For the longest time, US Internet users relied on cable modem and DSL for their connections. Cable had a big advantage in speed, where it emphasized downloads over uploads.

The tables have turned and the cable industry is in bad shape, losing 431,000 subs in Q4 2024 per MoffettNathanson. This the fifth consecutive quarter of record subscriber loss with no apparent end in sight.

Meanwhile, competitive services Fiber to the Premise (FTTP), Fixed Wireless Access (FWA), and Low-Earth Orbit (LEO) satellite broadband are adding customers: FWA added 924,000 subs and FTTP added 558,000 during this quarter. LEO is still relatively insignificant, with less than two percent of the market, but we can expect a growth spurt thanks to the pending reset of the BEAD program.

Shifting Usage Patterns

Not only are we using different networks, we’re also using different devices in different ways. Since 2017, mobile devices have dominated, with more than 95% of Internet users owning smartphones worldwide. But more than 60% of Internet users also use desktop and laptop computers.

Traffic volume continues to grow, but the rate of increase in download traffic has slowed to 8.6%. The rate of increase in upload traffic is impressive, currently at 14.6% and rising. The average upstream connection speed is now 32 Mbps, out of reach to most cable broadband subscribers.

Cable’s declining numbers follow a period where its linear TV services have also taken a beating; they now lag streaming services in hours of use. The cable industry is trying to improve the quality of its offerings – especially on the upstream side – but so far these efforts are more press release than product.

Treading Water

Cable is seeking to buy time for product improvement by blocking FCC allocation of new mid-band spectrum for FWA and mobile licensing. One means of pursuing this tactic involves funding fake, ersatz Wi-Fi advocacy groups such as Wi-Fi Forward and Spectrum for the Future.

Recent figures suggest slower growth in FWA as mobile networks inch closer to capacity. FWA tends to piggyback on mobile networks when it moves into new neighborhoods, but it has to carry a lot more data than smartphone service with limited ability for offload.

It’s possible to increase mobile network capacity by sectorizing and building new cells, but this costs twice as much as adding new spectrum to existing infrastructure. Given the public’s expressed desire to ditch cable for FWA, there is an obvious consumer benefit in accelerating the FWA buildout. Increased consumer choice also blunts the significance of net neutrality, an option that the courts disfavor.

Bundle of Contradictions

Winning in Wireless, a recent report by Spectrum for the Future, demonstrates cable’s desperation to halt the advance of FWA and FTTP. The report rejects arguments for auctioning full-power licensed with a series of objections that apply equally to its preferred networks, Wi-Fi and CBRS.

It then tries to show that these objections don’t really apply to its favorites for the lamest of all possible reasons. Not surprisingly, WiW was written by lawyer Jennifer M. McCarthy rather than a team of engineers and economists.

It closes with the now common refrain that unlicensed spectrum is the “workhorse” of the Internet despite the fact that Wi-Fi doesn’t actually connect devices all the way to the Internet. Conveniently, cable modem does this very thing.

Out of Date or Fully Modern?

 McCarthy claims the US is stuck in a spectrum policy playbook from the 1900s rather a shiny new approach suitable for 5G and Wi-Fi 7. In reality, the US follows a very sophisticated scheme, the Simultaneous Multiple Round Auction (SMRA), developed by Stanford game theorists Paul Milgrom and Robert Wilson along with the FCC’s Evan Kwerel.

Milgrom and Wilson won the Nobel Prize in Economics in 2020 for this system. Wi-Fi spectrum is assigned by agency fiat, the exact system that was fashionable in the era of the Radio Act of 1912. The licensed by rule feature is relatively novel, dating back to the 1985 Spread Spectrum Order, but its main effect is to take spectrum licenses off the table. 

Unlicensed spectrum is only applicable to a limited set of use cases in very small networks where interference isn’t much of an issue. As the paper shows, large enterprise networks work better under the CBRS licensing regime with private 5G than under unlicensed.

