CBRS Has 420,000 Radios and Almost None of Them Are in Buildings

There is a number that gets quoted at every CBRS conference, and it is a good number: more than 420,000 Citizens Broadband Radio Service devices deployed in the United States. It sits alongside a second number that gets quoted rather less often: more than 1,000 operators using General Authorized Access. Put those two facts next to each other and something awkward appears.

Four hundred and twenty thousand radios divided by a thousand operators is several hundred radios each.

What several hundred radios per operator actually describes

Nobody builds a hospital with three hundred CBRS radios in it. Nobody builds a distribution centre with three hundred either — the published cost benchmark for a six-radio distribution centre network is about USD 90,000 installed, before devices, and a single-radio small warehouse network runs about USD 10,000. An operator with several hundred radios is not fitting out buildings. It is covering a service area.

That is what the CBRS deployment actually is. The OnGo Alliance’s own industry survey describes the installed base as spanning “private and fixed wireless networks” and lists fixed wireless access as one of three principal formats alongside neutral hosts and public networks. Fixed wireless access is a last-mile broadband technology. It competes with fibre and DSL. It has nothing to do with whether a building has coverage in its stairwells.

On the best reading of the available evidence, in-building private cellular accounts for something on the order of 1.4 per cent of installed CBRS radios — perhaps six thousand radios out of 420,000. That is an inference, not a published statistic; no one publishes the split. But the arithmetic points one way and the alternative readings require an enterprise deployment pattern that the cost data does not support.

This is not a failure. It is a measurement error being read as one.

The temptation is to file this as CBRS underperforming. That is the wrong conclusion, and it is wrong for a specific reason: CBRS was never underwritten as an in-building technology. It was underwritten as spectrum.

The Priority Access License auction that closed in August 2020 raised USD 4,858,663,345 in gross bids across 20,625 licences won by 228 bidders, roughly USD 235,000 per licence for an unpaired 10 MHz block in a band where the other 70 MHz is free to anyone who registers a radio. The bidders were buying coverage certainty in counties where they wanted to build networks. Wireless ISPs bought most of the licences. Wireless ISPs then built fixed wireless networks.

The deployment followed the buyers. What did not follow was the enterprise building, which was always the story told to justify the band in the first place.

The quieter number: the administrators are leaving

Underneath the radio count sits something more consequential, and almost nobody outside the spectrum engineering community tracks it. Every CBRS radio in the United States must register with a Spectrum Access System before it transmits. The SAS assigns the channel and the power at that location and can order the radio to stop. Without an administrator, the radio is a paperweight.

The FCC certified four SAS administrators on 27 January 2020: CommScope, Federated Wireless, Google and Sony. Since then:

• CommScope stopped operating its SAS on 1 April 2023. Its certification survived — renewed through January 2030 — but the service did not, and the company must notify the FCC thirty days before resuming.
• Google stopped accepting new CBRS customers on 10 June 2026 and will shut the system down on 10 June 2027. Existing users have to move their radios to another administrator before then.
• Nokia was certified in December 2025 and RED Technologies in September 2023, bringing new entrants in as incumbents left.

The Wireless Innovation Forum’s directory now lists five working administrators. The FCC said in March 2026 that six of the eight it has approved still operate. A vendor census compiled in October 2026 puts the working list at five and explicitly flags one as discontinued and one as exiting.

Read that sequence as a market signal rather than a technical footnote. The best-capitalised company in the world entered shared-spectrum coordination, ran it for six years, and decided to stop. Two of the four original administrators have now walked away from the service. The companies that remain are a spectrum specialist, a regional operator, a network vendor, a Finnish software firm and Sony.

Meanwhile, look at what happened to the coordination layer for the other unlicensed option. The FCC approved seven automated frequency coordination systems for 6 GHz in February 2024 — Qualcomm, Federated Wireless, Sony, Comsearch, Wi-Fi Alliance Services, the Wireless Broadband Alliance and Broadcom. Three of the seven are non-profit or standards bodies. The Wireless Broadband Alliance launched a commercial AFC service in May 2025 built on the OpenAFC open-source specification.

So in shared spectrum the administrative layer is consolidating; in unlicensed 6 GHz it is commoditising toward open source. Those are opposite trajectories, and they are being chosen by the same kinds of companies at the same time.

What is actually winning, and it is not a radio

If the question is “who is winning in-building projects”, the honest answer has nothing to do with which radio performs better. It is about which budget the project can be charged to.

A distributed antenna system is charged to construction, or to a compliance obligation, or to a mobile operator’s capex. A CBRS network is charged to a private-network project — a new line item that has to be justified, staffed and renewed. A 6 GHz Wi-Fi upgrade is charged to the IT refresh, which was going to happen anyway.

That last one is a structural advantage that no amount of engineering can overcome. On our build, a 150,000 square foot building costs about USD 240,000 for a cellular DAS at USD 1.60 per square foot, USD 105,000 for a public-safety system at USD 0.70, USD 60,000 for a four-radio CBRS network, and USD 68,400 for a 38-access-point Wi-Fi upgrade. The DAS costs three and a half times the Wi-Fi and is the only one of the four that a building code can force anyone to buy.

Which is why the fire code turns out to be the most reliable buyer in this market. Section 510.1 of the International Fire Code requires approved emergency-responder coverage in all new buildings at 95 per cent of general floor area and 99 per cent of critical areas at a delivered audio quality of 3.0. That is a specification with a permit attached. It does not have to win an argument.

The exit that settles the question

If you want one transaction that tells you what the professionals think in-building infrastructure is worth, it is not a spectrum auction. It is the USD 8.5 billion Crown Castle sale that closed on 1 May 2026, in which the fibre solutions business went to Zayo and the small-cell business to Arium Networks, an EQT fund, with more than USD 7.0 billion of the proceeds going to debt reduction.

A tower company with the best sites, the best fibre routes and the deepest operator relationships sold the in-building and small-cell complex to a private infrastructure fund. When the incumbent with the lowest cost of capital in the sector is the seller, the asset’s returns are below that cost of capital. Everything downstream — integrator margins, neutral-host fees, venue lease economics — inherits the conclusion.

What would change this

Three things, all visible from outside.

If a Priority Access licensee with spectrum in counties where it has no network starts selling in-building capacity priced against a DAS rather than against a carrier plan, the CBRS story changes and the sunk spectrum cost becomes an advantage. If a large venue chain commits to neutral-host CBRS at scale, the reference customer that has been missing for six years finally exists. And if the fire-code enforcement pattern shifts from complaint-driven to audit-driven, the public-safety line — already the second-fastest-growing part of the market at 9.8 per cent a year — becomes the fastest.

None of those has happened yet. What has happened is that the biggest single number in the CBRS story turned out to describe something other than what the story said it described, and that the administrative plumbing underneath it is being quietly dismantled by the companies best placed to judge whether it was worth keeping.

Wi-Fi is winning in-building projects. It is winning because the upgrade is already in the budget, because the containment already exists in the ceiling, and because nobody has to be persuaded that it is allowed.