Direct-to-Cell Explained: How Your Phone Reaches a Satellite

Direct-to-cell is a way of connecting an ordinary, unmodified smartphone to a satellite instead of a ground tower. The satellite carries a real LTE base station and broadcasts on the same licensed cellular frequencies your carrier already uses, so the phone treats it like a tower that happens to be 350 km (about 217 miles) overhead.

The catch is physics: the satellite needs an enormous antenna, it has to correct for its own speed, and each beam has very little capacity to share. That is why texting arrived first and voice is still being rolled out.

What direct-to-cell is

Until recently, “satellite phone” meant a special handset with a stubby antenna on satellite-only frequencies. Direct-to-cell (also called direct-to-device, D2D, or direct-to-handset) flips that around. The phone stays exactly as it is. The satellite does all the adapting.

Starlink describes its approach plainly: each Direct to Cell satellite carries an onboard eNodeB modem that “acts like a cellphone tower in space”, and carriers integrate it like a standard roaming partner. The company says existing LTE phones work “wherever you can see the sky”, with no hardware, firmware or special app changes (Starlink Direct to Cell).

An eNodeB is the formal LTE name for the base station in a cell tower. In direct-to-cell, that function rides on a satellite moving at orbital speed, and the “cell” it creates is a beam footprint painted onto the ground.

Three flavors of the idea exist today, and they are easy to confuse:

1. Unmodified-phone LTE/5G from LEO (SpaceX Starlink, AST SpaceMobile, Lynk): the satellite speaks standard cellular on carrier-licensed bands, so any recent LTE phone works.
2. Standards-based narrowband NTN (Skylo, and Iridium’s planned service): the phone needs a newer modem chip that understands 3GPP’s non-terrestrial network extensions and tunes to satellite (MSS) bands. Mostly text and SOS.
3. Proprietary (Apple with Globalstar): a custom protocol and custom antenna inside the iPhone, over Globalstar’s own satellite frequencies.

This article focuses on the first flavor, then compares it to the other two.

How a normal phone can reach a satellite

A smartphone’s cellular radio transmits at roughly 0.2 to 2 watts (NewSpace Tracker). The LTE standard caps the usual handset at 23 dBm, which is 200 milliwatts (TelecomHall). A cell tower, by contrast, radiates 20 to 40 watts and sits a few hundred meters to a few kilometers away (KeepTrack).

Radio signals weaken with the square of distance. Moving the receiver from 2 km to 350 km multiplies the distance by 175, so received power drops by a factor of about 30,000 before anything else. NewSpace Tracker puts the free-space path loss from a phone to a LEO satellite at roughly a billion times worse than to a nearby tower, since real towers are often closer and the slant path is longer than the altitude.

The big antenna is the whole trick

Antenna gain scales with physical area. Double the area and you collect twice the signal and can focus a narrower beam. That is why direct-to-cell satellites are dominated by one giant flat panel.

The verified sizes:

• Starlink Direct to Cell (V1): a 2.7 by 2.3 meter phased array, roughly 6 square meters, per satellite according to SpaceX, versus a roughly 1 meter antenna on a conventional Starlink broadband satellite, and up to 43 spot beams versus about 16 (Aerospace America).
• AST SpaceMobile BlueBird 1 to 5 (Block 1): a 64 square meter array, launched September 12, 2024, operating at 520 km (KeepTrack: AST BlueBirds).
• AST SpaceMobile BlueBird 6 (Block 2): about 2,400 square feet, roughly 223 square meters, described by the company as the largest commercial communications array ever deployed in low Earth orbit, 3.5 times the size of the first five BlueBirds (Converge Digest; Telecoms.com).
• SpaceX next-generation D2C satellites (filed, not yet flown): about 250 square meters and 2,500 kg each, at 330 km, per the FCC application filed September 17, 2025 (Space Intel Report).

A 223 square meter array is bigger than a tennis court. That panel is what makes a 200 mW uplink readable from space, and it lets the satellite form many narrow beams and steer each one electronically as the ground slides past.

Why the satellites fly so low

Lower is better for the link budget, so direct-to-cell satellites fly lower than almost anything else in commercial use. SpaceX’s constellation page lists its V1 direct-to-cell shell at a mean altitude of 360 km (53 degrees inclination) and 358.5 km (43 degrees), with V1 DTC satellites occupying 330 to 370 km, versus 450 to 490 km for today’s broadband satellites (Starlink Space Safety). A brightness study cited by N2YO rounds that to about 350 km, versus about 550 km for the original broadband shell (N2YO).

