How the rules treat non-terrestrial networks

The FCC rewrote two rules in twelve weeks

Two rules did most to shape low-orbit broadband in the United States. One was a power limit written at WRC-2000 to protect geostationary satellites. The other was a licensing process that treated each application as its own case. The Commission rewrote both in 2026, and one of the rewrites is in force (Exhibit 1).

The first is the power limit. The equivalent power-flux-density limit, EPFD, caps the signal a non-geostationary constellation may put on the ground in the 10.7 to 12.7, 17.3 to 18.6 and 19.7 to 20.2 GHz bands, so that geostationary links underneath keep working. FCC 26-26, adopted on 30 April 2026, “replace[s] the EPFD framework with modern, performance-based GSO protection criteria”. The rule took effect on 13 July 2026, apart from two provisions delayed indefinitely. DIRECTV’s petition for reconsideration, with replies closed on 31 July, is pending.

The replacement is a measure of harm, and the harm is lost throughput (Exhibit 2). Where operators have no coordination agreement, a geostationary link may lose no more than 3% of its time-weighted average throughput over the long term. Over the short term it may suffer no more than a 0.1% absolute increase in unavailability. Links without adaptive coding, which cannot trade speed for resilience, are protected instead at an interference-to-noise ratio of −10.5 dB for 80% of the time. Low-orbit satellites must also keep at least 3° away from the geostationary arc. None of this is new to the Commission: the same throughput test has governed sharing between low-orbit systems since FCC 24-117 took effect on 13 January 2025, with a looser 0.4% short-term test.

The Commission puts the benefits at “well over $2 billion, with capacity increases of 100% to 700% possible using the same number of in-orbit NGSO satellites”. Those figures come from the parties with the most to gain: SpaceX’s and Amazon’s comments supplied the range, on the argument that a system can use eight satellites at once over a location “instead of being effectively limited to one”.

Part 100 replaces case-by-case review with a default to yes

The second rule is Part 25. FCC 26-47, adopted on 22 July 2026, creates a new Part 100 and says “The Commission will sunset part 25 once part 100 becomes effective.” It has not become effective. The FCC’s own fee order, published on 14 September 2026, keeps the current approach “until the newly adopted Part 100 rules become effective and replace the existing rules”. No Federal Register publication of FCC 26-47 had appeared by 7 October. When it does take effect, Part 100 changes seven things.

• Review. A “default to yes” regime, with targeted review only for listed issues, among them waivers, market access, foreign ownership and federal coordination. Completeness is decided within 30 days and public notice runs 15 days.
• Licence term. A default of 20 years for most systems.
• Processing rounds. Opt-in, with annual windows from 1 January to 31 October, starting in the Ka, Ku, Q and V bands.
• Bond and milestones. For processing-round systems only: a $10 million surety bond, posted within 30 days of grant and declining to zero at 90% deployment, with half the system due in six years and all of it in nine. Other systems follow the ITU milestones (Exhibit 3).
• Earth stations. A “Nationwide, Non-Site license” under which individual sites are registered. An operator with a current, approved Form 312 asking for a licence without waivers may operate on a non-interference, unprotected basis once its application is on public notice.
• Orbital debris. Certification that satellites are designed and operated to de-orbit within five years of the end of the mission, which is the rule FCC 22-74 applied to satellites launched after 29 September 2024.
• New categories. Licences for several geostationary satellites at one location, for vehicles that change orbit and for systems that span several orbits.

Filed FCC 26-47, adopted 22 Jul 2026, released 23 Jul 2026 · FCC 26-59, 91 FR 58258, 14 Sep 2026 · FCC 22-74

Of the seven, the milestone change is the one an investor can put a price on. Under Part 25, as DA 26-553 restates it, a licensee that misses the half-way milestone forfeits its bond. Its authorisation is also “reduced to the number of satellites that are in orbit and operating on the date of the missed milestone”. The second penalty is the severe one, because it shrinks the licence to whatever happens to be flying. Under Part 100 a processing-round system that misses the milestone keeps its authorisation and loses its place in the queue instead. The ITU’s own milestones run on a longer clock (Exhibit 3).