Statistical Sleight of Hand

Leading nations have assigned more mid-band spectrum to 5G than has the US. As Accenture puts it in its report The Looming Spectrum Crisis, the gap is serious and growing:

In 2023, the U.S. trailed five leading nations – Japan, UK, France, China, and Saudi Arabia – by an average of 202 MHz in available spectrum. This deficit is expected to more than double by 2027, with the U.S. expected to trail five projected leading nations – China, Japan, UK, South Korea and Saudi Arabia* – by an average of 520 MHz of available spectrum. [*footnote: By 2027, several countries (listed in order of projected spectrum advantage) with more licensed mid-band spectrum than the U.S. are projected to be China (1660 MHz), Japan (1100 MHz), the UK (790 MHz), South Korea (700 MHz) and Saudi Arabia (600 MHz).]

McCarthy seeks to sweep the spectrum deficit under the rug by adjusting the US and China spectrum allocations by national population, as if spectrum can’t be reused from city to city. China’s population is four times that of the US, hence China would have to allocate four times as much midband spectrum as the US to be on par by McCarthy’s analysis.

But that’s not the way spectrum use works. Wireless signals have limited range; hence frequencies can be reused by antennas sufficiently isolated from each other so as not to cause interference. There are many TV stations on UHF Channel 2 in different parts of the country.

Similarly, lots of Wi-Fi access points use channel 6 in the 2.4GHz band and many use channel 40 in the 5GHz band without interfering with each other. This is Spectrum 101 but the insight appears to be lost on McCarthy.

Geopolitical Realities

Additionally, the China system for spectrum allocation is quirky, reflecting the nation’s political and economic structure, so it’s not directly comparable. But Japan, Korea, and the UK are very clearly relevant and we lag behind all of them by considerable margins.

The FCC had high hopes for 5G mmWave spectrum that proved too expensive to deploy at scale and also for the clever but clunky CBRS system with its very limited range. Our military also uses much more midband for two of its three radar bands than do our NATO allies.

These are the policy errors from which we need to recover. The solution lies in reducing the footprint for radar systems in the C Band and other parts of the mid-band. Radar is a very flexible system that can easily use frequencies that are too low or too high for mobile broadband; indeed, it does so already.

Reuse Can Be Costly

After denying the possibility of spectrum reuse across the broad geography of nations, McCarthy abruptly reverses course and mandates it at the neighborhood level. She argues that a broader portfolio of spectrum does nothing to relieve congestion in cellular networks but reuse via densification (AKA sectorization) solves the problem:

Significantly, increasing capacity in high demand areas does not require more exclusive licensed spectrum for the nation’s largest carriers, but it does require those carriers to increase their investments in network densification. In other words, use the spectrum they already have, but do it more efficiently and effectively…By serving smaller coverage areas located closer to subscribers, small cells can reuse the same frequencies more often, increasing network capacity without radical increases in spectrum allocation.

This is hardly news to the wireless industry. By 2024, wireless operators had deployed 432,469 outdoor cell sites in the US, a 24% increase over the previous year. And these new sites aren’t just single network facilities, they’re increasingly shared under colocation agreements.

An additional 802,500 indoor small-cell nodes supplement outdoor deployments. These deployments are costly, running some $30B per year and adding up to $190B since the advent of 5G. The typical macrocell costs $200K and can be as expensive as $1M.

Reuse Can Also be Cheap

If we knock three zeros off the cost of a macrocell we’re in the realm of the typical state of the art Wi-Fi 7 router. Yet McCarthy insists that the only way to increase Wi-Fi coverage and performance is by allocating more unlicensed spectrum:

To maintain our leadership position in future generations of Wi-Fi technology and accommodate ever increasing Wi-Fi demand, U.S. policymakers should make additional spectrum available for unlicensed use in the adjacent 7 GHz band, which would generate a projected additional $79.62 billion between 2025-2027.

How does that follow? While Wi-Fi is simpler and less sophisticated than 5G, it’s certainly capable of modulating its power levels and seeking out unused channels, the essential elements of densification. Consumer-grade Wi-Fi mesh systems avoid interference by using just these practices.

Instead of relying on one powerful router to cover the entire home, heavy Wi-Fi users install multiple access points, generally one in each room where Wi-Fi use is intense. These satellite nodes are tied together by Ethernet (or even wireless in some cases) and managed automatically by vendor-supplied software.  

