The cost collapse that changed everything
In 2010, launching a kilogram of payload to low Earth orbit cost roughly $54,500. By 2024, SpaceX's Falcon 9 had driven that number below $2,700 — a 95% reduction. Starship, still in development, targets below $100/kg at scale.
This isn't incremental improvement. It's the kind of step-change that opens entirely new economic possibilities, the same way transistor cost reduction opened software.
The infrastructure layer is being built now
The orbital economy has three layers:
Launch — getting mass off Earth. SpaceX dominates, Rocket Lab captures small-sat launches, RocketLab, Relativity, ABL, and others compete for share. China's Long March family is increasingly commercial.
Satellite infrastructure — the pipes. Starlink (SpaceX), OneWeb (Eutelsat), and Amazon Kuiper are building competing broadband constellations. Planet Labs operates the largest commercial imaging fleet. Spire runs atmospheric sensors.
Applications — the actual services. Navigation, broadband, Earth observation, weather prediction, and increasingly, satellite-to-satellite networking.
The battle for the infrastructure layer is the interesting one. Whoever controls the satellite buses, the ground stations, and the spectrum has leverage over everything built on top.
Spectrum is the real constraint
You can build more rockets. You cannot create more radio frequency spectrum.
Spectrum is allocated by the ITU in Geneva through a coordination process that has, historically, moved at diplomatic speed. The emergence of mega-constellations (Starlink has 6,000+ satellites, targeting 42,000) has turned spectrum coordination into a geopolitical contest.
Countries filing spectrum claims for satellites they haven't built yet — a phenomenon called "paper satellites" — is now a recognized regulatory problem. The ITU's 2019 deadline rules were meant to address it. They haven't, fully.
What the orbital economy actually looks like
As of 2025, the orbital economy generates roughly $400B annually across:
- Satellite services (broadband, GPS): ~$200B
- Launch services: ~$15B
- Satellite manufacturing: ~$40B
- Ground equipment: ~$100B+
The growth drivers are broadband penetration in underserved markets (3B+ people without reliable internet access), precision agriculture and climate monitoring via imagery, and direct-to-device communications — the ability to ping a satellite from a phone without a specialized terminal.
Apple added satellite SOS to iPhone 14. AST SpaceMobile is building a network that can serve standard smartphones. The terrestrial and orbital networks are merging.
Defense and dual-use
The Russo-Ukrainian war became the first conflict in which commercial satellite infrastructure played a decisive tactical role. Starlink terminals enabled Ukrainian forces to maintain communications after traditional networks were degraded.
The U.S. DoD has spent years trying to reduce dependence on expensive, exquisite geosynchronous satellites in favor of proliferated LEO constellations that are harder to take out. The Space Development Agency (SDA) is building a 1,000+ satellite mesh for military broadband and missile tracking.
This dual-use nature creates complexity. Commercial operators don't want to be treated as military targets. Adversaries don't care.
The moon and beyond
Artemis (NASA), CLPS (commercial lunar payload services), and commercial entities like Astrobotic and ispace are building the lunar economy infrastructure — rovers, landers, eventually a cislunar GPS network.
The economic case for the moon is long-term: helium-3 for fusion (speculative), water ice at the poles for rocket propellant depots (nearer-term), and platinum-group metals from asteroids (much longer-term).
The more immediate value is operational: a propellant depot in lunar orbit changes the economics of deep space missions dramatically.
What to watch
The next five years will determine whether Starship achieves its cost targets (if it does, the orbital economy inflects again), whether Amazon Kuiper can credibly challenge Starlink, and how ITU spectrum policy responds to congestion.
The infrastructure layer is still being built. The applications layer hasn't been written yet.