When the FCC authorized SpaceX’s Gen2 satellites in November 2024, it allowed operation at 340, 345, 350 and 360 km, with the condition that SpaceX coordinate with NASA for anything below 400 km to protect the International Space Station (TechCrunch).

The trade-off is drag. At 350 km there is still a whisper of atmosphere, so satellites need propulsion to stay up and live short lives: SpaceX told the FCC its next-generation D2C satellites would last five to seven years (Space Intel Report).

How the satellite copes with moving at 7.5 km/s

Terrestrial LTE assumes the base station is bolted to the ground and the phone moves at highway speeds. Direct-to-cell breaks that in two ways: the base station is screaming across the sky, and the signal takes much longer to arrive.

Doppler shift

The 3GPP reference scenarios for non-terrestrial networks put a 600 km LEO satellite at 7.56 km/s relative to the Earth’s surface (3GPP TR 38.821, section 7.1.1). As it rises over your horizon it is coming toward you; as it sets it is moving away. Just as an ambulance siren rises then falls in pitch, the satellite’s radio frequency arrives higher than transmitted on approach and lower on departure.

3GPP’s reference table gives the maximum Doppler shift for a 600 km LEO satellite as 24 parts per million, and the maximum rate of change as 0.27 ppm per second (3GPP TR 38.821, section 4.2). At a 1,900 MHz PCS carrier, 24 ppm works out to a shift of roughly 46 kHz, and it swings from positive to negative over a single pass. That is far outside the frequency error an LTE handset is designed to tolerate from a fixed tower. Left uncorrected, the phone would simply fail to lock on.

Here is the elegant part. The phone is unmodified, so it cannot fix this. The satellite fixes it on the phone’s behalf. It knows its own orbit and where each beam points on the ground, so it calculates the Doppler shift for that beam at that instant and deliberately transmits slightly off frequency, so the signal arrives exactly where the phone expects it. It applies the mirror-image correction to what it receives. Industry white papers describe this as compensating for Doppler and round-trip time “on the network side” (Wiley / IEEE). Each beam has its own geometry, so each needs its own correction, updated continuously.

Timing advance

The second problem is delay. LTE slices time into slots, and each phone must transmit so its signal lands in the correct slot. A phone far from the tower is told to send a little early; this is the “timing advance”. Terrestrial LTE is designed for cells of a few kilometers, rarely more than a few tens (TelecomHall).

For a 600 km satellite, 3GPP gives a round-trip propagation delay of 12.89 ms for the service link alone, when the base station is on the satellite (3GPP TR 38.821, section 7.1.1). That is far more than an LTE slot. 3GPP’s non-terrestrial additions handle it with a “common timing advance” broadcast by the base station plus a per-device offset computed from the device’s GNSS position and the satellite’s orbital data (5G from Space, arXiv). Those additions need a new modem, which an unmodified phone lacks. So direct-to-cell operators absorb the bulk delay inside the satellite’s base station software, and the phone sees a large but otherwise ordinary cell. The remaining variation within one beam is small enough for standard timing-advance commands.

Handover every few minutes

A satellite at this altitude crosses the visible sky in roughly four minutes (KeepTrack). The network holds the beam footprint fixed on the ground as long as it can, then hands your session to the next satellite. The phone sees an ordinary cell handover.

What happens after the satellite receives your text

A cell tower is useless without a connection back to the carrier’s core network, called backhaul.

Starlink’s Direct to Cell page says the satellites “will immediately connect over laser backhaul to the Starlink constellation” (Starlink via Telecoms.com; Starlink Direct to Cell). Your message hops from the D2C satellite, over an optical link, to one or more broadband Starlink satellites, then down to a ground station. From there it enters the carrier’s core network as if from a roaming partner’s tower, and on to the internet or the phone network.

AST SpaceMobile uses a different ground architecture, with nearly 50 gateways in various stages of completion, installation or planning as of August 2026 (AST SpaceMobile Q2 2026 update, SEC).

The point of the lasers is coverage: a D2C satellite over the middle of Nevada may have no ground station in view, but a neighbor does.

