On the morning of September 16, 2026, the Gravity-1 Y3 rocket lifted off from waters east of Shanghai and placed eight SpaceSail constellation satellites into orbit. China’s Ministry of Industry and Information Technology subsequently announced that 256 SpaceSail satellites had now been launched.[1] The data centers, fiber-optic cables and base stations underpinning terrestrial internet services are largely hidden from everyday view. Satellite internet moves part of that network into space—but why does it require hundreds or even thousands of satellites?
China SatNet is another major developer of satellite internet infrastructure in China.[10] Separately, the China National Space Administration said that the Satellite Internet Low-Orbit Group 24 satellites were successfully launched on August 16.[2] In an erratum issued in August, the U.S. Federal Communications Commission cited a SpaceX disclosure stating that Starlink had more than 10,200 broadband and direct-to-device communications satellites in orbit.[3]
SpaceSail’s cumulative launch total, the group numbers assigned to China’s low-orbit satellite internet missions and Starlink’s reported in-orbit fleet all reflect different stages of constellation deployment. They nevertheless point to the same question: How can a low-Earth orbit broadband network remain continuously available?
The answer can be understood by considering a ship at sea. The satellite serving it does not remain fixed overhead. It rises from one side of the horizon, crosses the sky and disappears over the other. The ship’s antenna must continuously hand the connection from one satellite to the next. Nearby ships, aircraft and islands may also be competing for limited communications resources. Allowing one location to connect occasionally and enabling many locations to remain reliably connected are networks of entirely different scales.
One Satellite Passes Overhead—and Then It Is Gone
The appeal of low-Earth orbit is straightforward. Because the satellites are closer to the ground, radio signals travel shorter distances, offering advantages in propagation delay and link loss compared with higher-altitude systems. But these satellites also move rapidly around Earth. A satellite in a circular orbit at an altitude of 550 km completes one orbit in about 95 minutes. A particular location on the ground can see it for only a small fraction of that time.[4]
Merely glimpsing a satellite above the horizon is not enough to support a stable communications link. Terminals generally require the satellite to exceed a minimum elevation angle, both to avoid buildings and terrain and to provide sufficient link margin against the longer propagation path encountered at low elevations.
Consider a simplified calculation. At an altitude of 550 km and a minimum elevation angle of 25 degrees, the usable ground-arc radius measured from the satellite’s subpoint is approximately 941 km. Lowering the threshold to 10 degrees expands that radius to about 1,664 km. The ground area that one satellite can effectively serve is therefore highly sensitive to terminal performance and link conditions.
Now assume that the satellite passes directly over the user and, for simplicity, ignore Earth’s rotation. Under the 550 km altitude and 25-degree elevation assumptions, a single usable pass lasts about four and a half minutes. A satellite that only skims the edge of the visible area would remain usable for even less time.
Looking only at handovers along the centerline of the orbital track, maintaining continuous coverage using satellites in the same orbital plane would require roughly 20 satellites spaced evenly around a 95-minute orbit. A real network also switches between satellites in different orbital planes.
Figure 1. How minimum elevation angle changes the usable coverage area of a single satellite. The 550 km altitude and 10-degree and 25-degree thresholds are illustrative assumptions used in this article. Source: Diagram produced for this article using spherical-Earth geometry.
The circles shown in the figure represent only geometric visibility. Data is actually delivered through beams generated by the satellite’s antennas. Payload capabilities and network scheduling determine where those beams are directed, how many can operate simultaneously and where frequencies can be reused.[5] A city may therefore lie inside a satellite’s geometric coverage area and still experience slow service because beam or capacity resources are insufficient.
The Next Satellite Must Be in the Right Place at the Right Time
Imagine that a terminal aboard a ship is downloading navigational charts. As the satellite currently serving it moves lower in the sky, the system must already know where the next satellite will appear so that the two links can be joined during the handover window.
Ephemeris predictions tell the terminal where to point. The antenna then tracks the satellite or electronically steers its beam, while the network transfers the session and routing path. If the handover is slightly delayed, a video call may briefly freeze. If the replacement satellite is not available, the connection will be interrupted.
A constellation must therefore maintain two kinds of spacing. Within each orbital plane, neighboring satellites must be positioned closely enough to take over from one another. Different orbital planes must also fill the east-west gaps between their coverage areas.
Orbital inclination determines the highest latitude reached by a satellite’s ground track, while Earth’s rotation continually moves ground users beneath different orbital planes. Concentrating large numbers of satellites in only a few planes may therefore be less effective than distributing them appropriately across the constellation.[4]
A historical change to Starlink’s deployment plan illustrates the scale involved. In 2019, the FCC approved SpaceX’s proposal to redistribute 1,584 satellites in the 550 km orbital shell from 24 orbital planes with 66 satellites each to 72 planes with 22 satellites each. The total number of satellites remained unchanged, but their distribution among orbital planes changed.[6]
With dozens of orbital planes and dozens of satellites in each plane, a single orbital shell can quickly reach more than 1,000 satellites. The quality of handovers and ground coverage depends not only on the total number of spacecraft but also on how they are arranged.
The required number of satellites also changes with the coverage objective. A constellation serving only a particular landmass can be designed around the target latitudes and coverage requirements, with beam and capacity allocation adjusted during peak demand.
If the network must also serve ships on the open ocean, polar air routes and mobile terminals worldwide, its orbital planes, inclinations and handover requirements will change accordingly. The required fleet size must be calculated by working backward from three questions: Where must service be available, when must it be available and how long can an interruption last?
