Who pays for sensing, and who may

No carrier reports sensing revenue

The simplest test of whether a business exists is whether anyone reports it. On that test, sensing does not yet exist anywhere (Exhibit 1).

The one figure that gives a sense of scale in China comes from the press rather than from a filing. In April 2025 China Mobile said, as Xinhua reported, that it had “部署超过500个通感一体基站”, more than 500 integrated sensing base stations, covering 42 low-altitude routes and serving 130,000 industrial drones. Set that against the more than 2.4 million 5G base stations its 2024 annual report counts at the end of that year, and the sensing sites come to about two in every ten thousand. China Mobile’s own filings describe a technology system and trial networks, and nothing more: no sensing revenue, no contract, no site count.

No carrier has disclosed sensing revenue, and no priced sensing contract is on the record anywhere. Any sensing revenue in a model today is an assumption, and it should be labelled as one.

The buyers on the record are public bodies

Every sensing offer or trial on the record is a demonstration, a laboratory exercise or a subscription with no published price, and every named buyer is a state or a security agency (Exhibit 2).

In China the buyer is, in effect, the state’s airspace programme. The money flows through operator capital spending against a national priority, not through a sensing tariff, and four facts give a sense of the scale.

• Policy. The low-altitude economy has been in the national Government Work Report in each of 2024, 2025 and 2026, as Chongqing Daily recorded in March 2026.
• Guidelines. Caixin reported on 12 February 2026 that the Ministry of Industry and Information Technology and four other departments had issued guidelines requiring the operators to strengthen communication, positioning and surveillance below 300 m.
• Projection. The China Academy of Information and Communications Technology projects 360,000 new low-altitude base stations and RMB 111 billion of investment by 2035. That is a projection, not a contract.
• Cost. Chongqing Mobile says sensing cut its low-altitude network construction cost by 25%.

In the United States the first paid offer is a contractor’s, and no release names a price or says who pays whom. Light Reading’s report on NetSense is silent on both. AT&T’s demonstration with Ericsson in July set out a roadmap aimed at “public-sector stakeholders”. Ericsson Federal’s chief executive, Christopher Ling, told RCR Wireless in August that “ISAC will start as a private network [solution] just so we can make sure it works locally”. Ericsson’s release on its Finnish demonstration says early deployments “will likely focus on defense and public safety users”. The vendors, in other words, are saying the same thing in different words.

It is worth remembering that those buyers already buy drone detection, from radar makers. Echodyne’s radar sits inside Trust Automation’s $490M indefinite-delivery contract for US Air Force counter-drone systems, announced in August 2025, as Business Wire reported in April 2026. That is the competitor a network-sensing offer will meet in a security tender. Europe’s only entry so far is a laboratory one: Vodafone and Tiami Networks detected unconnected objects across a 5G network at Vodafone’s Malaga lab, as ISPreview reported in February 2026.

A US mobile licence covers fixed and mobile services only

A flexible-use licence is wide, but it is not unlimited, and the limit is written in two sentences of the Commission’s rules (Exhibit 3).

Under section 27.2(a), a Part 27 licensee “may provide any services for which its frequency bands are allocated, as set forth in the non-Federal Government column of the Table of Allocations”. Section 30.6(a) says the same for the millimetre-wave bands. Section 2.1 defines the services. Radiodetermination is “The determination of the position, velocity and/or other characteristics of an object, or the obtaining of information relating to these parameters, by means of the propagation properties of radio waves”. Radiolocation is “Radiodetermination used for purposes other than those of radionavigation”. The mobile service is “A radiocommunication service between mobile and land stations, or between mobile stations.” Read those definitions plainly and detecting a drone by its echo is radiodetermination.

The federal government’s only statement on the point was written before the technology existed. On 17 August 2020 the FAA, the Justice Department, the FCC and Homeland Security issued a joint advisory which says: “For example, use of radar to detect UAS requires a Radiolocation Service license from the FCC.” An executive order of 6 June 2025 directed the agencies to revise that advisory within 30 days, but the FAA’s page still lists only the 2020 version and we found no revision. The FCC’s drone docket, opened with DA 26-314 on 1 April 2026, asks about the 24 GHz band “for radar and detection operations” and about Echodyne’s petition for radiolocation there. It observes that the Communications Act “may nonetheless pose barriers to Counter-UAS deployment”, but it does not ask about cellular network sensing. Ericsson, CTIA, ATIS and an aviation coalition do mention integrated sensing in their filings in that docket. The NTIA’s National Spectrum Strategy of November 2023 does not use the word sensing at all.

