folks who plan to build fiber with BEAD grants have built in a possible cost increase for fiber construction of 10% per year. That would mean that fiber built in the fourth year might be 33% more expensive than what was built in the first year.

The report is well worth reading for anybody planning to build fiber because it’s full of interesting statistics. For example, an interesting statistic is that a buried construction crew of 3 can build around 1,000 feet per day, while a crew of 7 can build 1,585 feet per day. A crew of 3 can build 1,800 feet of aerial fiber per day, while a crew of 6 can build 4,000 feet per day. These statistics show there isn’t a big benefit from using large crews compared to using multiple smaller crews.

Another interesting statistic was drop costs. The average cost for a buried drop under 100 feet was $556, while the cost for a 500+ foot drop was $675. The average cost for an aerial drop under 100 feet was $338, while the cost for a 500+ foot drop was $400. These numbers also blend the cost of ISPs that build drops using in-house staff versus those using all contract labor.

New Mexico’s New Broadband Affordability Plan

The legislature in New Mexico approved a broadband affordability plan, which it labels as LITAP (Low-Income Telecommunications Assistance Program). This plan is intended as a direct replacement of the expired federal Affordable Connectivity Plan (ACP).

Like the ACP plan, the New Mexico plan would provide a $30 monthly subsidy to qualified households, with up to $75 for those living on tribal lands. The total subsidy is capped at $10 million in the first year, and up to $45 million in future years.

The plan will be funded by the existing State Rural Universal Service Fund (SRUSF), which is administered by the Public Regulatory Commission. This program has historically been funded by surcharges on customer bills for telephone, VoIP, or cellular service. The SRUSF is currently being used to subsidize telephone bills, carrier support for access charges, and for broadband programs to help bring broadband to rural areas. The current surcharge on customers is $0.61 per month for 2026, down from $1.13 in 2024. It’s estimated that when the LITAP plan goes fully into effect, the surcharge might climb to $2 per month.

One feature of the legislation is that this new broadband subsidy will supersede the existing subsidies for telephone service. The legislation requires any ISP that uses the revised program to participate in the Lifeline program. The federal plan currently provides a $9.25 monthly subsidy for eligible households, and up to $34.25 on qualifying tribal lands.

Households will be eligible for the subsidy if they participate in two New Mexico plans for need-based health care assistance, or for “at-risk” students in the home. LITAP will also be available for any customer who is eligible for the federal Lifeline program. When fully funded at $45 million, the new fund will subsidize approximately 120,000 households.

The legislation was prompted by a recognition by legislators that affordability is the largest barrier to residents buying broadband. This makes sense in the state since New Mexico is ranked 46th in terms of household incomes. The state has also consistently been in the top three of states with the highest level of poverty.

This is the first state plan I’ve seen that largely mimics the now-defunct federal ACP subsidy. New York has a plan that requires larger ISPs to offer $15 rates to qualified low-income households. Hopefully the New Mexico plan will be welcomed by ISPs in the State, particularly those who are building grant-funded networks. This plan will make it far easier to achieve needed customer penetration rates.

It seems unlikely that the new plan can easily be challenged in court. The SRUSF was created in 1999 and has been providing subsidies for telephone bills since its inception.

Perhaps this will prompt other states to do something similar. There are a lot of other states that already have a Universal Service Fund for telecom that is based on surcharges to customers.

A Telehealth Success Story

I’ve been hearing for years about how good broadband can open access to telemedicine. It’s certainly something that is increasingly needed as rural hospitals and clinics are closing across the country. But telehealth is not just a rural issue, and a recent article by Sean Gonvales points out a successful telehealth program in San Francisco.

The SF Tech Council, a collaboration of government and nonprofit entities, undertook an effort to train patients on how to use patient medical portals. Patient portals are online systems that allow patients to communicate with doctors, manage appointments, check lab results, or ask for refills on prescriptions. I’ve been using a portal with my own family doctor, and it’s a great tool for handling things that used to require phone calls or even a doctor visit. A well-run portal makes life easier for patients and physicians.

What the group in San Fransico realized is that a lot of patients were unfamiliar with portals and didn’t understand how to establish a portal connection or use a portal. Using funding from a State digital inclusion grant, the group created a hands-on training effort to help patients learn how to navigate a portal. In San Francisco, the training has been aimed at low-income patients, those with low computer literacy skills, or the many people in the community who speak Chinese or Spanish.