What spectrum direct-to-cell uses

A phone can only listen to the bands it was built for, and a carrier only has rights to specific licensed blocks. For direct-to-cell to work with unmodified phones, the satellite must transmit on the carrier’s own terrestrial spectrum. That takes regulation.

In the United States, the FCC created a framework for exactly this. The Supplemental Coverage from Space (SCS) Report and Order was adopted on March 14, 2024 and took effect May 30, 2024 (Global Policy Watch / Covington). The order lists the bands eligible for SCS:

• 600 MHz: 614 to 652 MHz and 663 to 698 MHz
• 700 MHz: 698 to 769 MHz, 775 to 799 MHz and 805 to 806 MHz (including the FirstNet segments)
• 800 MHz: 824 to 849 MHz and 869 to 894 MHz
• Broadband PCS: 1850 to 1915 MHz and 1930 to 1995 MHz
• AWS-H Block: 1915 to 1920 MHz and 1995 to 2000 MHz

(FCC SCS Report and Order draft)

The conditions matter. A satellite operator can only provide SCS if the terrestrial licensee (or group of licensees) holding all the licenses on that channel across a defined geographic area leases the spectrum to it. The satellite service is secondary: it must not cause harmful interference to terrestrial services and gets no protection from them. Interim rules require 911 calls and texts to be routed to a public safety answering point (FCC).

In Starlink’s case, the lease is with T-Mobile for the PCS G Block, 1910 to 1915 MHz uplink and 1990 to 1995 MHz downlink (TelecomHall; Converge Digest). That is a single 5 MHz block in each direction. Keep that number in mind for the capacity section.

AST SpaceMobile’s satellites are designed to span 700 to 2,700 MHz, the full range of terrestrial 4G and 5G bands (KeepTrack: AST BlueBirds), and the FCC granted AST SCS authorization for commercial U.S. service covering a network of up to 248 satellites (AST Q1 2026 update, SEC).

The spectrum story is changing. On September 8, 2025, EchoStar agreed to sell its AWS-4 and H-block licenses to SpaceX for about $17 billion, for a next-generation Starlink Direct to Cell constellation (EchoStar). Unlike a carrier lease, that spectrum would be SpaceX’s own. What SpaceX intends to do with it is the subject of Starlink Mobile vs AT&T, Verizon and T-Mobile.

Why capacity is so small, and why it only works outdoors

Two facts explain almost every limitation you have read about.

The beam is wide and the spectrum is narrow

When SpaceX launched its first six D2C satellites in January 2024, Elon Musk posted that the system supported “approximately 7 Mbps per beam” and “does not significantly compete with existing terrestrial cellular networks” (Wireless Estimator). In SpaceX tests in March of that year, the measured rate was 17 Mbps (Aerospace America).

Now consider what a beam covers. A tower serves a cell a few kilometers across. A satellite beam covers tens of kilometers; AST’s design uses spot beams about 48 km wide (Aerospace America). Every phone in that footprint shares the same handful of megabits. Users on the current service see hundreds of kilobits per second at best (KeepTrack).

That is why texting came first. An SMS is a few hundred bytes; a beam that cannot stream video to one person can still deliver thousands of texts per minute. Messaging apps came next, because they are also small bursts. Voice is hardest: a continuous, low-latency stream in both directions, from a 200 mW transmitter, for the whole call.

AST SpaceMobile’s answer is more antenna. Its Block 1 satellite set a record peak of 98.9 Mbps to an unmodified smartphone, and the company expects Block 2 to approach 200 Mbps per satellite (AST Q1 2026; AST Q2 2026). Those are single-phone peaks under test conditions, still shared across a beam in real use.

SpaceX’s answer is also more antenna, plus its own spectrum. The V2 D2C satellites, planned to fly on Starship from mid-2027, are described by Starlink as having 100 times the data density of the V1 generation (5Gstore, citing MWC 2026 remarks).

Walls win

Apple’s support page is blunt about the outdoor requirement for its satellite features: “Be outside with a clear view of the sky and the horizon”. Light foliage may slow the connection; dense foliage, hills, mountains, canyons and tall structures may block it (Apple Support). Starlink’s phrasing is “wherever you can see the sky” (Starlink).

