Industries / Technology / Telecommunications / Internet providers and rural broadband
Fixed wireless
Fixed wireless is internet sent by radio from a tower to an antenna fixed on a home or business, with no wire running to the building.
6
Milestones achieved, from 1992 to 2026.
3
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1992
Internet over unlicensed radio
On 9 May 1985 the FCC adopted rules that let anyone use spread-spectrum radios, which spread a signal across a wide band, without a license in bands set aside for industrial, scientific and medical (ISM) equipment. Michael Marcus, the FCC engineer behind the rules, noted in a 2004 talk that point-to-point links under them were "unexpected but permitted." In 1992 Brett Glass, an electrical engineer, used such radios to start LARIAT in Laramie, Wyoming, first as a nonprofit co-op, because the town had no fast internet outside the University of Wyoming. Harvard's Berkman Center, introducing a 2009 talk by Glass, called LARIAT the first terrestrial wireless internet service provider (WISP), and said WISPs then served about a million accounts.
Credit
Michael Marcus and the FCC; Brett Glass and LARIAT
2001
Licensed fixed wireless goes bust
Companies including Advanced Radio Telecom, Winstar and Teligent had paid billions of dollars at FCC auctions for LMDS and MMDS licenses, bands for beaming broadband by radio, so they would have spectrum to themselves. Advanced Radio Telecom filed for bankruptcy on 30 March 2001 and Winstar on 18 April 2001, and Teligent was close to running out of cash, RCR Wireless reported that month. The same report cited a Strategis Group count of at least 250 active operators using free unlicensed bands with cheaper, near plug-and-play equipment.
LessonOwning spectrum did not make the business pay, and small operators on unlicensed bands kept growing.
Credit
Advanced Radio Telecom, Winstar and Teligent; reported by RCR Wireless
2010
Empty TV channels, few takers
TV white spaces are the TV channels (470 to 608 MHz in the US) left unused in a given place, where signals travel far. The FCC finalized rules in September 2010 letting devices use them without a license after checking an online database of protected TV stations. Microsoft launched its Airband program in 2017, aiming to connect 2 million people in the US by July 2022, and in May 2021 said it had reached 1.5 million people; the broadcasters' trade group NAB counted only about 300 white space devices registered nationwide, down from about 800 four years earlier. In 2026 researchers at Iowa State University and Microsoft wrote that rural deployments "did not materialize," for lack of chipmakers, regulatory uncertainty and no protection from interference, and that use effectively ceased in the UK when its last database provider left.
LessonSpare spectrum went largely unused because the rules for using it were too uncertain for chipmakers to build for.
Credit
the FCC; Microsoft; Iowa State University
2018
Home 5G before phones
On 1 October 2018 Verizon started 5G Home, a fixed wireless service, in limited areas of Houston, Indianapolis, Los Angeles and Sacramento, promising typical speeds of 300 Mbps and peaks of 940 Mbps. It ran on millimeter-wave spectrum at 28 GHz, far above the bands phones then used, and on Verizon's own pre-standard 5GTF specification, ahead of its mobile 5G planned for 2019, Light Reading reported.
Credit
Verizon
2025
Measured against cable
Fixed wireless access (FWA) from mobile carriers uses spare capacity on their 4G and 5G networks to serve homes. Opensignal, which measures networks from tests run on users' devices, reported in October 2025 that T-Mobile's FWA scored 76.8% on its Consistent Quality measure, which tracks how often users have a connection strong enough for everyday tasks, against averages of 81.5% for cable, 65.7% for fiber to the node and 56.0% for DSL. It found FWA held 7.2% of broadband subscribers in markets with one FWA provider, 10.7% with two and 13.4% with three.
Credit
T-Mobile, Verizon and AT&T; measured by Opensignal
2026
Rural India on 5G radios
India's telecom regulator, TRAI, counted 12.94 million 5G fixed wireless subscribers at the end of June 2026, up from 12.73 million a month earlier, and 6.49 million of them, 50.20%, were in rural areas. It counts separately 4.93 million FWA subscribers on unlicensed band radio, 71.28% of them urban, all reported by Reliance Jio.
Credit
Reliance Jio and Bharti Airtel; counted by TRAI
Watch for
Gigabit radios for thousands
Tarana says its G2 fixed wireless platform supports up to 6.4 Gbps and 2,048 customers per site, combining licensed and unlicensed spectrum in the 3, 5 and 6 GHz bands, and calls it "fiber-class." Operators quoted at the launch spoke of selling 500 Mbps and gigabit plans.
Achieved whenan outside party measures gigabit plans delivered at busy evening hours to many homes on one G2 site.
