Orbital Communications

LEO broadband constellations

Low Earth Orbit Communications Constellations

The market that ate the manifest.

The defining development of 2020s orbital communications is the mega-constellation — thousands of LEO satellites delivering broadband internet across the globe. Starlink dominates. With 12 million subscribers as of June 2026, over 10,400 active satellites, and $11.4 billion in 2025 connectivity revenue (61% of SpaceX’s total), it is the first satellite internet service to cross $10 billion in a single year. Its satellites represent roughly 65% of all operational spacecraft in orbit.

Behind Starlink, the field is crowded. Amazon Leo (formerly Project Kuiper) has about 392 production satellites in orbit and is racing an FCC deadline that requires 1,618 by July 30, 2026 — a deadline it has requested an extension for. Eutelsat’s OneWeb constellation is complete at 648 satellites with global coverage from 1,200 km, and Eutelsat has ordered 440 next-generation replacements from Airbus. China is deploying two parallel constellations — the state-backed Guowang (~190 satellites launched) and Shanghai-backed Qianfan (~200 satellites launched) — with a combined target exceeding 28,000 satellites.

Satellite counts are approximate and change weekly. Starlink figures from SpaceX S-1 SEC filing (May 2026) and Jonathan McDowell tracking. Amazon Leo figures from orbital catalog tracking. Chinese constellation counts from China in Space tracking. “Planned total” reflects current FCC/ITU filings, which operators frequently amend upward.

GEO incumbents

Geostationary Communications Satellites: The Legacy Fleet

The $18.7 billion market that LEO is compressing.

Before LEO constellations, all commercial satellite communications ran through geostationary orbit — 35,786 km above the equator, where a satellite’s orbital period matches Earth’s rotation and the satellite appears fixed over one point. GEO satellites cover a third of the planet each, so three can serve the whole globe. The trade-off is latency: a round trip to GEO and back takes roughly 600 ms, roughly 10× a LEO link. For decades, that latency was acceptable because there was no alternative.

There now is. The GEO market was valued at $18.7 billion in 2026 and is projected to grow at only 4% annually through 2030 — well below the 13–18% growth rates for LEO services. The incumbents are consolidating to survive.

The pattern is consolidation. SES acquired Intelsat. Eutelsat merged with OneWeb. Viasat acquired Inmarsat. EchoStar merged with DISH. No major GEO operator stands alone anymore. The survivors are building “multi-orbit” offerings that combine their GEO fleets with LEO access, either through owned constellations (Eutelsat/OneWeb) or partnerships (Viasat/Telesat, SES/OneWeb).

Direct to device

Satellite Direct-to-Cell and Direct-to-Device Services

Your phone connects to orbit without a dish.

The next frontier in orbital communications is direct-to-device (D2D) — satellite connectivity to unmodified smartphones without a ground terminal. The FCC established its Supplemental Coverage from Space (SCS) framework in March 2024, and commercial authorizations followed. Two architectures are competing.

SpaceX / T-Mobile. SpaceX adds D2C (direct-to-cell) hardware to the Starlink satellites it already mass-produces. T-Mobile’s T-Satellite service launched commercially July 2025, starting with messaging and adding broadband data in October 2025. Over 650 D2C-equipped Starlink satellites are now in orbit, serving 22 countries. Voice is in testing during 2026, with full voice and data tied to Starlink V2 satellites (targeted second half of 2027). ARPU is $10/month as an add-on. In August 2026, SpaceX signaled plans for a standalone “Starlink Mobile” consumer cellular service — competing directly with terrestrial carriers, not just filling their gaps.

AST SpaceMobile. The opposite architecture — fewer, much larger satellites (each with a 64 m² deployable phased-array antenna) aiming for full 5G broadband from orbit. Partnerships with AT&T, Verizon, and Vodafone. FCC authorized a 248-satellite constellation in April 2026. Seven BlueBird satellites in orbit as of early 2026, with plans for up to 60 in 2026 — delayed after the New Glenn pad explosion in May 2026 pushed the launch schedule. The bet: higher per-satellite bandwidth than Starlink’s D2C, at the cost of a much slower deployment cadence.

Apple / Globalstar. Emergency SOS and satellite messaging on recent iPhones via Globalstar’s existing LEO constellation. Amazon acquired Globalstar in April 2026 for its spectrum and ground infrastructure. Iridium, Skylo, and Lynk Global round out the field with narrower offerings.