The reason is the same link budget. Under a tower, the signal has so much margin it can lose most of its strength through a roof and still be fine. The satellite link has no margin to spare; the giant antenna was needed just to close it in open air. A roof, a canopy or a car body takes it back below the threshold.

Who the direct-to-cell providers are

Status as of October 2026, from the sources listed at the end.

SpaceX Starlink Direct to Cell. The first six D2C satellites launched in early January 2024, and the first text went through the T-Mobile network on January 8, 2024 (Gizmodo; SDxCentral). T-Mobile’s T-Satellite service launched commercially on July 23, 2025 with texting, and data for selected apps followed from October 1, 2025. The constellation is over 650 D2C satellites, covering more than 500,000 square miles of U.S. territory without tower coverage (SatelliteInternet.com; Cybernews; KeepTrack). Voice runs through WhatsApp; a guide updated in July 2026 says native voice calling “is still in development” (SatelliteInternet.com). Partner carriers include T-Mobile (U.S.), Rogers (Canada), Optus (Australia), One NZ, KDDI (Japan), Salt (Switzerland) and Entel (Chile, Peru) (Starlink). If you want the practical details, see our guide to whether you can make phone calls with Starlink.

AST SpaceMobile. Thirteen BlueBirds were in orbit as of August 10, 2026, with production running through BlueBird 46 and a target of about 45 in orbit by the first quarter of 2027. The company is preparing a 2026 beta with strategic carrier partners, with 3,000 initial digital cells activated across the continental U.S., and has over 60 carrier agreements including AT&T, Verizon, Vodafone and Rakuten (AST Q2 2026, SEC; Advanced Television). AT&T has offered messaging and emergency services over the first five BlueBirds to a subset of customers since 2025, with call support targeted for late 2026 (PhoneArena).

Lynk Global, now Elveo Mobile. Lynk and Omnispace completed their merger on August 14, 2026 to form Elveo Mobile, with SES as a significant shareholder. It has nine satellites (six Lynk LEO, two Omnispace LEO, one MEO), 56 carrier and distribution contracts across more than 60 countries, and plans a 320-satellite LEO constellation starting in the fourth quarter of 2027. Initial services are narrowband: SMS and messaging apps (Light Reading).

Skylo. Skylo owns no satellites. It runs a 3GPP Release 17 narrowband NTN network over geostationary satellites from Viasat, EchoStar, Inmarsat and Terrestar, from 10 earth stations. It powers satellite SOS and messaging on Pixel 9 and 10 phones and Pixel Watch 4, and Verizon uses it for direct-to-device messaging (TelecomTV). Deutsche Telekom, Skylo and Qualcomm demonstrated SMS over GEO satellite on a Snapdragon X80 modem in November 2024 (Deutsche Telekom).

Apple and Globalstar. Apple announced Emergency SOS via satellite with the iPhone 14 on September 7, 2022, over Globalstar’s 48-satellite constellation at 1,414 km (Universe Today). In November 2024 Apple committed $1.5 billion to Globalstar, including $1.1 billion to expand its infrastructure and $400 million for a 20 percent stake (GSMArena). iPhone 14 and later offer Emergency SOS, Messages, Roadside Assistance and Find My via satellite, free for two years after activation (Apple Support). For a device-by-device walkthrough, see satellite texting on iPhone, Pixel and Galaxy.

Iridium. Iridium’s “Project Stardust” adds NB-IoT over NTN to its L-band LEO constellation, using 1616 to 1626.5 MHz, with service planned for 2026. Devices need radios built for those frequencies (Developing Telecoms). Deutsche Telekom signed on as a launch partner in September 2025 for IoT messaging and tracking (Via Satellite).

How the approaches compare

Sources for each cell are listed in the Sources section.

Where direct-to-cell is going next

Three things are in motion.

First, the FCC approved 7,500 more Gen2 Starlink satellites on January 9, 2026, bringing SpaceX’s authorized Gen2 total to 15,000 and allowing shells down to 340 km, with half of the new satellites required to be operational by December 2028 (RCR Wireless; Broadband Breakfast). A separate 15,000-satellite D2C constellation using the EchoStar spectrum is still pending.

Second, antennas are getting much bigger, the only lever that raises per-beam capacity without more spectrum. AST’s 223 square meter Block 2 is flying; SpaceX’s 250 square meter V2 is planned for Starship in 2027. Whether that makes satellite a real alternative to a tower or just a far better fallback is the question behind our comparison of Starlink Mobile vs the carriers.

