On December 23, 2025, an Indian LVM3 rocket lifted a 6,100-kilogram communications platform off a launch pad in Sriharikota and into low Earth orbit. It was BlueBird 6, the first of AST SpaceMobile’s production-class direct-to-cell satellites. Two months later it finished unfolding a 223-square-meter phased-array antenna, the largest commercial communications array ever deployed at LEO, Scientific American reported. Its job is to beam a 4G and 5G cell signal directly down to ordinary, unmodified smartphones. A second BlueBird flew to orbit on a Blue Origin New Glenn from Cape Canaveral on April 19, 2026.

T-Mobile, on a different fleet of SpaceX Starlink satellites, has been selling a $10-a-month direct-to-cell add-on since July 23, 2025. AT&T plans to start a beta service for FirstNet users on AST in the first half of 2026; Verizon signed a commercial agreement in October 2025. Three of the four largest US wireless carriers have now placed a bet on cellular-from-space. The companies launching the satellites — SpaceX, AST SpaceMobile, Lynk Global, Globalstar — have raised and spent billions on rockets, ground stations, and football-field-sized radios.

None of them own the patents that make a phone know how to find a satellite.

A search of the US patent grants in our 9.3M-document corpus, filtered for the cellular-from-space technology spec known as “non-terrestrial network” or NTN, surfaces 79 grants issued in 2025. Twenty-three of them belong to Qualcomm. Thirteen to Apple. Five to MediaTek. Four to Samsung. Three to Ericsson. Together the cellular chip and infrastructure clique accounts for more than 60 percent of the year’s grants. AST SpaceMobile’s count under that exact filter is zero. So is SpaceX’s.

The chip industry has spent the last four years quietly building the tollbooth around a road the satellite industry is still paving.

How a tollbooth gets built

Patents on US wireless infrastructure are not a free-for-all. The 3rd Generation Partnership Project, the global standards body that defines what counts as 5G, freezes specifications in numbered “releases.” Once a feature lands in a release, every chip that claims to comply has to implement it. And every patent that claims to cover an essential way of implementing it — a standard-essential patent, in the trade — gets paid by every chip that claims to comply. This is how Qualcomm built a business out of phones it does not manufacture.

Direct-to-cell satellite is brand new on this conveyor belt. 3GPP Release 17, frozen in 2022, was the first to specify NTN: how a base station in low orbit, racing past at 7.5 kilometers per second, can hand a phone call back and forth with another satellite or a ground tower without dropping the line. Release 18, locked in 2024, extended the spec to higher frequency bands. Release 19, due to wrap this year, adds delay-tolerant store-and-forward modes for IoT.

The patents being granted now are the working drawings of those specifications. Qualcomm’s August 2025 grant US12389300 covers handover between a terrestrial network and an NTN, including the trick of preserving the phone’s session context across the swap so a video call doesn’t drop. US12382369, granted the same week, claims a particular way to broadcast satellite ephemeris — orbital position data — inside the system information block that every phone listens to on connection. US12298412, granted in May, describes how to pin down a phone’s location using just one satellite by combining Doppler and range measurements. US12262264 from March packages a satellite’s round-trip-time and ephemeris-validity duration into a handover command. The titles read like a how-to manual for the spec.

Apple’s filings sit on the same protocol shelf. US12388521 (August 2025) indicates a transmission gap on the uplink based on the distance to the serving satellite. US12231222 (February) constrains a satellite radio so it conforms to the regulations of whichever country happens to be passing under it. US12356458 enhances the random-access procedure to accommodate the seconds-long round trip a phone-to-LEO conversation can incur.

These are not vague gestures at “satellite stuff.” They are specific, testable, infringeable mechanisms that any chip must implement to shake hands with a satellite under the 3GPP standard.

The independent patent analytics firm GreyB ran a similar count in 2024 against ETSI’s official declared-essentials list. Their finding: Qualcomm holds 18.66 percent of declared 5G NTN essentials in the United States, the largest single share. Ericsson, Samsung, Xiaomi, Huawei, and Nokia split most of the rest. The number has likely grown since.

The other side of the airlock

What AST and SpaceX do own is the other half of the radio link: the satellite itself.