The D2D market is evolving rapidly. SpaceX’s announcement of standalone “Starlink Mobile” service — competing directly with terrestrial carriers rather than partnering with them — is a potential inflection point for the entire telecommunications industry. If Starship enables mass deployment of high-bandwidth V2 satellites, the distinction between terrestrial and satellite cellular may dissolve.

Optical inter-satellite links

Laser Crosslinks: The Backbone in the Sky

The network routes through space, not through ground stations.

A satellite constellation without inter-satellite links is a set of independent relay towers — each satellite receives from the ground and transmits to the ground, and all routing happens terrestrially between gateway stations. Optical inter-satellite links (OISLs) change that. A laser crosslink between two satellites in orbit allows data to route through the constellation itself, hopping from satellite to satellite until it reaches the one closest to the destination. Starlink’s V2 satellites carry laser crosslinks, and the full mesh now spans the constellation. Amazon Leo’s satellites include 100 Gbps OISLs from launch.

The physics makes this faster than fiber for long distances. Light travels about 47% faster in vacuum than in glass fiber. A Starlink-routed path from London to Tokyo — hopping through the constellation — can beat the fastest submarine cable on latency. The advantage grows with distance: transcontinental and transoceanic routes see the largest gains. High-frequency trading firms have already begun exploring satellite links for latency-sensitive routes where microseconds matter.

OISLs also reduce dependence on ground infrastructure. A constellation with crosslinks needs fewer ground stations, because data can enter the network at one point and exit at another without touching the ground in between. This is operationally important for serving oceanic, polar, and remote coverage areas where ground stations are sparse or nonexistent.

The market

Satellite Communications Market Size and Structure

$34 billion in 2026, growing at 9–18% depending on what you count.

The satellite communications market was valued at roughly $34 billion in 2026. Market sizing varies by definition — whether you include ground equipment, launch services, or only connectivity revenue — but the structural picture is consistent: LEO services are growing at 15–18% annually, GEO services at 4%, and the center of gravity is shifting from fixed broadcast to mobile broadband and direct-to-device.

Starlink alone generated $11.4 billion in 2025 connectivity revenue and was running at a ~$17 billion annualized pace in Q2 2026. It is the only profitable segment inside SpaceX, which posted a $4.9 billion GAAP net loss in 2025 (with Starlink’s $4.4 billion operating income offset by Starship development costs). The satellite internet market overall was valued at roughly $17 billion in 2026 and is projected to reach $38 billion by 2031.

The verticals that matter are consumer broadband (Starlink’s core), enterprise connectivity, maritime, aviation in-flight connectivity (IFC), government and defense, and IoT. Aviation IFC is where GEO operators like Viasat and SES/Intelsat retain strong positions — an airliner crossing the Atlantic wants continuous coverage from a satellite that doesn’t move relative to the ground. Maritime is similar. Government and defense command premium pricing and are increasingly multi-orbit, mixing classified GEO capacity with LEO for resilience.

The GEO inversion

How LEO Displaced GEO Consumer Broadband

HughesNet filed for bankruptcy. That is the inversion.

In August 2026, HughesNet — the GEO satellite broadband pioneer that had served North American consumers for over two decades via the Jupiter GEO satellite series — filed Chapter 11 bankruptcy. The cause was not a technology failure. It was Starlink. GEO consumer broadband cannot compete with LEO on latency (600 ms vs. 20–50 ms), on throughput (Starlink’s per-user speeds routinely exceed 100 Mbps, a figure GEO systems struggle to match at scale), or on price trajectory (Starlink’s ARPU is falling because it can afford to, not because it has to). New customer acquisition for GEO consumer broadband has, in the words of one industry analysis, “fallen off a cliff wherever LEO is available.”

This is not the death of GEO. GEO satellites retain structural advantages for broadcast (one satellite covers a third of the planet), for aviation and maritime (continuous coverage without handoffs), and for government capacity (sovereign, hardened, positioned over theaters of interest). The ViaSat-3 series, each delivering 1 Tbps of throughput, serves these markets. SES’s O3b mPOWER MEO constellation serves enterprise and government. The inversion is specific: GEO lost consumer broadband to LEO, and the merger wave that followed — SES/Intelsat, Eutelsat/OneWeb, Viasat/Inmarsat, EchoStar/DISH — is the industry restructuring around that fact.

The launch market reflects this directly. The majority of global launch mass in 2025 was communications satellites, and the majority of those were Starlinks. The communications market does not merely use orbit — it is the economic engine that funds most of the launch cadence that makes everything else on Earth orbit possible.