What makes more sense, demanding carriers install more $10-200,000 cells or asking Wi-Fi users to solve their performance problems with a couple of $75 mesh nodes?

U.S. Leadership in Wireless

The paper frequently touts US leadership in wireless. Indeed, the US has pioneered some important aspects of spectrum allocation policy: auctions (both one-sided and two-sided); licensing by rule (AKA, unlicensed;) and multi-tiered licensing (CBRS.) But other countries can and do utilize these mechanisms.

Wireless networking is a global business subject to global regulation. American companies have often been at the forefront of wireless semiconductor chip production: Qualcomm dominates the market for high-quality 5G and Wi-Fi chips, Intel is a major player in consumer-grade Wi-Fi, and Broadcom is major mover of Wi-Fi access point chips.

But foreign firms MediaTek, Samsung, and Huawei control significant market share in both 5G and Wi-Fi. The CBRS chip market is currently dominated by US companies Sequans and Sierra, but their ability to excel isn’t assured outside the US. CBRS is largely used for private 5G networks, which suggests that it will ultimately become a feature of 5G chips rather than a distinct product. The standards and patents for wireless chips are international productions, unlikely to be dominated by any one nation for very long.

The aspect of leadership that means the most is the deployment and use of advanced networks. This is where economic value comes from, and deployment is heavily influenced by government spectrum policy. Wi-Fi has access to more spectrum in the US than in Western Europe and East Asia, but the economic value of Wi-Fi is hard to assess.

SFTF and Wi-Fi Forward engage in the conceit that Wi-Fi is just like 5G, but by range it’s simply a local area network alternative to Bluetooth and Ethernet. When Wi-Fi becomes congested, consumers simply connect their video streamers to their home routers over Ethernet. When a wide area network such as 5G FWA or cable modem congests, the solution is a lot more complicated…and expensive.

Private 5G over CBRS May Accelerate Economic Growth

While Wi-Fi is predominately an entertainment network, 5G and CBRS address the needs of manufacturing and other business applications. They harken to the early years of IEEE 802 local area networks when manufacturing was a major target and General Motors and Boeing were major players. Manufacturing networks need limited latency, a need that Wi-Fi has never been able to meet. They also need mobility, a need that 5G was designed to provide.

Hence, private 5G networks over CBRS are emerging as major market opportunities. CBRS is poorly understood. Its most important feature is coordination, another property that unscalable Wi-Fi fails to provide. The CBRS Spectrum Access System (SAS) is a control point that controls frequency assignments, power levels, environment sensing, and interference protection in addition to authorization. A similar system should have been made a mandatory part of unlicensed in the beginning, but unlicensed was too speculative in 1995 for much forecasting and wireless LANs were several years away.

CBRS is one answer to the question of what regulators would do differently if they were devising rules for Wi-Fi all over again. While Wi-Fi shares spectrum bands and makes users fight each other over every transmission opportunity, CBRS seeks to provide each user with clean spectrum in much the same way that exclusive licenses do. CBRS is not a system for dynamic sharing, it’s a way to find clean spectrum without ongoing contention. While Wi-Fi advocacy groups like to claim CBRS as near relative, it’s naturally more aligned with 4G, 5G, and the next G.

Hence, CBRS is a great (or very good) enabler of private 5G networks. Its economic potential is seen in making manufacturing and campus-wide networking more efficient and reliable than it could ever be over Wi-Fi. In reality, CBRS is the antithesis of Wi-Fi.

Unique Needs

SFTF becomes most creative when it endeavors to explain why the solutions it proposes for cellular networks don’t apply to Wi-Fi. As we’ve seen, they oppose assigning more spectrum to FWA in areas of congestion, but they somehow manage to convince themselves that Wi-Fi has a unique need for larger channels:

First, and most importantly, hundreds of megahertz in the 6 GHz band are occupied by incumbent systems, limiting the channels available for unlicensed devices. Second, the latest generations of Wi-Fi require access to much wider channels to deliver gigabit to multi-gigabit speeds and greater capacity for data-intensive applications; other ISM bands don’t have enough contiguous spectrum to meet those needs, making 6 GHz essential to Wi-Fi’s continued growth, reliability and success.

