Amazon has applied for United States regulatory approval to deploy as many as 5,105 low-Earth-orbit satellites capable of connecting directly to mobile phones, extending its contest with Elon Musk’s Starlink from fixed broadband into conventional cellular services.

The proposed Amazon Leo Direct-to-Device network would support voice calls, messages, data and emergency communications in places where terrestrial towers are unavailable or disabled.

The filing is an application, not an operating licence.

Amazon plans to begin deployment in 2028, subject to approval from the Federal Communications Commission and additional authorisations in every country where it intends to offer service.

Building the constellation, securing international spectrum rights and integrating it with mobile operators would take several years.

The new system is separate from, but designed to complement, Amazon’s existing Leo broadband constellation, formerly called Project Kuiper.

That network connects users through purpose-built antennas and has regulatory approval for 3,236 satellites.

Amazon says more than 390 are now in orbit, enough to begin introducing fixed service within its initial coverage bands.

Starlink, by comparison, operates more than 10,000 satellites across its much larger network and already serves customers in more than 150 countries.

Direct-to-device technology addresses a different market.

Instead of requiring a satellite dish or dedicated terminal, it allows a compatible smartphone, wearable device or remote sensor to communicate with an orbiting satellite.

The satellite effectively acts as a cellular base station in space, filling gaps between terrestrial networks rather than necessarily replacing them in towns and cities, where ground infrastructure provides greater capacity at lower cost.

Amazon intends to distribute the proposed satellites among five orbital shells between 510 and 580 kilometres above Earth.

Three would concentrate capacity over densely populated mid-latitude regions, while two higher-inclination shells would extend service towards high latitudes and polar areas.

This arrangement is intended to provide global reach while placing more capacity over regions where demand is greatest.

Compatible devices would communicate with the satellites through L-band and S-band radio frequencies.

Separate Ka-band and V-band connections would carry aggregated traffic between orbit and Amazon’s ground infrastructure.

Optical links between satellites would allow data to be routed through the constellation by laser before reaching a suitable ground station.

The spacecraft would process mobile signals in orbit rather than merely retransmitting them unchanged.

Amazon says digital beamforming, beam-hopping, dual-polarisation reception and adaptive coding would help extract usable connections from the comparatively weak signals produced by handheld devices.

Those technologies could also direct capacity towards temporary demand, such as a disaster zone or a remote industrial operation, while limiting interference with neighbouring frequency users.

Real-world performance will depend on available spectrum, the number of satellites overhead, local regulation, handset design and agreements with network operators.

Satellite links remain constrained by power and capacity, meaning early services may prioritise emergency messages and basic connectivity before approaching the speed or consistency of terrestrial mobile broadband.

Buildings, terrain and foliage can also obstruct the clear view of the sky required for reliable communication.

Amazon plans to work with mobile-network operators rather than sell every service independently.

That approach would let carriers extend existing coverage into rural districts, oceans, transport corridors and disaster areas without constructing towers across every location.

It could also allow customers to retain their telephone numbers, billing relationships and service plans when their devices switch between ground and satellite networks.

The proposal follows Amazon’s agreement to acquire Globalstar for approximately $11.6 billion.

The transaction, expected to close in 2027 subject to regulatory approval, would give Amazon Globalstar’s satellites, ground infrastructure, technical operations and valuable mobile-satellite spectrum rights.

Those frequencies are central to the direct-to-device plan because usable spectrum is scarce and heavily regulated.

Globalstar already provides satellite connectivity for supported Apple devices, including emergency messaging, roadside assistance, location sharing and other safety functions.

Amazon has agreed to continue supporting compatible iPhone and Apple Watch models and to collaborate with Apple on future services using the expanded network.

This gives Amazon an established consumer relationship while its own constellation remains under development.

Starlink holds the strongest position in the emerging market.

Its parent, SpaceX, controls both satellite production and a reusable launch fleet, allowing it to deploy spacecraft at a cadence rivals have struggled to match.

Starlink has partnered with telecommunications companies including T-Mobile in the United States and operators in Europe and other regions to connect ordinary mobile devices outside ground coverage.

Amazon lacks an equivalent operational launch system under its direct corporate control.

It has contracted launches from several providers, including SpaceX, United Launch Alliance, Arianespace and Blue Origin, the space company founded by Amazon founder Jeff Bezos.

Using multiple providers reduces dependence on a single rocket, but launch availability, vehicle delays and cost remain material constraints.

Amazon previously required additional time to satisfy deployment milestones attached to its original broadband authorisation.

Competition is also expanding beyond Amazon and Starlink.

AST SpaceMobile is developing direct connections between standard phones and large satellites through partnerships with mobile operators, while established satellite companies are modernising older networks.

Governments increasingly view these systems as strategic infrastructure because they can preserve communications during natural disasters, military operations, cyberattacks and failures of terrestrial networks.

A constellation of this scale will face scrutiny over orbital congestion, collision avoidance, atmospheric re-entry, astronomical interference and the cumulative environmental cost of manufacturing and launching thousands of short-lived spacecraft.

Amazon must demonstrate that its satellites can coexist with other operators’ frequencies, manoeuvre away from collision risks and be removed safely when they reach the end of their service lives.

The competitive significance is broader than connecting isolated mobile users.

Amazon could combine satellite access with its cloud-computing, logistics, retail and device businesses, while Starlink remains integrated with SpaceX’s launch and spacecraft operations.

Regulatory review will now determine whether Amazon may begin constructing the 5,105-satellite system for deployment from 2028.