I found this story to be intriguing because it demonstrates the need for hands-on digital skills training. Portals are being increasingly used because they allow the entities we communicate with online verify who they are talking with. In addition to medical portals, portals are used to communicate with an increasing number of federal, state, and local government agencies like FEMA and Social Security. Portals are being used by banks, credit card companies, mortgage companies, and student loan processors.

The folks in San Francisco realized that people were leaving a doctor visit with instructions to follow up through the portal, but then never made the digital connection. The grants funded the ability to establish a hands-on session with a digital navigator who walked people through using a portal in a 45-minute training session.

The feedback was that nearly all participants in the training walked away knowing how to establish and navigate a portal. It’s likely that this training will also give folks more confidence in using the increasing number of other kinds of portals. The feedback was that once people knew how to use a portal, they saw the benefits and were eager to use it.

Every time I read about one of these success stories, I’m saddened that the federal government backed out of funding for digital skills training. This particular success story comes from a State grant, and States can certainly fill in some of the gap created by the cancelled federal funding. However, the federal grant funding for these kinds of training efforts was only going to last for a few years, so there was already going to be a need to establish ongoing programs for these kinds of training efforts.

One thing that the opportunity for federal funding for digital inclusion did was to get a lot of people interested in establishing local digital equity efforts. Hopefully, this momentum will mean these efforts will move forward without the federal funds. At the local level, there are a lot of government agencies and nonprofits that might offer basic skills training like this one. Perhaps a coalition of doctors and hospitals might take on the role to make sure that patients know how to use the needed digital tools. It would make sense if insurance companies would embrace this – they all acknowledge that preventive care saves money, and making sure that patients follow-up visits through a portal is part of the preventive care process.

Broadband Shorts February 2026

The following are a few interesting stories that don’t warrant a standalone blog, but that I found interesting.

Amazon One Delays. Amazon asked the FCC for a two-year delay in meeting its commitment to launch half of its promised constellation of 3,236 satellites. The deadline for having more than 1,600 satellites in orbit was July 30 of this year. As of the date of writing this blog, the company only has 212 satellites in orbit. Amazon says it has contracted for ten additional 10 SpaceX launches to speed up its process.

This is bad news for homes where Amazon One was the BEAD winner, like a number of counties where I live in Western North Carolina. If Amazon follows the historical path of Starlink, it will need at least 1,600 satellites in orbit to provide reasonable service, and more is better. With no additional delays, the service likely won’t be available until at least late summer 2028.

In related news, the FCC granted Amazon One the ability to add 4,500 satellites to its broadband constellation. This increase would result in the following: 3,232 Generation 1 satellites, 3,212 second-generation satellites, and 1,292 satellites in a polar orbit.

NTIA Opposes Relaxation of BEAD Compliance Requirements. Starlink sent a request to numerous states asking for relaxed BEAD implementation rules, which included a request that Starlink would not have to conduct speed tests for BEAD-funded customers. NTIA quickly added a new item to its Frequently Asked Questions that states that, “no BEAD statutory requirements or other program rules may be altered by a BEAD subgrant agreement”.

Charter Grows Cable Subscribers. Charter surprisingly grew cable customers in the fourth quarter of 2025 after losing customers every quarter for years. I laughed when I read that, because I know how they did this since I am one of their new cable subscribers. We called and asked about upgrading to gigabit broadband and were offered a lower price than what we pay today for 500 Mbps, but only if we agreed to take cable TV. I am now a phantom Charter cable customer who will never watch any of the content. Like many millions of households, I am no longer interested in linear TV that forces me to surf through channels I don’t want to watch.

This is not the first time this has happened to me. Twelve years ago, I was forced to take cable TV from Comcast in order to get broadband. The settop box went into the closet the first day and was never plugged in. I have to assume both Comcast and Charter paid programmers on my behalf, which is an odd and expensive financial choice.

Rural Backhaul. WISPA, the trade association for wireless ISPs, complained to the FCC that Charter is refusing to renew contracts to provide backhaul to many of its WISP members. WISPs are reliant on having at least some of the towers in a network connected by fiber, and if companies like Charter stop selling backhaul, some WISP networks will be degraded or could even go dark.

