When the International Telecommunication Union processed the first satellite filings of the space
age, the assumption was that geostationary orbit would remain the center of gravity for
commercial space. Half a century later, that assumption has not proven accurate, as spacecraft
have moved into low Earth orbit, and there are enormous numbers of them. If current filings are
even reasonably accurate, the number of satellites in orbit will become effectively
unmanageable, even with the best efforts at maneuvering, de-orbiting, and technological
advances. So, let’s look at the numbers themselves.
The reason operators stuck to GEO for five decades was that the GEO geostationary slot at
35,786 km covers roughly a third of the Earth from a single spacecraft. A LEO platform at 500
km sees perhaps a few hundred kilometers of useful footprint before geometry and link budget
push the user outside the beam. To deliver continuous global coverage from LEO, operators need
thousands of satellites with extremely low latency.
SpaceX alone now operates more than 10,000 Starlink satellites, or about two-thirds of those in
orbit. The next-largest operator, OneWeb, has fewer than 700. Between 2017 and 2022, national
regulators collectively filed requests with the ITU covering more than a million satellites across
300-plus constellation systems. Rwanda alone filed for 337,320. In January, SpaceX filed a
notification for roughly a million orbital data-center spacecraft, compute platforms intended to
run AI workloads in orbit. Blue Origin followed up with a 50,000-satellite data center filing, so
the projected satellite population for 2040 sits around half a million.
Every active LEO broadband satellite consumes downlink bandwidth in Ku, Ka, and,
increasingly, V-band, alongside feeder-link allocations that must coexist with terrestrial fixed-
service and radio-astronomy users. The ITU coordination process assumes a manageable number
of operators negotiating in good faith. It was not designed for a regime in which a handful of
players propose cumulative spacecraft populations exceeding the historical total by two orders of
magnitude.
The result is that interference disputes have become routine, with radio astronomy sites
documenting out-of-band emissions from LEO spacecraft at levels that compromise observations
across portions of the 10.7 to 12.7 GHz band. The massive Square Kilometer Array project has
raised these concerns, and SpaceX has cooperated by adjusting antenna patterns, but the
electromagnetic brightness of the sky will continue to climb.
As Ukraine has ably demonstrated, military use of Starlink has been a crucial element for
fighting off Russian advances. It uses Starlink for drone operations, artillery correction, and
tactical communications. The broader principle — that a commercial constellation can function
as wartime communications infrastructure — is now a fixed assumption in defense planning.
Not to be outdone, Russia has been developing counterspace systems intended to produce orbital
debris clouds against LEO targets, and it tested a direct-ascent ASAT against its own Kosmos-
1408 spacecraft in November 2021. That single event produced more than 1,500 trackable
fragments and forced the ISS crew to return to their capsules. That pales in comparison to
China’s 2007 Fengyun-1C test, which involved more than 3,000 tracked pieces, an estimated
150,000 smaller fragments, and a debris cloud at 865 km that will persist for decades.
The orbital data center filings from SpaceX and Blue Origin are another possible achievement,
and with them, some very big challenges. For instance, a compute platform in orbit must get data
in and out, which means either extremely high-throughput feeder links to ground stations or
dense optical inter-satellite links into an existing broadband constellation. There’s also a thermal
problem: while terrestrial AI data centers have well-understood water- and air-cooling
requirements, in LEO the only heat-rejection mechanism is radiation. So, it’s limited by available
surface area so for a platform dissipating megawatts of power, it must carry radiator panels that
can measure 170 m, according to Elon Musk.
Fortunately, Starlink-heavy altitudes of 500 km have one saving grace: atmospheric drag
eventually eliminates smaller fragments within 5 to 8 years. Spacecraft debris in higher orbits
doesn’t have this ability on any useful timescale, so if an event occurs at 800 to 1,000 km, where
the Fengyun-1C debris still exists, the resulting debris will effectively be permanent.
There are also environmental factors to consider, as a 550-lb. satellite releases about 66 lb. of
aluminum oxide nanoparticles during reentry. Multiply that across projected annual deorbit rates,
and aluminum oxide injection into the mesosphere and upper stratosphere becomes a major
problem. There are also potential issues with lithium, copper, and carbon-fiber combustion
products that researchers are just beginning to model.
It is important to keep in mind that the announced numbers—a million satellites, orbital data
centers, hundred-thousand-spacecraft national constellations are only filings that must undergo
review. History suggests the number of satellites deployed may be far less than this, but even if it
is, space is going to become a very crowded place in the decades to come.