Coverage Is Only the Beginning—Capacity Still Has to Be Calculated
Suppose 20,000 users in a coastal city are using satellite broadband simultaneously during the evening, each consuming an average of 5 megabits per second. Peak demand would reach 100 gigabits per second.
Now assume that, under the applicable beam, spectrum and interference conditions, each satellite can provide 10 gigabits per second of effective capacity to that area. Simply adding the required capacity would mean that approximately 10 satellites must serve the city at the same time. Those satellites must also take turns passing over the city while supporting users in other regions, pushing the total network size still higher.[5]
Another easily overlooked factor is the rest of the connection beyond the city. Data transmitted from a user to a satellite generally travels through a feeder link to a ground gateway before entering the wider internet. In systems equipped with inter-satellite links, traffic can instead be forwarded through space until it reaches a suitable gateway.
The additional user capacity created by deploying more satellites must ultimately be supported by feeder links and available gateways. Inter-satellite routing can direct traffic toward other landing points with spare capacity. The European Space Agency’s MATRIX research identified feeder links as a potential bottleneck in high-throughput satellite systems.[7]
Peak-hour performance is therefore determined by the most constrained segment of the chain: whether the terminal can track the satellite, whether the spacecraft has an available beam, whether spectrum can be reused in the same area, and whether feeder links and gateways can carry the traffic onward.
Inter-satellite links create more routing options, but radio resources on the user side must still be allocated beam by beam. Operating a low-Earth orbit network is fundamentally an exercise in real-time scheduling across a constantly moving topology.
Even After a Constellation Is Built, It Must Be Replenished Every Year
If low-Earth orbit requires so many satellites, why not place them higher?
Raising the orbital altitude expands the geometric area visible to each satellite. Satellites in geostationary orbit can also remain nearly fixed relative to the ground, allowing a small number of spacecraft to provide broad communications coverage. The longer propagation distance, however, imposes a different set of constraints on low-latency services and compact user terminals. Low-Earth, medium-Earth and geostationary orbits each serve different purposes.[4][9]
The size of a low-Earth orbit constellation is also not fixed once its first generation has been launched. Satellites age, fail and retire. Operators must maintain their orbital positions, launch replacements and introduce upgraded spacecraft while minimizing disruptions to existing services.
At the same time, increasingly crowded orbital regions create more work involving conjunction warnings, collision avoidance and end-of-life disposal. The European Space Agency’s Space Environment Report 2026, released in September 2026 using data through the end of 2025, identified congestion in low-Earth orbit as a growing and persistent engineering challenge.[8]
The number of satellites launched is therefore only a starting point for evaluating a network. More useful metrics include how many hours per day the target land and ocean areas have service, how long connections are interrupted during satellite handovers, how much bandwidth each user actually receives during urban peak periods, and whether service can continue after the loss of a satellite or gateway.
Those answers can only come from operating the network in orbit and testing it with ground terminals.
Returning to the day’s launch, each new batch of satellites adds more nodes to a network that is constantly in motion. To keep video streaming aboard ships, prevent aircraft terminals from losing their connections and maintain usable speeds for suburban customers during the evening peak, those nodes must appear in the right orbit at the right time—and the rest of the communications chain must be able to carry the traffic.
Low-Earth orbit constellations need so many satellites because continuous broadband service with short transmission distances and low latency creates overlapping requirements for coverage, time and capacity—and each layer adds to the total.
Building a LEO network requires more than adding satellites; it also depends on payloads, ground systems, testing infrastructure and system-level integration. China’s expanding commercial space industry is bringing greater manufacturing capacity and more competitively priced options to the global market, and international customers can contact STARPATH GLOBAL to discuss their constellation requirements. The team can help assess suitable Chinese satellites, payloads, AIT equipment and integrated solutions for different technical requirements and budgets.
References
[1] Radio Regulation Bureau, Ministry of Industry and Information Technology, “SpaceSail Constellation Networking Satellites Successfully Launched, Bringing Total to 256,” September 16, 2026.
[2] China National Space Administration, “China Successfully Launches Satellite Internet Low-Orbit Group 24 Satellites,” August 17, 2026.
[3] U.S. Federal Communications Commission, Erratum to the 2026 Section 706 Report, August 18, 2026; the in-orbit figure was cited from SpaceX’s August 2026 quarterly report.
[4] European Space Agency, “Orbits”; satellite passes and continuous coverage for low-Earth orbit communications systems.
[5] ITU-R Report M.2460-0, 2019; non-geostationary-orbit coverage, spot beams and frequency reuse.
[6] U.S. Federal Communications Commission, “Space Exploration Holdings, LLC—Request for Modification of the Authorization for the SpaceX NGSO Satellite System,” DA 19-1294, December 19, 2019, pp. 1–2.
[7] European Space Agency, “Multiple Access Telecom Reconfigurable Inter-satellites X-factor”; feeder-link bottlenecks.
[8] European Space Agency, ESA Space Environment Report 2026, September 14, 2026; data through the end of 2025.
[9] European Space Agency, “Types of Orbits”; geostationary-orbit coverage and position relative to the ground.
[10] State-owned Assets Supervision and Administration Commission of the State Council, “Hao Peng Calls for Accelerated Development of a World-Class Satellite Internet Company During Visit to China SatNet,” August 27, 2021; China SatNet’s business positioning.