Third, standards are catching up. 3GPP Release 17 defined the timing advance, Doppler pre-compensation and scheduling offsets that let future phones help correct for the satellite instead of leaving it all to the spacecraft (5G-MAG). As new modems ship with them, the two worlds will converge.

Sources

• Starlink Direct to Cell
• Starlink Space Safety: Constellation Altitudes
• Telecoms.com: Starlink indicates its satellite-to-phone service will drop next year
• FCC: Supplemental Coverage from Space Report and Order, draft for March 14, 2024 open meeting (DOC-400678A1)
• Covington / Global Policy Watch: FCC’s Supplemental Coverage from Space Rules Take Effect May 30
• TechCrunch: SpaceX gets FCC green light for Starlink direct-to-phone deal with T-Mobile
• Converge Digest: SpaceX gains FCC approval for Direct-to-Cell coverage with Gen2 Starlink
• RCR Wireless: SpaceX wins FCC’s approval for 7,500 additional Starlink satellites
• Broadband Breakfast: FCC Approves 7,500 More Starlink Gen2 Satellites
• Space Intel Report: SpaceX to FCC: we’ll use EchoStar mobile satellite spectrum to launch a 15,000-satellite constellation into a 330-km orbit
• EchoStar: EchoStar Announces Spectrum Sale and Commercial Agreement with SpaceX
• Aerospace America: Direct-to-device satellite internet sparks competing concepts
• KeepTrack: Starlink Direct to Cell deep dive
• KeepTrack: AST SpaceMobile BlueBirds deep dive
• NewSpace Tracker: Direct-to-smartphone satellites technology explainer
• TelecomHall: How SpaceX Direct to Cell will overcome technical challenges using the LTE standard
• Wiley / IEEE: Direct-to-Cell Technology: Enabling Satellite Connectivity for Legacy Devices
• 3GPP TR 38.821, section 4.2: Non-Terrestrial Networks reference scenarios (via iTecSpec)
• 3GPP TR 38.821, section 7.1.1: Delay (via iTecSpec)
• arXiv: 5G from Space: An Overview of 3GPP Non-Terrestrial Networks
• 5G-MAG: Non-Terrestrial Networks
• Wireless Estimator: Starlink sets in orbit six satellites as America’s towers remain a fixed stable for communications
• Gizmodo: SpaceX and T-Mobile send first texts via Starlink Direct to Cell satellites
• SDxCentral: Starlink Direct to Cell blasts off, puts cell towers in space
• N2YO: SpaceX’s new direct-to-cell Starlink satellites are way brighter than the originals
• SatelliteInternet.com: Starlink Direct to Cell and T-Satellite Guide
• SatelliteInternet.com: Starlink Direct to Cell provider page
• Cybernews: T-Mobile to launch Starlink-powered satellite data network this fall
• 5Gstore: T-Mobile T-Satellite Is Getting a Huge Upgrade
• Converge Digest: AST rolls out largest commercial LEO antenna to power 4G/5G
• Telecoms.com: AST SpaceMobile pledges 2026 service launch
• Nasdaq: AST SpaceMobile Gears Up for Its BlueBird 6 Launch Next Week
• AST SpaceMobile Q1 2026 Business Update (Form 8-K exhibit, SEC)
• AST SpaceMobile Q2 2026 Business Update (Form 8-K exhibit, SEC)
• Advanced Television: AST SpaceMobile readies next 3 BlueBird satellites
• PhoneArena: Half of AT&T customers can now access satellite services
• Light Reading: Lynk and Omnispace merger creates D2D challenger Elveo Mobile
• TelecomTV: Satellite comms pioneer Skylo expands its D2D ecosystem
• Deutsche Telekom: Telekom, Skylo and Qualcomm demonstrate SMS over satellite directly from smartphones
• Apple Support: Use satellite features on iPhone
• GSMArena: Apple invests $1.1B in satellite messaging provider Globalstar
• Universe Today: iPhone 14 will have satellite connectivity. How exactly it will work
• Developing Telecoms: Iridium to support direct 3GPP-based 5G IoT connections by 2026
• Via Satellite: Deutsche Telekom Partners with Iridium to Launch Integrated Satellite, Terrestrial 5G NTN Service

All sources accessed October 19, 2026.