Buying rural backhaul has always been a challenge in many parts of the country due to the small number of companies with fiber that are willing to sell bandwidth to other ISPs. Many rural ISPs are forced to pay far more per gigabyte to buy rural backhaul from telcos and cable companies. It’s hard to speculate why Charter would do this. The company might want to focus its sales and marketing efforts elsewhere. But it could also be for a more sinister reason if Charter is doing this to eliminate or cripple rural competition.

Comcast Class-Action Settlement. Comcast settled a class-action lawsuit and agreed to pay $117.5 million as a result of a data breach that released sensitive customer data. That may sound like a lot, but the settlement covers 31.7 million customers and works out to a settlement of $3.70 per customer before subtracting out the amount that will go to lawyers. This settlement doesn’t place much value on customer privacy and data.

Broadband Grant Deadlines

The industry and the press have been laser-focused on BEAD grants for the last few years. It’s easy to forget that there are a lot of federal broadband grants that have been issued under other grant programs, many of which are facing completion deadlines. This includes grants from programs like CAF II, RDOF, ReConnect, the Capital Projects Fund, ARPA grants funded through SLFRF, NTIA middle-mile grants, and the NTIA Tribal Grants.

A lot of projects under these grants have a deadline to be completed this year or in 2027. Today’s blog looks at the consequences of not finishing a grant project by the legislative deadline.

The grants that face the most immediate deadlines at the end of 2026 include State grants that were funded by the Capital Projects Fund or through SLFRF. These grants would have been awarded through individual State broadband grant programs, but the underlying money came from federal legislation. ARPA grants might have also come from counties or cities. These grants have a hard completion deadline of December 31 of this year.

I’ve been hearing from State Broadband managers that the Federal government has no appetite for any extensions of this funding. This is not news, and they’ve been saying the same thing for the last year, but lately, they have been reminding states of this.

It’s been routine in the past for ISPs to get extensions for grant construction as long as they had a good story of why they needed the extra time. In today’s environment, there are a lot of reasons why construction might be delayed. It could be due to challenges in getting permits, rights-of-way, or easements. It might mean having problems getting onto poles because of recalcitrant pole owners. It might be due to supply chain problems in receiving needed materials. It might come from labor shortages that slow construction vendors. If the federal agencies behind the grants enforce the legislative deadlines, none of these excuses will matter.

Grant recipients need to do everything in their power to finish construction this year. With no extensions, any work done after the deadline will not be reimbursed. I’ve been hearing rumors that failure to complete a federal grant program on time might also make an ISP ineligible for any future federal broadband funding.

If hard deadlines are enforced, this could also be coupled with a hard deadline for submitting grant paperwork quickly. Most grant programs in the past have allowed grantees some leniency after construction is completed to submit invoices. Hard deadlines might result in rejection of invoices sent after the deadline. Most grants also require some kind of close-out reporting to document that the project construction is completed, and ISPs need to get any such report completed quickly.

Anybody working with a State Broadband Office needs to understand their specific requirements. For example, there might be states that want paperwork submitted and completed by the legislative deadline, meaning construction has to be completed even earlier than the end of the year. ISPs also need to push vendors to invoice for the grant-funded projects quickly to provide proof of the spending.

I am positive that some ISPs will be surprised if they don’t get reimbursed for work they have completed. Grant offices have sent out deadline warnings for years, but have often given routine exceptions for lateness. It sounds like, starting this year, that deadlines are firm with little or no room for exceptions.

Cellular Backup for Broadband

Amazon announced a new device that offers broadband backup for broadband outages. The new device is the eero Signal, which is an add-on to Amazon’s eero WiFi system that will automatically connect a customer to cellular broadband any time the primary broadband connection fails.

For now, the device works with 4G and retails for $99.99. For the purchase price, a customer gets the device plus six months of backup service. After that, there is an annual $99.99 subscription. A 5G version of the products is coming soon that will have a $199.99 annual subscription fee. In both cases, the customer also has to have an eero subscription, which can cost as much as $9.99 per month or $99.99 for a year.

The Signal device will connect with the strongest carrier signal in your area, meaning that Amazon has made arrangements with AT&T, T-Mobile, and Verizon. There are usage limitations, and the cellular connection can’t be used as a primary broadband connection. I’m curious how long the device will function for somebody who has a multiday outage. One of the other caveats of the device is a warning that customers can’t connect to 911 through the device.