AST’s 51 issued grants are a catalogue of what it takes to put a soccer-pitch-sized phased array in orbit and keep it pointed at the right point on Earth. US12280894 covers the scheme by which square antenna tiles fold for launch and unfurl in vacuum. US12263962 protects a reaction-momentum wheel design that can roll and pitch the array in flight. US12261678 holds the attitude-control system that uses carrier-phase differential GPS to keep the antenna planar against orbital flexing. US12200508 is the adaptive amplitude-tapering each beam needs to suppress sidelobes when the satellite is forming hundreds of cells simultaneously across its footprint. They are extraordinary engineering. They are not the cellular protocol.

SpaceX’s 133 grants tell the same story one altitude lower. The recent ones are nearly all user-terminal hardware: pan-and-tilt antenna mounts, in-line calibration for phased arrays, radome assemblies. Where SpaceX does file communication-protocol patents, they are about traffic on the Starlink network’s own internal links, not the standardized 3GPP-NTN link a Galaxy or iPhone uses to find the bird.

Of AST’s 51 lifetime grants, just four contain any of the words LTE, cellular, 5G, handover, or HARQ. Two are different revisions of the same patent on compensating for Doppler-induced frequency error in LTE signals. One is the SAT-RAN beam handover from August 2025. Their job is to make a satellite look enough like a tower that an unmodified phone connects. Qualcomm’s job is to define what “enough like a tower” means in the spec, and to own the patent on the answer.

The Snapdragon Satellite detour

Qualcomm did not arrive at this position by accident. In January 2023 it announced Snapdragon Satellite, a proprietary direct-to-device service built on Iridium’s L-band fleet, intended to ship in Android phones the same year. By November the deal was dead. As Qualcomm and Iridium acknowledged at the time, smartphone makers had refused to integrate the chips. The reason, Computer Weekly and SDxCentral both reported, was that Android OEMs would only build satellite phones around an open 3GPP standard, not a one-vendor solution.

The proprietary play failed in November 2023. By the next December, Qualcomm had filed and begun to receive grants on the standardized version. The company filed one NTN grant in 2021. By 2023 it was twenty-one. By 2025 it was twenty-three. Qualcomm did not abandon the satellite business after Iridium; it joined the standards body that was about to make satellite mandatory and started drafting the patentable parts.

In March 2025, Ericsson, Qualcomm, and Thales Alenia Space ran the first end-to-end 5G NTN call over a simulated LEO channel. They followed up later in the year with over-the-air handovers between terrestrial and satellite cells. The press release reads like a victory lap. It is also a market-creation event for a royalty stream that did not exist before.

Who pays and how

If history is the guide, the chip vendors will not collect on satellite calls directly. They will collect from the modem makers, who will pass it through to the handset OEMs, who will pass it through to the carriers, who will pass it through into the $10-a-month T-Mobile add-on and the AT&T tier and the Verizon bolt-on. Standard-essential royalties move along a supply chain like a quiet weight pressing down on every contract.

For a venture investor staring at a satellite-broadband cap table, the implication is awkward. The money you raise to launch a constellation pays SpaceX or Blue Origin to lift it. The money your subscribers pay you to use it pays Qualcomm and Ericsson and Samsung for the spec they own. The cellular royalty machine has already consumed Wi-Fi, automotive V2X, and NB-IoT in the last decade. Outer space is the next domain that has a radio.

The satellites are the ones with the rocket flames. The patents are the ones with the rent.

Method note. Counts come from a working corpus of 9.3M US patent grants sourced from the USPTO bulk grant XML. The NTN filter requires a grant’s title or abstract to mention satellite together with one of non-terrestrial, NTN, or direct-to-cell; year buckets use issue date, not application date. Assignee normalization is name-based and case-insensitive, so each count combines variant spellings and subsidiary filings. Standard-essential percentages come from GreyB’s 2024 analysis of the ETSI 5G declared-essentials list and may have shifted since. Counts measure what has issued, not what has been applied for; pending applications skew further toward the chip vendors. Partnership and launch dates come from Qualcomm, Ericsson, T-Mobile, Apple, AST SpaceMobile, and Globalstar press releases, plus reporting by Scientific American, Via Satellite, SatNews, SDxCentral, Capacity, and GeekWire.