Research brief · Connectivity · October 2026 · SpaceX filings, FCC, analysts

Starlink mobile:
can it be stable?

Short answer: yes, but as a layer, not a replacement. A cell tower in orbit already sends reliable texts to an ordinary phone, voice and basic data are rolling out, and a stable off-grid service is realistic around 2027 to 2030. What it cannot do, by physics and not just hardware, is replace the towers in a city. We read the filings, the link budgets and the analysts, and mapped why, and when.

See the arc
~600Direct to Cell satellites in orbit by 2026
$10/moT-Satellite in the US, texting live since July 2025
4-18Mbps per beam, shared across a whole region
2027-30When stable supplemental voice and data is realistic
Star trails over a satellite dish at night
The arc · 2024 → 2030 · satellite direct to phone

Texting now, the rest in order

Starlink put an LTE base station in orbit that talks to a normal phone over a partner carrier's spectrum. Texting works today. Voice and data are next. The question is not whether it becomes reliable, it already is for low-bandwidth use, but how far up the capability ladder it can climb before physics caps it at supplemental coverage.

2024 · The tower in space

First satellites up

6 → 100+

The first 6 Direct to Cell satellites launch in January 2024, over 100 by July. Each carries an LTE base station and talks to unmodified phones on T-Mobile's PCS band.

2025 · Texting goes live

Reliable SMS

$10/mo

T-Satellite launches commercially in July 2025, texting and 911 for $10 a month, open to any carrier. App data (weather, maps, chat) follows in October. Reliable for text.

2026 → 2027 · Voice and data

Up the ladder

~20x

Satellite voice is targeted for 2026. Starship-launched V3 satellites carry roughly 20x the capacity, and a $17bn EchoStar deal adds dedicated mid-band spectrum.

2027 → 2030 · Stable, supplemental

It just works, off-grid

Gap-filler

A reliable supplemental service where there is no tower: rural, maritime, disaster. Usable voice and data under open sky. Never a replacement for the city network.

The constellation scales, the beam does not

Direct to Cell satellites in orbit

Cross-check · SpaceX, FCC, Ookla, academic link budgets, analysts

Where they agree

  • All
    It is a cell tower in space

    Each Direct to Cell satellite carries an LTE base station and talks to unmodified 4G phones on a partner's spectrum, T-Mobile's PCS band in the US. About 600 are in orbit by 2026.

  • All
    Texting works now

    T-Satellite launched commercially in July 2025, SMS and 911 for $10 a month, with app data from October and voice calls targeted for 2026. Reliable for low-bandwidth use today.

  • All
    It is supplemental coverage, not a replacement

    The FCC built its 2024 Supplemental Coverage from Space framework around exactly this: fill the dead zones, do not serve the dense areas. That is how SpaceX and T-Mobile pitch it too.

  • All
    Capacity is the binding limit, and it is physics

    A beam carries about 4 to 18 Mbps, shared across a cell 15 to 22 km wide. A near-zero-gain phone antenna, a 500 km path and a borrowed 2x5 MHz of spectrum cap it. The satellite does the heavy lifting.

  • Most
    V3 and more spectrum raise the ceiling

    Starship-launched V3 satellites carry roughly 20x the Direct to Cell capacity, and a $17bn EchoStar deal adds about 50 MHz of mid-band. SpaceX targets 150 Mbps per user around 2028. Still supplemental.

  • Most
    Everyone is building it

    AST SpaceMobile chases real broadband with giant antennas, Apple with Globalstar, Amazon buying Globalstar, China's Qianfan. 3GPP folded satellite into the 5G standard, so phones support it by default.

Where they disagree

When does it "just work" off-grid for voice and data?

Not emergency texting, but a service you rely on. Placed by capability, from here today to never.

Here todayNever (physics)

Can it replace cell towers?

A supplement for dead zones, or real broadband to the phone? The two architectures bet differently.

Supplement onlyReal broadband

Which approach wins direct-to-device?

Many small satellites and text-first, or a few giant antennas and broadband-first?