Amazon is clearly counting on folks willing to pay extra for instantaneous broadband backup. The service seems pricy to me. I have several broadband outages every month on my cable broadband connection, and most last for 15 minutes or less. If I’m working on my computer,  I can activate my cellular hotspot within a minute if I really need an Internet connection, but I often just wait for the outage to clear. But I’m not the target customer for the product since my quick fix is only for the computer I am using. I don’t operate any critical devices on WiFi, like a security system, fire alarms, or a health monitor. The biggest draw of the service for folks with those devices is that the eero Signal product will support WiFi for all of the devices in the home, including when the customer is away from home.

In December, AT&T also announced a new broadband backup service. This is a free product that offers cellular backup to customers who buy both AT&T broadband and cellular service. This backup also provides cellular broadband for the whole home, but only if the activated cellphone is within range of the home Internet gateway. The broadband gateway will instantly start using cellular broadband if the normal broadband connection is lost. One of the best features of this service is that it requires no new hardware or a subscription. But this wouldn’t provide backup for critical devices if the customer isn’t home or in range with the cellphone.

Both companies obviously think there is a need and a market for a backup service. AT&T is oddly acknowledging that its broadband has enough outages for this service to be an issue. I’m going to guess that Amazon isn’t going to be the only hardware company to offer this. The idea of creating a recurring revenue stream has to be attractive to folks who normally sell hardware.

Broadband Usage in 4Q 2025

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 per household in gigabytes. Below is the trend in average monthly U.S. download and upload volumes since the third quarter of 2021. These averages include broadband used by residential and small business customers.The average U.S. broadband customer used 59 more downloaded gigabytes and 10 more uploaded gigabits per month than a year earlier. This growth means continued pressure on broadband networks because if we assume roughly 120 million broadband subscribers nationwide, this growth means over 8.3 billion more gigabytes of data cross the Internet than a year earlier.

As can be seen in the table above, upload usage has been growing at a faster pace than download usage. In a recent quarterly report, OpenVault credited the growth of upload usage to the increasing usage of video calls, cloud backup, IoT uplinks, and similar uses. To put the 10-gigabyte increase in average upload into context, it’s the equivalent of every household uploading an additional 7 standard definition movie files or 3 high definition movie files every month compared to a year earlier. I think the average household would be surprised by the volume of data they are uploading each month.

The report made an interesting comparison between HFC technology and fiber for an unnamed cable company. When comparing customers that subscribed to equivalent fast download speeds, the average customer using cable technology uploaded 58 gigabytes, while the average fiber customer uploaded 93 gigabytes. Clearly, the slow upload speeds on cable (an average of 17.3 Mbps) are restricting uploading compared to fiber customers with symmetrical data plans.

Another interesting statistic is the percentage of U.S. subscribers at different speed tiers. For the last several years, there has been a steady migration of subscribers from slower speed tiers to the fastest tiers, and there is continued erosion in customers subscribing to speeds under 100 Mbps. What’s most interesting about the two years is a big jump in subscribers in the 200-499 Mbps tier and a decrease in subscribers buying speeds faster than 500 Mbps. My best guess is that this is reflecting the continued migration of millions of homes to FWA wireless. Those households seem to be willing to accept slower speeds as a trade-off for the lower prices.

One thing this table demonstrates is the absurdity of the FCC’s current definition of broadband at 100/20 Mbps. At the end of 2025, 87.4% of U.S. broadband subscribers are buying a product with speeds of 200 Mbps download or faster.

OpenVault always includes other interesting statistics in its quarterly reports:

  • The annual average increase in usage is growing over time. In 2022, households used 50 gigabytes more than the previous year. That’s increased to 54 GB in 2023, 57 GB in 2024, and 69 GB in 2025.
  • Super Power Users (those that use more than 2 terabytes of data per month grew by 22.5% in 2025, to become 7.4% of all households.

Median usage grew from 461.2 GB in 2024 to 531.8 GB in 2025. The median is the number where 50% of customers use less and 50% use more. That’s an increase of 15.3%, which is a faster growth rate than for average usage (9.9%). OpenVault credits the faster increase in median speeds to faster growth than average for customers with smaller monthly usage.