Editorial Article
The DAS Market Is Paid For by a Fire Code and Costed by a Carrier
What the spectrum auction actually told us
The Federal Communications Commission’s Auction 105 ran from 23 July to 25 August 2020 and closed after 76 rounds. It offered 22,631 Priority Access Licenses in the 3.5 GHz band, seven in each of 3,233 county-based licence areas, each an unpaired 10 MHz block. Two hundred and seventy-one bidders qualified, 228 won something, 20,625 licences were won, and 2,006 were left with the Commission. Gross bids came to USD 4,858,663,345 and net bids to USD 4,543,232,339.
Two derived numbers matter. The gross bid per won licence was about USD 235,572. And 2,006 unsold licences is 8.86 per cent of the total, which tells you that in roughly one county in eleven, nobody would pay a reserve for ten megahertz of shared mid-band spectrum.
Five years later the Commission is still administering the consequences: in December 2025 it issued a notice on one winning bidder’s interim default payment obligation. A licence that was won at auction was worth less than the commitment attached to it. That is the most useful single fact about the indoor wireless market, because it says the constraint is not spectrum price.
Two engines that do not share a metric
Distributed antenna systems are bought for two reasons that are measured completely differently, and conflating them is why most DAS forecasts are neither right nor wrong but unfalsifiable.
The first engine is a code. NFPA 1225, the 2022 edition of the emergency services communications standard, contains the in-building public safety radio coverage requirement where a jurisdiction adopts it. Adoption is decided one authority at a time. UpCodes’ list of jurisdictions adopting the 2022 edition is short and specific: Delaware, Georgia, Clark County in Nevada and Las Vegas, with amendments in most of them. That is the whole point. Whether a building needs a public safety DAS is a permit-by-permit question, answered by the authority having jurisdiction, not a national rule.
A code-driven project is not discretionary. It is a condition of occupancy, it is scoped by a radio coverage test, and it is paid for by the building owner. A carrier-driven project is optional, is scoped by a business case, and is paid for by whoever can be persuaded to fund it. The first behaves like fire sprinklers. The second behaves like capex.
The number, and what is inside it
Our own figure for the global distributed antenna system market is USD 10.8 billion in 2025 rising to USD 22.6 billion by 2033, a stated CAGR of 9.7 per cent; the two endpoints compound at 9.67 per cent. The largest region is North America and the fastest-growing is Asia-Pacific, and active systems are the largest segment, which is consistent with everything below.
The composition question is more interesting than the rate. A DAS project’s cost stack is fibre, power, head-end equipment, remote units, installation labour, commissioning, and the coverage test that proves compliance. The radio is not where the money goes. It has not been where the money goes for a decade; the remote unit became a commodity while the fibre run and the head-end software did not. That is why the value in an active DAS sits with whoever controls the head-end and the building relationship, and why an antenna vendor cannot price on differentiation.
CBRS created a carrier-free alternative, not a neutral-host market
The original thesis for the 3.5 GHz band was neutral hosting: a third party builds coverage in a building, sells capacity to several operators, and the building gets a network nobody has to subsidise. The General Authorised Access tier was supposed to make that cheap, because GAA use requires no licence at all.
What actually happened is that GAA gave the enterprise a route to a private network that does not involve a carrier. If a warehouse can light up private LTE in shared spectrum with no operator in the value chain, it has replaced the neutral host with itself. That is a smaller market for DAS in unit terms and a better one for whoever supplies the head-end and integrates it, which is the same conclusion as the cost stack, reached from the other side.
The PAL outcome reinforces it. Twenty-two thousand six hundred and thirty-one licences, an average gross price of a quarter of a million dollars, and eight point eight six per cent unsold is not a market clearing at a premium. It is a market where the buyers who could use the spectrum cheaply did, and the rest of the map was left on the shelf.
The two things a supplier should track
First, the adoption map. Every jurisdiction that adopts the in-building coverage requirement converts a discretionary capex decision into a permit condition, retroactively, for a building that is already designed. That is the closest thing to a demand trigger this market has, and it is published in state and municipal code amendments rather than in a forecast.
Second, the secondary market in PALs. If licences trade below their auction price, the case for building neutral-host coverage weakens; if they trade above it, somebody has found a use that a national carrier did not. Either way it prices the alternative to a DAS, which is the only competition that matters.
What we will not pretend to know
We do not split this market into equipment and services, and the published figures that do should be treated carefully, because a DAS deployment is sold as a project and a project can be counted as equipment revenue, as an integration contract or as a lease. The same building can appear in three different totals. The 9.7 per cent is a market rate for a heterogeneous bundle; it is not a volume, and it is not a price index.
What is defensible is narrower and more useful: the demand is created by codes and by carriers, in that order of reliability, and the profit accrues to the parts of the system that a code requires and a carrier will not pay for.
Frequently Asked Questions
How much did the FCC's CBRS PAL auction raise?
Auction 105 closed on 25 August 2020 after 76 rounds with gross bids of USD 4,858,663,345 and net bids of USD 4,543,232,339, across 20,625 licences won by 228 bidders. Of 22,631 licences offered, 2,006 were left with the Commission.
Is in-building public safety radio coverage required by national code in the US?
No. NFPA 1225 contains the requirement, but it binds only where a jurisdiction adopts it. The list of jurisdictions adopting the 2022 edition is short - Delaware, Georgia, Clark County and Las Vegas in Nevada, most with amendments - so the obligation is decided permit by permit.
What is the distributed antenna system market size?
Pheonix puts the global market at USD 10.8 billion in 2025, rising to USD 22.6 billion by 2033 at a 9.7 per cent CAGR; the endpoints compound at 9.67 per cent. The figure bundles equipment, head-end software, installation and integration, so it is a market rate rather than a volume or a price index.
Does CBRS General Authorised Access replace distributed antenna systems?
It replaces the neutral-host business case more than the technology. GAA lets an enterprise deploy a private network with no carrier in the value chain, which removes the third party that a neutral-host DAS depended on for revenue.