Blog · May 2026 · By Craig Miles · 8 min read

Four Technical Approaches to Direct-to-Device Satellite Services Explained

Not all direct-to-device satellite services work the same way. The phrase direct-to-device, or D2D, has become shorthand for any satellite service that connects directly to an ordinary handset without specialist hardware. But underneath that umbrella are four distinct technical architectures, each making different bets about spectrum, satellite design, and where the intelligence in the system should live.

In March 2026, the International Telecommunication Union hosted its first Space Connect webinar, bringing together four of the leading D2D providers to describe their approaches. The session made clear that D2D is not a single technology. It is a market with multiple competing visions of how to solve the same problem — and the engineering choices behind each vision have significant consequences for coverage, capacity, regulation, and handset compatibility. I have been tracking this space closely since my 2023 TEDx talk on LEO satellite connectivity, and the pace of change since then has been considerable.

This article explains each of the four main D2D technical approaches, what makes them different, and why those differences matter.

Why Technical Architecture Matters in D2D

Before examining each approach, it is worth understanding why the underlying architecture matters at all. In a D2D service, the satellite must receive a signal from a standard smartphone — a device with a small antenna, low transmit power, and no specialist RF hardware. That is an extremely difficult link budget problem.

Different operators have chosen different ways to solve it. Some have built larger satellites with bigger antennas to compensate for the weak signal from the handset. Others have kept the satellite simple and put the intelligence on the ground. Some have used spectrum already associated with satellite use; others have struck deals with mobile network operators to reuse terrestrial cellular spectrum. Each of these choices flows through the entire system design.

Approach 1: Traditional MSS Bent-Pipe (Globalstar)

Globalstar represents the most established D2D architecture, and in some respects the most conservative one. It operates on mobile-satellite service spectrum in the L-band (around 1.6 GHz uplink) and S-band (around 2.4 GHz downlink) — frequencies that have been associated with satellite use for decades.

The defining characteristic of Globalstar’s architecture is bent-pipe design. In a bent-pipe satellite, the spacecraft acts purely as a relay — it receives signals from the ground, shifts the frequency, and re-transmits them back down without any onboard processing. The intelligence, the routing logic, and the software all sit at Globalstar’s ground stations.

This has a significant practical advantage: the satellite itself is relatively simple, and upgrades to the system can be made at the ground station without launching new hardware. As Globalstar’s General Counsel Barbee Ponder noted at the ITU Space Connect session, the company is able to adapt quickly to technological innovation simply by upgrading its ground-based electronics.

The trade-off is geographic limitation. Because the signal must reach a ground gateway on both the uplink and downlink path, the user must be within range of a gateway, which limits coverage compared with fully regenerative systems.

In practice, Globalstar’s most visible D2D deployment is through Apple. iPhones from the iPhone 14 series onwards include embedded chipsets that can access Globalstar’s network, initially for emergency SOS messaging and later for broader satellite messaging. This is proprietary — Apple’s iPhones cannot use other D2D networks through this mechanism — but it demonstrated that mass-market consumer D2D was viable and kickstarted the broader market.

Approach 2: Large-Aperture LEO Regenerative (AST SpaceMobile)

AST SpaceMobile has taken the most hardware-intensive approach to D2D, and in many ways the most technically ambitious. Rather than keeping the satellite simple, AST has built the largest commercial communications arrays ever deployed in low Earth orbit, with the explicit goal of delivering broadband-grade services to completely standard, unmodified smartphones.

The company’s next-generation BlueBird Block 2 satellites — with BlueBird 6 launched in December 2025 and BlueBird 7 in early 2026 — each feature a phased array antenna spanning nearly 2,400 square feet, more than three times the size of the first-generation BlueBirds. Each satellite is powered by a proprietary AST5000 application-specific integrated circuit that supports 10 GHz of processing bandwidth, peak speeds of up to 120 Mbps per coverage cell, and more than 2,000 individual cells per satellite.

This is a regenerative architecture: unlike Globalstar’s bent-pipe design, the satellite processes the signal onboard rather than simply relaying it. That removes the dependency on a ground gateway being within the signal path and enables genuinely global coverage — or at least coverage wherever the satellite passes overhead.

The enormous aperture compensates directly for the weak signal from a standard handset. More antenna area means more gain, which closes the link budget without requiring the phone to transmit at higher power or use specialist hardware. AST SpaceMobile’s Jennifer Manner noted at the ITU session that the large arrays allow the system to precisely shape cells and respect national borders, which is important for regulatory coordination.

AST SpaceMobile is targeting 45 to 60 satellites in orbit by the end of 2026 to provide continuous coverage across the United States and selected international markets, including the UK through its partnership with VodafoneThree. It uses terrestrial mobile spectrum through MNO partnership agreements — a deliberate choice to sit inside the existing mobile ecosystem rather than operate as a separate satellite service.

Approach 3: Asset-Light GEO Software-Defined (Skylo)

Skylo represents perhaps the most unconventional approach. Rather than building or launching its own satellite constellation, Skylo has built a software platform that runs on top of existing geostationary satellites operated by other providers. Its strategy is asset-light by design.

The Skylo approach uses NB-IoT NTN — narrowband IoT over non-terrestrial networks — as defined in 3GPP Release 17. Because the platform is built on an open, standardised protocol, devices that are 3GPP NTN compliant can use Skylo’s service without bespoke hardware. Marko Keskinen, Managing Director for Europe at Skylo, put this clearly at the ITU session: if devices already comply with 3GPP NTN standards and use a standard chipset, they should be treated as standard consumer devices.

