Orbital Shells 101: Why LEO, MEO, and GEO Are Not Interchangeable

2026-06-09 · by Mira Okafor · Senior Editor · Orbital Mechanics · Article

When people say a satellite is "in orbit," they almost never mean the same thing. A Starlink satellite at 550 km and a DirecTV satellite at 35,786 km are both in orbit. They are not interchangeable, and the reason tells you almost everything about how the modern space industry is organized.

This article walks through the orbital shells — the altitude bands that the space industry sorts satellites into — and explains, in plain terms, why each shell exists, what lives there, and what the engineering tradeoffs are.

The shape of the problem

Every orbit is a tradeoff. The two biggest knobs are altitude and inclination, and they interact with three physical constraints:

Gravity drops with altitude. Higher orbits are slower, and satellites at higher altitudes take longer to circle the Earth.
The atmosphere doesn't end at 100 km. It fades out gradually. Below ~400 km, atmospheric drag is large enough that satellites deorbit in months. Above ~800 km, drag becomes negligible for most practical purposes.
Radio line-of-sight is a function of altitude and Earth curvature. A satellite at 35,786 km can see ~42% of the Earth's surface at any moment. A satellite at 550 km can see ~2%.

Those three constraints, plus the obvious one of launch cost per kg to altitude, are why the shells exist. Different missions have different optimal altitudes. The shells are the industry clustering the optimums.

The shells, at a glance

|---|---|---|---|---|

The rest of the article walks through each shell with examples and tradeoffs.

LEO — Low Earth Orbit

LEO is the workhorse shell. Almost every commercial constellation lives here, and so does the International Space Station.

The defining characteristics:

Altitude: 400 to 2,000 km. The lower bound is set by drag; the upper bound is more a convention than a hard limit (the atmosphere is functionally gone above ~600 km).
Orbital period: 88 to 127 minutes. Starlink satellites at 550 km circle the Earth in about 95 minutes.
Speed: About 7.8 km/s at 400 km, about 6.9 km/s at 1,200 km. Lower altitude = faster.
Coverage: A single LEO satellite sees a "footprint" of about 2% of the Earth's surface. To cover the planet continuously, you need a constellation.
Latency: Round-trip signal time is 20-50 milliseconds, which is the key reason LEO broadband is competitive with fiber for some use cases.

LEO is the most crowded shell by number of satellites. As of mid-2026:

Starlink has ~7,500 satellites, mostly in the 540-570 km range.
Amazon Leo (formerly Project Kuiper) has 330+ satellites on orbit, with planned shells at 590, 610, and 630 km.
ISS orbits at ~420 km.
Hubble at 540 km.
Planet Labs, ICEYE, Capella, BlackSky have hundreds of Earth-observation satellites between 450-600 km.

Why LEO wins for broadband: low latency. A signal goes up to a Starlink satellite and back down to a ground station in under 30 ms. GEO takes 600+ ms round trip, which is fine for video and voice, but bad for gaming, video conferencing, and high-frequency trading. LEO wins on user experience.

The cost of LEO: you need a lot of satellites. Starlink's 7,500-strong constellation is the only way to give continuous global coverage at 550 km. Amazon Leo's planned 3,236-satellite constellation is similarly sized.

The new worry in LEO: collision risk. With 10,000+ active satellites in LEO by 2027, the conjunction rate (the rate of "close approaches" that need human intervention) is going up fast. The FCC's 5-year deorbit rule is the regulatory response — we have a separate piece on that.

MEO — Medium Earth Orbit

MEO is the navigation shell. GPS, Galileo, GLONASS, BeiDou, and the upcoming QZSS regional system all live here.

Altitude: 2,000 to 35,786 km. Most navigation constellations cluster in the 19,000-23,000 km range.
Orbital period: 2 to 24 hours. GPS satellites at 20,200 km have a 12-hour period, which means they complete exactly two orbits per sidereal day.
Latency: Round-trip signal time is ~70-130 ms.
Coverage: Each MEO satellite sees a much larger footprint than a LEO satellite — about 38% of the Earth's surface at 20,000 km.

Why MEO wins for navigation: a small number of satellites at MEO can give continuous global coverage. The full GPS constellation is 31 operational satellites, all at ~20,200 km. Your phone receives signals from at least 4 of them at any time from almost any point on Earth. You can do the same with LEO satellites, but the satellites would be in different positions every minute and the math is much messier.

The new interest in MEO: broadband. A handful of startups (O3b mPOWER, others) are putting broadband constellations in MEO to get the latency improvement over GEO without the constellation count of LEO. It is a niche bet, but the math works for trunk data and maritime markets.

GEO — Geostationary Orbit

GEO is a specific orbit, not a shell in the same sense as LEO and MEO. It is the circular orbit at 35,786 km altitude directly above the equator, where a satellite's orbital period exactly matches the Earth's rotation period. From the ground, the satellite appears to hang motionless in the sky.

Altitude: 35,786 km (circular, zero inclination, equatorial).
Orbital period: exactly 23 hours 56 minutes (one sidereal day).
Latency: Round-trip signal time is ~600 ms. This is the key limitation of GEO.
Coverage: A single GEO satellite can see ~42% of the Earth's surface — basically an entire hemisphere, with overlap for redundancy.