Claimed by
Tarana, September 2025
Millimeter wave returns to cities
Verizon's chief executive, Dan Schulman, said Verizon plans a fixed wireless service capable of gigabit speeds in urban areas, on millimeter-wave spectrum, moving a heavy service onto spectrum he said is largely unused. Earlier in 2026 Verizon closed its purchase of Starry, which served apartment buildings by millimeter wave. Verizon had about 6.4 million fixed wireless subscribers at the end of September 2026 and said it would go "well beyond" its target of up to 9 million in 2028.
Achieved whenan outside measurement, such as Opensignal's or Ookla's, reports gigabit-class speeds for paying customers on the new service.
Claimed by
Verizon, October 2026
A tenth of fiber's cost
Carl Guardino of Tarana said fixed wireless can be built for "a 10th to a 20th of the cost" of fiber. Fixed wireless providers won about 11% of the locations in BEAD, the US federal broadband grant program, against about 67% for fiber, by Connected Nation's count as reported by Fierce Network.
Achieved whenthe per-location grants in states' approved BEAD awards, or an outside cost study, show fixed wireless built at that fraction of fiber's cost in the same areas, and the awarded networks pass their speed and latency tests.
Claimed by
Tarana, June 2026
Open
01
Check radio claims with lidar
A public test of fixed wireless coverage claims on the FCC's National Broadband Map. For every location a wireless ISP says it can serve in a county, check with lidar, laser scans of the ground and trees flown by aircraft, whether a radio path to any of the provider's towers within range is clear of hills, trees and buildings, and flag the claims with no clear path.
The briefCredit · unclaimed02
Count the empty TV channels
A public, measured count of unused TV channels (470 to 608 MHz) at many rural sites, set beside the channels that the FCC's white space rules would allow there, to show how much spectrum the rules leave idle where rural broadband could use it.
The briefCredit · unclaimed
Open 01 · the brief
Check radio claims with lidar
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public test of fixed wireless coverage claims on the FCC's National Broadband Map. For every location a wireless ISP says it can serve in a county, check with lidar, laser scans of the ground and trees flown by aircraft, whether a radio path to any of the provider's towers within range is clear of hills, trees and buildings, and flag the claims with no clear path.
Why it is still open
Fixed wireless is the most overstated technology in BroadbandNow's manual checks of nearly 65,000 addresses against the FCC map, at 58.4%. A 2025 UC Santa Barbara study tested fixed wireless claims against Ookla speed tests, not against terrain. The USGS describes lidar line-of-sight maps as a way to plan wireless coverage and says Cambium Networks uses its lidar for that, but a search in November 2026 found no public audit that tests map claims this way.
Built with what exists
USGS 3D Elevation Program lidar; PDAL to build surface models; the viewshed tool in GDAL or WhiteboxTools; the FCC map's provider availability downloads and CostQuest's location points under license, as the UC Santa Barbara team used them; tower sites from the FCC Antenna Structure Registration and providers' own tower maps; Python with GeoPandas.
First step
Pick one county with lidar flown since 2020, download its lidar tiles and the state's fixed wireless availability file, and run a line-of-sight map from one known tower.
Done looks like
For three rural counties, every fixed wireless location claim tested for a clear path (at least 60% of the first Fresnel zone, the football-shaped space around the direct line that a radio signal needs, unobstructed) to a tower within 15 km; the share flagged published with code and data; and 100 flagged and 100 unflagged addresses checked against the provider's own availability tool.
Open 02 · the brief
Count the empty TV channels
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public, measured count of unused TV channels (470 to 608 MHz) at many rural sites, set beside the channels that the FCC's white space rules would allow there, to show how much spectrum the rules leave idle where rural broadband could use it.
Why it is still open
The 2026 Iowa State and Microsoft study measured only inside its own 560-square-mile test area in central Iowa, where it found 90 MHz unused but the database allowed only 18 MHz for high-power use, and it notes that TV license renewals in 2028 to 2031 give a window to revisit the rules. A GitHub search in November 2026 found database clients and student detectors, but no multi-site measurement set compared with what the rules allow.
Built with what exists
RTL-SDR receivers, which only listen and need no license; rtl_power or GNU Radio to sweep the band; a GPS logger; the FCC's separation rules for white space devices in 47 CFR 15.712 and TV station records to work out the channels a database would allow; Python.
First step
Log 24 hours at five sites in one county with one receiver and compare the empty channels with those the rules allow there.
Done looks like
Ten rural counties in five states, 30 sites each, each site logged for 24 hours, and a channel-by-channel table of measured-empty against rule-allowed channels, published under an open license.