The trade-off for Skylo’s asset-light approach is performance. Geostationary satellites orbit at approximately 35,786 km, compared with the 500 to 600 km typical of LEO constellations from AST and SpaceX. That altitude means higher latency — around 600 milliseconds round-trip — and more path loss, which limits the service to lower data rate applications. In practice, Skylo focuses primarily on IoT and messaging use cases rather than voice or broadband.

The strategic advantage is speed and flexibility. Skylo can scale globally by partnering with satellite operators who already have spectrum rights and orbital slots, without the capital expenditure or launch risk associated with building a new constellation. The company describes itself as software-defined, cloud-native, and modular — characteristics that allow it to move at software development speed rather than hardware development speed.

Approach 4: Hybrid IMT and MSS LEO (SpaceX Starlink)

SpaceX is pursuing D2D as an extension of its existing Starlink broadband constellation, which gives it a scale advantage no other D2D operator currently matches. Starlink’s LEO network already consists of thousands of satellites, providing the orbital coverage density that enables a continuous service rather than intermittent pass-based access.

SpaceX’s D2D approach is deliberately hybrid: the company seeks to use both terrestrial mobile spectrum (IMT) and MSS spectrum depending on the market and the regulatory environment. As SpaceX’s Spectrum and Regulatory Affairs Manager Udrivolf Pica described at the ITU session, Starlink’s current satellites function as towers in the sky, using existing gateway infrastructure deployed around the world to support both broadband and direct-to-cell service.

In the UK, SpaceX is working with Virgin Media O2 (VMO2), which received the first Ofcom D2D licence variation in February 2026 under the new authorisation framework. In the United States, T-Mobile and SpaceX launched the first nationwide D2D messaging service in mid-2025, available to any compatible smartphone on any US carrier. Voice and data services have followed as the constellation has matured.

The Starlink D2D service does not require specialised hardware on the handset. Compatible smartphones — which increasingly means most modern Android and Apple devices depending on chipset — connect to the Starlink constellation through standard cellular signalling, appearing as a roaming partner to the device.

The challenge for SpaceX’s hybrid approach is regulatory complexity. Using terrestrial mobile spectrum from LEO satellites creates interference management obligations with co-channel and adjacent-channel users on the ground, which is why the Ofcom framework requires MNOs to hold a specific licence variation and comply with detailed technical conditions before offering the service.

Comparing the Four Approaches

The Spectrum Boundary Problem

One theme that ran through the ITU Space Connect session was the extent to which D2D is blurring previously distinct regulatory categories. Session moderator Patricia Cooper summarised it directly: D2D is blurring spectrum allocation lines.

Traditional satellite services operated in clearly defined MSS spectrum, separate from terrestrial mobile spectrum. D2D — particularly the AST SpaceMobile and SpaceX models — deliberately crosses that boundary by using terrestrial mobile spectrum from orbit. That creates interference coordination obligations, new regulatory frameworks, and the need for agreements between satellite operators and mobile network operators.

This is not just a UK issue. The ITU’s next World Radiocommunication Conference (WRC-27) will address spectrum policies directly relevant to D2D, and how governments resolve competing claims over mobile spectrum between terrestrial and non-terrestrial users will shape the industry for the following four years.

What This Means for Handset Compatibility

For engineers and procurement professionals, one of the most practical questions is which handsets can use which D2D services. The answer varies by approach.

Globalstar’s Apple integration requires specific embedded chipsets, which means it is currently limited to iPhones from iPhone 14 onwards and is not available to Android devices through this path.

Skylo’s 3GPP NTN approach offers the broadest potential compatibility: any device with a compliant NTN chipset should be able to access a 3GPP-standard NTN service. Skylo’s argument — that NTN-compliant devices should be treated as standard consumer devices — is essentially an argument for removing any additional regulatory burden on the handset side.

AST SpaceMobile and SpaceX both target standard unmodified smartphones, but the specific devices that can use their services depend on chipset support and software enablement. Not every current smartphone can connect, although as chipset generations turn over the compatible base is expanding.

Why Four Approaches Are Likely to Coexist

It would be tempting to look at these four architectures and expect one of them to win. That is probably the wrong frame. The more likely outcome, consistent with what the ITU session suggested, is that different D2D architectures will serve different use cases, different regulatory environments, and different parts of the market.

Skylo’s low-rate IoT model serves a category that high-bandwidth LEO approaches are not optimised for. Globalstar’s MSS model has decades of spectrum rights and a major handset OEM as an anchor customer. AST SpaceMobile and SpaceX are competing for the broadband and voice segment, but through different constellations and at different scales.

As with most mature communications markets, the outcome is likely to be a layer cake rather than a single winner — with different services operating at different altitudes, in different spectrum bands, and through different commercial relationships, combining to provide coverage where no single approach currently reaches alone.

Final Thoughts

The four main technical approaches to direct-to-device satellite are the traditional MSS bent-pipe model pioneered by Globalstar, the large-aperture regenerative LEO model pursued by AST SpaceMobile, the asset-light GEO software-defined model used by Skylo, and the hybrid IMT and MSS LEO model deployed by SpaceX. Each solves the D2D link budget problem differently, each makes different regulatory and spectrum choices, and each targets a different part of the market.

What unites them is the goal: seamless satellite connectivity to mass-market devices, in places and at times when terrestrial networks cannot reach. That goal is now close enough to delivery that the ITU is already planning WRC-27 spectrum policy around it. D2D has moved from experiment to infrastructure.