Why GEO wins for TV, weather, and fixed comms: a satellite that doesn't move is a huge simplification for ground antennas. A DirecTV dish on your house can be mounted once and pointed at one spot. A weather satellite can take images of the same area every 15 minutes, simplifying forecasting models. A defense early-warning satellite in GEO can watch an entire continent for missile launches.

The famous GEO positions: there are about 500 operational GEO satellites, clustered around a small number of "popular" longitude slots. The slot at 105° West is shared by several major satellites; the slot at 0° (over the equator off the coast of Africa) is the canonical reference.

The cost of GEO: launching to GEO requires either a heavy-lift rocket with a high-energy upper stage, or a smaller rocket plus a multi-month electric-orbit-raising burn. Either way, launch cost to GEO is 3-5x the cost to LEO. The satellites themselves are also heavier, more expensive, and built for 15-20 year lifetimes.

The new worry in GEO: crowding. The GEO belt is finite (you cannot place two satellites in exactly the same slot — radio interference). The FCC and the ITU manage slot allocation, but the demand is high.

HEO — Highly Elliptical Orbits

HEO is a set of orbits, not a single shell, but the most famous type is the Molniya orbit, used by Russia for decades and now used by several other countries for high-latitude coverage.

Shape: Highly elliptical. The satellite spends most of its time near apogee (the highest point) and zooms quickly through perigee (the lowest point).
Apogee: Typically 35,000-50,000 km, often over the Northern Hemisphere.
Period: Usually 12 hours (half-sidereal-day), so the satellite appears to spend most of its time over the same part of the Earth.

Why HEO exists: GEO satellites sit over the equator. From Moscow, a GEO satellite is at a very low elevation angle and gets blocked by terrain. A Molniya satellite with a 12-hour period and the right argument of perigee hangs over Russia for ~8 hours out of every 12. The Soviet Union pioneered this, and Russia, Canada, Japan, and others use variants.

HEO is a small but persistent niche. New HEO constellations are being proposed for Arctic broadband, including some commercial ventures.

Cislunar and beyond

Outside GEO, the shells stop being well-defined. The space between GEO and the Moon (~384,000 km) is cislunar space, and it includes:

NRHO (Near-Rectilinear Halo Orbit) — the orbit NASA picked for the Lunar Gateway. Stable with low propellant cost.
DRO (Distant Retrograde Orbit) — a stable orbit far from the Moon, used by some planned science missions.
Lagrange points — five special points where the gravitational pull of two large bodies (Earth-Sun, Earth-Moon) balances with centripetal force, allowing a satellite to "hover" with minimal station-keeping. L1 (between Earth and Sun) is home to solar observatories like SOHO. L2 (opposite side of Earth from the Sun) is where James Webb Space Telescope and Euclid orbit.

These are not "shells" in the same way as LEO, MEO, and GEO, but they are increasingly important as the commercial space economy pushes outward.

Putting a satellite in a shell — the real-world process

When a company says "we are putting a satellite in LEO," the actual sequence is:

Choose an altitude and inclination. Altitude sets the latency, period, and coverage. Inclination sets which latitudes you pass over. A satellite at 0° inclination (equatorial) never sees the poles. A satellite at 90° inclination (polar) passes over every latitude twice per orbit.
Choose a launch vehicle. This is mostly a function of mass to the target orbit. Starship V3 and Falcon Heavy are the only US vehicles that can do large GEO payloads. Smaller rockets can do small GEO payloads with electric orbit-raising, or do LEO constellations in batches.
Reserve a slot or trajectory. If you are going to GEO, you need a slot assignment from the ITU (international) and the FCC (US). If you are going to LEO, you need to file a constellation plan with the FCC (US) and equivalent regulators elsewhere, plus notify other operators of your proposed shells.
Launch and station-keep. Once on orbit, the satellite is not "done" — drag (in LEO), gravitational perturbations (in MEO and GEO), and the pull of the Moon and Sun all move the orbit. Operators use small thrusters to maintain the satellite in its assigned slot or shell.
Deorbit or graveyard at end of life. In LEO, the FCC now requires operators to deorbit satellites within 5 years of end of mission. In GEO, end-of-life satellites are pushed ~300 km above GEO into a "graveyard orbit" so they don't interfere with active satellites.

Why this matters in 2026

Two converging stories make orbital shells a hot topic in 2026:

The LEO broadband buildout is in full swing. Starlink is past 7,500 satellites. Amazon Leo is at 330+ and accelerating. China's Guowang constellation is at several hundred. The LEO shell between 500 and 650 km is now the most populated single orbital regime in history, and the conjunction rate is rising fast.
The FCC's 5-year deorbit rule is now in active enforcement. Operators are being required to demonstrate end-of-life plans within 5 years (down from the older 25-year guideline). This changes the unit economics of LEO constellations — a Starlink satellite that used to amortize over 10 years now has to pay back in 5.

The shells are not abstract. The regulatory, economic, and engineering choices that put a satellite in one shell versus another are the choices that determine whether a commercial space business works.

Try it

Want to feel what happens when you put a satellite in the wrong shell? The Orbital Puzzle sim drops a satellite into LEO, MEO, or GEO and lets you see the speed, period, and coverage difference for yourself. A communications satellite at 600 km feels very different from a communications satellite at 36,000 km.

Orbital Puzzle — drag a satellite into the wrong shell and see what happens →