How the radios work

There are two ways to buy back the link budget you lose in that distance, and the two leading operators each picked one. They are not variations on a single design. They are opposite bets on where the gain should come from, and almost everything else about each constellation (beam count, fleet size, interference management, replacement cadence) follows from which bet a company made.

The two designs make opposite choices

AST buys the link budget back with aperture. Its authorization order DA-26-391A1 was adopted on 21 April 2026 and describes each satellite carrying a large phased-array antenna operating thousands of electronically steerable beams simultaneously, with the array sized so that the aperture produces narrow primary beams and low sidelobe levels. Those beams are actively shaped to match service-area boundaries, and the same order records that they can be switched off where they would otherwise cause interference. The underlying physics is unremarkable: a bigger antenna concentrates energy into a smaller patch of ground, so there is more signal where you want it and less where you do not. What is demanding is the engineering needed to fly an antenna that large, and to keep it pointed.

SpaceX buys the same link budget back the other way, with proximity and numbers. The FCC’s order DA-24-1193A1 of 26 November 2024 authorises supplemental coverage across up to 7,500 Gen2 satellites operating in shells at 340, 345, 350 and 360 km. Flying lower shortens the path; flying a great many of them shortens the wait for one to be overhead. Neither lever asks anything of the antenna, which is what makes the two approaches opposite rather than merely different.

That is the authorisation. The orbital record shows it being executed: our own figure, computed from CelesTrak two-line elements by converting mean motion to altitude, puts 824 catalogued Starlink objects (7.7%) below 400 km, which is the altitude band the order describes. Public catalogues do not identify which of those objects carry direct-to-cell hardware, so this is evidence that the low shells are being populated rather than a measure of the service capacity in them.

Neither approach is obviously right; they simply put the cost in different places. Aperture is expensive per satellite and hard to launch, which makes it a capital and manufacturing problem, concentrated before the first satellite reaches orbit. Proximity is cheap per satellite and expensive in fleet size, atmospheric drag and replacement cadence, which makes it an operating problem that never stops. Nothing in the public record settles which of those is the better trade over a decade.

The bands follow the design

Service spectrum is where the two designs stop being a matter of engineering taste. Lower frequency is worth real link budget: at 700 MHz a signal penetrates walls and diffracts around terrain far better than at 1.9 GHz, which is why the low-band holdings of AST’s terrestrial partners are the ones being leased rather than anything further up the band plan. The gateway and telemetry bands follow the same logic in reverse, sitting far higher in the spectrum, where bandwidth is plentiful and nothing has to reach a handset through a wall.

The trade is width, and width is the whole capacity story. Those 700 MHz blocks are 6 MHz wide; the PCS G block is a single 5 MHz pair. Whatever the constellation above is doing, the service link is metered through a narrow pipe shared by everyone under the beam, and neither aperture nor altitude changes that arithmetic.

Satellite count improves availability and revisit; it does not widen the channel.

The capacity story is routinely mistold. Capacity per user comes from bandwidth, spectral efficiency and how tightly beams can be reused, which is exactly why aperture and beam control matter as much as fleet size. It is also why a headline satellite count is an availability claim rather than a throughput claim: more satellites mean a shorter wait and fewer gaps, not a wider channel.

The constraint nobody puts on a coverage map

Because this spectrum is terrestrial, the binding limits are not really about reaching the handset at all. They are about interference into everyone else who uses the band on the ground, which is a constraint no coverage map has ever shown. AST’s order records that it filed power flux-density and maximum EIRP density values with the FCC, and that it can adjust EIRP across operating altitudes and elevation angles to stay compliant. That is a design in which the regulator, rather than the link budget, sets the upper bound on how hard a satellite is allowed to transmit.

On the SpaceX side, one of those limits has not been settled at all. The Commission neither granted nor refused the request to waive the aggregate out-of-band power flux-density limit of −120 dBW/m²/MHz for these operations; it deferred the question. That means an authorized service is operating with a significant power-limit question still open.

Filed The FCC deferred SpaceX’s request to waive the aggregate out-of-band power flux-density limit of −120 dBW/m²/MHz for these operations (order DA-24-1193A1).

For an engineering buyer, that is a live risk to plan around rather than a footnote to skim past. For a strategist, it is a reminder that regulatory headroom, not technology, may set the ceiling on how much this service can ultimately deliver. It is also one of the two technical items that belong in diligence on this market and almost never appear there.

A cross-check that costs nothing

None of this requires taking an operator’s word for anything, and one check costs nothing at all to run. The SatNOGS database, telemetry logged by volunteer ground stations, records AST BlueBird beacons at 430.5–439.5 MHz (GFSK/FSK, 2.4–19.2 kbaud) and S-band telemetry at 2225 and 2235 MHz (BPSK, 200 kbaud). Those S-band frequencies sit inside the 2200–2290 MHz command-and-telemetry band in the FCC authorization, so the filing and the observed radio agree.

Judgement SatNOGS transmitter records at 430.5–439.5 MHz and at 2225 and 2235 MHz fall inside the 2200–2290 MHz command-and-telemetry band in the FCC authorization: two unrelated sources describing the same radio.

Amateur observation cannot see the service payload: beacon observations are telemetry, not service payload. It does independently confirm that the satellites are where the paperwork says, transmitting what the paperwork says. An unrelated source agreeing with the filing is not a measurement of the service; it is a reason to believe the documents that describe it, and the rest of the record with them.