Text-first scaleBroadband-first

SpaceX FCC filing (18.3 Mbps peak per 5 MHz beam) · Musk (~10 Mbps/beam shared) · arXiv 2025 (~4 Mbps measured outdoors) · Ookla/ISPreview (beam radius 15-22 km) · SatNews 2026 (V3 ~20x, EchoStar 50 MHz, 150 Mbps/user target ~2028) · AST SpaceMobile + Verizon 2025 (broadband to phone) · ComSoc 2026 (analyst forecasts, skeptic quotes).

Probability of outcome · House of 47 synthesis

The odds

Synthesis estimates across the sources below, scenario estimates, not forecasts. The through-line: reliability climbs the ladder from text to voice to data as the constellation densifies, while the per-area capacity ceiling keeps it a supplement to the ground network.

Capability map · what a satellite phone link can do

What it can do, and when

The ladder runs from a text you can always get out, to voice and light data under open sky, and stops hard at high-capacity mobile broadband in a city. The last row never turns to yes, and the reason is spectral reuse, not next year's hardware.

UseToday~2027~2030

Directional, synthesised from SpaceX, T-Mobile, FCC and analyst material. "Emerging" = launching or in beta. The last row stays no because one beam shares a few Mbps across a region, so orbital spectral reuse cannot match a city of towers.

House of 47 read · How to think about it

Coverage, not capacity

Judge it by reach, not speed

Terrestrial networks cover about 15% of the planet's surface. Starlink mobile is a resilience layer over the rest: rural roads, open ocean, disaster zones. It is also the always-on link for the machines that work beyond coverage, the remote and autonomous devices we track at Robots Comparison. Measure it by where it works, not by megabits, and it is already remarkable.

Texting is solved, voice is the next proof

Reliable SMS and 911 from a bare phone is here. The milestone that matters next is off-grid voice and basic data that just work without pointing at the sky, and that lands around 2027 to 2030 as the constellation densifies.

The ceiling is physics, not hardware

A beam shares a few Mbps across a cell tens of km wide. V3 satellites and new spectrum lift the ceiling, but spectral reuse, the thing that makes ground networks fast, barely exists from orbit. It will not replace the city network, ever.

Related reading

This is the second space-scale bet we have mapped. For the other one, moving compute itself off the planet, see our read in Data Centers in Space.

Base case

Plan for a stable supplemental network by the late 2020s. Reliable texting now, voice and light data rolling out through 2027, and a service that just works off-grid by around 2030, bounded to coverage gaps and resilience, never urban capacity.

Tail

Stress-test the capacity story. If V3 at scale and the EchoStar spectrum land as claimed, or AST's broadband-to-phone works, the usable data tier could arrive faster and richer. The physics ceiling on dense areas holds either way.

Starlink mobile will be stable, for texting now, for voice soon, for everything short of the city. The sky fills the gaps, it does not replace the towers.

Sources · figures checked against filings, physics and reporting

Sources

Starlink, Direct to Cell (how it works), official

T-Mobile, T-Satellite service & devices, official

Engadget 2025, first D2C constellation complete (Musk ~10 Mbps/beam), report

ISPreview / Ookla 2026, beam size & FCC contours (15-22 km), analysis

arXiv 2025, measured D2C throughput (~4 Mbps outdoors), study

SatNews 2026, 150 Mbps/user target, EchoStar 50 MHz, V3 ~20x, report

ConvergeDigest, Starlink V3 ~1 Tbps per satellite, report

FCC 2024, Supplemental Coverage from Space framework, order (PDF)

Drive Tesla 2025, T-Satellite dates (texting Jul, data Oct, voice 2026), report

TechRadar 2024, AT&T/Verizon FCC objections (out-of-band limits), report

Satellite Today 2025, AST SpaceMobile + Verizon broadband deal, report

Apple / Globalstar 2024, $1.5bn investment, Emergency SOS, report

Amazon / Globalstar 2026, D2D acquisition, report

Satellite Today 2025, China's Qianfan / SpaceSail D2C, analysis

ComSoc 2026, D2D analyst forecasts & skeptic quotes, aggregation

Fierce Network, satellites will not replace towers, analysis

WIA, D2D as a complement to terrestrial, overview

Satellite Today 2024, T-Mobile beta & rollout, report

Changelog · what changed in this brief, newest first

Changelog