The ITU kept Article 22

The EPFD limits are international law, written into Article 22 of the Radio Regulations, and the conference that could have reopened them chose not to.

At WRC-23 a CITEL proposal, co-signed by ten states including the United States, would have put a review of the EPFD limits on the WRC-27 agenda. It “was not adopted”. Instead, according to the plenary minutes FCC 26-26 quotes, the conference invited ITU-R to study the limits “and to inform WRC-27 of the results of the studies, without any regulatory consequences”, a diplomatic refusal. The only EPFD question on the WRC-27 agenda is item 1.4, on limits for low-orbit downlinks at 17.3 to 17.7 GHz in Regions 1 and 3. Outside the United States, Article 22 still applies to the same satellites the FCC has released from it at home.

WRC-27 will instead do a great deal of other satellite business. Thirteen of its nineteen agenda items, on our count, concern satellite and space communication systems directly (Exhibit 4). Of the rest, item 1.7 brings IMT into bands with satellite incumbents, while items 1.17 to 1.19 concern space weather sensors and passive Earth-observation satellites. The conference meets from 18 October to 12 November 2027. The FCC has proceedings of its own on several of the same bands and is not waiting. FCC 25-29, adopted in May 2025, aims to make “over 20,000 megahertz of spectrum available for satellite use across four bands”, although no final rules had appeared in the Federal Register by 7 October 2026. The direct-to-device series covers the FCC rules that allow satellite service in terrestrial bands in the United States.

3GPP put the base station in orbit between Release 17 and Release 19

In Release 17 the standard let a phone talk through a satellite that did nothing more than relay its signal. Two releases later it had put the base station itself in orbit and linked satellites to one another, a long way in under four years (Exhibit 5).

The difference between the two payload designs comes down to where the base station is (Exhibit 6). A transparent satellite relays to one on the ground. A regenerative satellite carries its own, and because it does, it can link to other satellites as well.

Two Release 20 items show where the standard goes next. The first undoes an assumption Release 17 started with: every satellite specification so far has assumed a device that knows its own position from GNSS and uses it to correct its timing and frequency for the satellite’s motion, the Doppler shift of up to 23 parts per million that part 1 calculates. Without GNSS that device is lost. A GNSS-resilient study finished in June 2026 and a normative item followed. TR 38.914 goes further by making it a 6G requirement that the radio “shall be able to operate without GNSS in both Idle and Connected states”.

The second, RP-262263, was approved in September 2026 with a target of the first quarter of 2027. It starts from an observation: “Some satellite systems perform downlink pre-compensation and uplink post-compensation in order to support NR UEs that are not capable of performing uplink pre-compensation”. It then asks whether those devices and standard satellite devices can share a cell. In our view this describes the direct-to-device systems that serve unmodified phones, whose radios the direct-to-device series describes. If so, 3GPP is bringing a design choice those systems made outside the standard back inside it.

Every one of the fourteen satellite band numbers in TS 38.101-5 is in satellite spectrum: L, S, Ku and Ka. 3GPP now specifies what an operator needs to run satellites inside its own radio network, base stations in orbit included. It does not specify a satellite band in terrestrial mobile spectrum, which waits for WRC-27 item 1.13 in November 2027.

Terms used

• EPFD. Equivalent power-flux density: the combined power that all of a non-geostationary system’s satellites may deliver at a point on the ground. Article 22 of the Radio Regulations sets it to protect geostationary links.
• Adaptive coding and modulation. A link that changes its coding and modulation as conditions change, trading throughput for resilience. The FCC’s new test measures harm as lost throughput on links like that.
• Processing round. The FCC’s way of considering competing satellite applications in the same band together; earlier rounds are protected against later ones for a period.
• Transparent and regenerative. A transparent satellite relays the signal to a base station on the ground; a regenerative one has the base station on board.
• FR1-NTN and FR2-NTN. 3GPP’s two frequency ranges for satellite bands: 410 MHz to 14.5 GHz, and 10.7 to 30 GHz.