The first and second points contradict each other. Access to wider channels isn’t guaranteed when spectrum is shared with incumbent systems outside the control of Wi-Fi. But more important is the claim that Wi-Fi needs multi-gigabit speeds. Wi-Fi is, after all, an entertainment network whose most data-intensive application is video streaming at 4-15 Mbps. Wi-Fi advocates insist that 320 MHz channels are necessary when each one of them occupies more than twice the total spectrum allocated to the entire more-efficient-than-Wi-Fi CBRS system. Are we supposed to take this outrageous claim seriously?

Wi-Fi is the tail of a dog composed of multi-megabit Internet service and multi-gigabit Ethernet. As more homes become connected to the Internet through 5G and LEO satellites, the demand for ultra gigabit in-home wireless networking would be declining if it had ever existed. None of the champions of multiple 320 MHz channels has ever been able to prove that genuine demand for such capacity has ever, or will ever exist.

The Internet is a large, holistic system in which capacities across the whole system simply need to be in balance.     

The Final Contradiction

SFTF is desperate to prevent the mobile network from improving through greater access to spectrum. As we’ve seen, they’re not averse to self-contradiction. In the report’s final section, they offer a breath-taking bit of babble:

As U.S. policymakers consider how to make additional spectrum available for the myriad consumer and industrial bandwidth-intensive use cases that are only beginning to emerge, it is critical to recognize that the traditional model of full-power exclusive licensing does not fit the moment and comes at a huge cost to America. First, there’s no unused spectrum left. So tough choices will have to be made if bands are to be cleared and auctioned for exclusive use. Shared licensing regimes, such as CBRS or unlicensed, on the other hand, are a more prudent approach and allow for continued innovation and connectivity without putting existing uses at risk.

Setting aside the fantasy about “myriad consumer and industrial bandwidth-intensive use cases” that no one can name, I’m impressed by the cuteness of “fitting the moment” in an era of spectrum scarcity where there’s “no spectrum left.” SFTF has already told us that the entire 7 GHz band is “left” for further fattening of its members’ coffers.

Additional underused spectrum awaits us in the mid-band at 4 and 8 GHz; in the high-band from 9 to 13 GHz; and in the mmWave band. The National Spectrum Strategy has identified 3.1-3.45 GHz, 7.125-8.4 GHz, and 12.7-13.25 GHz for active study.  WRC-23 identified 10-10.5 GHz for mobile and WRC-27 will study 4.4-4.8 GHz, 7.125-8.4 GHz, and 14.8-15.35 GHz for use as mobile bands.

Efficiency is the Answer

We don’t maximize the value of increasingly scarce spectrum by using it inefficiently. The CBRS system was created by policymakers who realized that the CSMA/CA contention system employed by Wi-Fi was not and is not working. The number one requirement of network systems is reliability, an area in which 30 years of Wi-Fi has failed to make progress.

The FCC has taken a step in the right direction by requiring Wi-Fi to use Automated Frequency Coordination in the 6 GHz band for outdoor and high-power indoor uses. In principle, AFC is something like the CBRS SAS that can actively coordinate concurrent usage of adjacent bands. AFC is a great target for improvement.

CBRS is not a shared spectrum, low-power system like Wi-Fi. It is a near cousin of exclusive licensing that would be much more effective at higher power levels. The FCC can make this happen as well.

Doing Right By Wi-Fi

Finally, future Wi-Fi allocations need to switch from the mid-band to the high-band, above 15 GHz. Today’s Wi-Fi runs out of steam after 20-25 feet of propagation. Such a system does not need to use the spectrum most valuable for outdoor-indoor systems such as mobile and FWA. Squatting on the mid-band benefits wireline broadband suppliers desperate to stave off FWA competition, but it does nothing for Wi-Fi users. This issue is under FCC control as well and we’ll know it’s a wise move when we hear the cable-funded ersatz Wi-Fi advocates shrieking.

In short, the FCC should ignore the pleas of self-appointed Wi-Fi advocates who don’t have the best interests of Wi-Fi users at heart as it develops AFC. It should increase the power of the otherwise perfectly sensible CBRS system, and it should make additional spectrum available to mobile services by exclusive or CBRS-PAL-style auctions.