Our own reading is that the line runs through the product. Sensing that reuses a base station’s ordinary communication transmissions is arguably just part of operating a mobile station. A drone-detection service sold to a third party looks much more like the radiolocation service the 2020 advisory says needs its own licence. So does a dedicated sensing waveform, such as the combined OFDM and chirp signal in the Huawei-based BSense trial that part 4 cites. There are four routes to an answer, and not one of them is pending.

• Declaratory ruling. The Commission simply says whether network sensing is a mobile service.
• Allocation. A table change or footnote adds radiolocation for the band’s licensees.
• Waiver. A licensee asks for one to run a sensing service.
• Rulemaking. A proceeding writes conditions for sensing, as CEPT is doing in Europe.

Permission is the one input a carrier cannot supply for itself. Its licence covers fixed and mobile services, and radiolocation in its bands belongs to the Federal Government. The only interagency statement there is says radar detection of drones needs a radiolocation licence. The FCC has opened no proceeding on network sensing. Europe has one, the ECC work item at 3.4 to 3.8 GHz that part 2 describes.

Congress widened the authority to act on a detection in December 2025

Detecting a drone and doing something about it are governed by different law, and the second half changed in December 2025.

The SAFER SKIES Act, division H of Public Law 119-60, enacted on 18 December 2025, rewrote 6 U.S.C. 124n. Homeland Security and the Justice Department keep the power to “detect, identify, monitor, and track” a drone “without prior consent” and to mitigate it, notwithstanding the aircraft, computer, wiretap and pen-register statutes. What the Act adds is a new class of actor: a State, local, Tribal or territorial law enforcement or correctional agency may, once it has completed the training the statute sets, mitigate a credible threat. The protected objects are people, facilities, “a venue or set of venues used for large-scale public gatherings or events, critical infrastructure, or correctional facilities”. The federal authority ends on 30 September 2031 and the state and local authority on 31 December 2031. The 2020 advisory adds a point that matters for a radar. Detection systems which emit signals and read reflections, “such as radar”, are “less likely to pose concerns under federal criminal surveillance statutes”, whereas detectors that read a drone’s control link may implicate the wiretap and pen-register laws.

For a carrier the division is a clean one. Selling detection, echoes in and tracks out, rests on firmer criminal-law ground than an RF detector does, subject always to the spectrum question above. Acting on a detection belongs to the agencies the statute names, until 2031. So the buyer who can use the whole chain, from detection to response, is a public-safety agency, which happens to be the first customer Ericsson names.

Privacy law written for networks protects customers, and sensing sees bystanders

Sensing sees people who are not customers, and the privacy rules written for networks protect customers. The first regulator to address that gap has said that consent cannot close it (Exhibit 4).

In the United States, customer proprietary network information is defined in 47 U.S.C. 222(h)(1) as information about a telecommunications service “made available to the carrier by the customer solely by virtue of the carrier-customer relationship”. An echo off a bystander is not made available by any customer. The Sixth Circuit held on 2 January 2025 that “mobile broadband is a private mobile service”. On our reading, then, the CPNI rules do not reach network sensing of bystanders. California’s privacy law reaches part of it. Its definition of biometric information includes “gait patterns or rhythms”, while “precise geolocation” is limited to data “derived from a device”; so device-free location may fall outside the law, while gait used to identify a person falls inside it. We found no statement on network sensing from the FCC, the FTC or any state attorney general.

In Europe the ePrivacy Directive defines location data as data about “the terminal equipment of a user”, so device-free sensing falls under the GDPR alone. One regulator has said what that means in practice.

The dates to watch

Seven dated events between December 2026 and the end of 2031 would change the answers above (Exhibit 5).

The one event that would settle the US permission question has no date at all: an FCC ruling, waiver or rulemaking on network sensing. In the meantime the payment record and the permission record point to the same first business: drone detection sold to a public-safety or defence buyer by a vendor or an integrator. It runs as a subscription or a private network around a named site, on a carrier’s radios. Whether the carrier can sell it under its own name and its own licence, at national scale, waits on the FCC. The licences going on sale in April 2027 come with the same mobile allocation as the bands in Exhibit 3 (see our spectrum position decision page).