Almost every claim about the commercialisation of space reduces to cost per kilogram delivered to orbit, and almost every attempt to reduce that cost runs into the same equation. Understanding why launch was expensive for fifty years, and what specifically changed, explains both the genuine shift that has occurred and the parts of the field where nothing has changed at all.
The equation that sets the budget
A rocket accelerates by expelling mass, and the relationship between the velocity change it can achieve, the efficiency of its engine, and the ratio of fuelled to empty mass is logarithmic. The logarithm is the problem. A modest velocity change needs modest propellant. A large one needs an amount that grows exponentially, because every additional kilogram of propellant must itself be accelerated by the propellant beneath it.
Reaching a stable low orbit requires roughly nine and a half kilometres per second once gravity losses, drag, and steering are accounted for. Orbital velocity itself is less than that, and the difference is energy spent fighting the atmosphere and gravity on the way up. Against the efficiency achievable from chemical propellants, that leaves an unforgiving mass budget.
A launch vehicle is overwhelmingly propellant by mass, its structure is a small fraction, and the payload is a small fraction of that. Staging exists because of this, since carrying empty tanks to orbit wastes performance, and the cost is that hardware is discarded on every flight.
What reuse actually changes
Recovering a first stage changes none of the physics above. It changes which costs recur.
The penalty is paid in payload. Propellant reserved for slowing, re-entering, and landing is propellant not used to accelerate the payload, and landing hardware adds mass carried the whole way. Returning a stage to its launch site costs more performance than landing downrange on a ship, because reversing horizontal velocity is expensive, which is why the recovery method is chosen per mission according to the margin the payload leaves.
Flight rate decides the economics
The gain is that the most expensive hardware is no longer thrown away, and whether that is worth the penalty depends almost entirely on one variable that is not technical. If a recovered stage flies many times, manufacturing cost spreads across those flights and the economics work. If it flies twice they do not, because refurbishment, inspection, and recovery operations are not free while the payload penalty is paid every flight.
Flight rate is therefore the pivotal number, and a reusable vehicle flying rarely is more expensive than an expendable one. This is why the shift toward reuse arrived together with a large increase in cadence, and why the two cannot be separated when assessing whether it worked.
A change in how things are bought
A structural change ran alongside the technical one, and it is easy to overlook because it is contractual rather than physical. Traditional government procurement reimbursed the contractor for costs incurred plus a fee, which suits programmes where the requirement is unprecedented and cost genuinely cannot be estimated, and which contains no incentive to reduce cost since a lower cost yields a lower fee.
The alternative, applied first to cargo and later to crew transport, was to specify an outcome, pay a fixed price, and let the supplier own the vehicle and the design. The supplier absorbs overruns, keeps savings, and can sell the same service to other customers. That is the genuine commercialisation in the field, not the appearance of private companies, which have always built launch vehicles under contract, but the shift of ownership and risk from the customer to the supplier.
Why constellations are hard in a different way
Cheaper launch made large constellations viable, and those have constraints unrelated to getting to orbit. Placing broadband satellites low rather than in geostationary orbit is driven by two factors. Round trip delay to geostationary altitude is a substantial fraction of a second purely from distance, which is intolerable for interactive traffic, and received power falls with the square of distance, so a low satellite closes a link with far less power and a smaller antenna.
The costs are severe. A low orbiting satellite is above a given point for minutes, so continuous coverage needs thousands rather than dozens. Residual atmosphere produces drag, so satellites expend propellant continuously to hold altitude and re-enter within a few years once they stop. The constellation is not a fixed asset but a consumable one, requiring permanent manufacturing and launch cadence simply to continue existing.
That is a fundamentally different business from a few large geostationary satellites operating for fifteen years, and it became possible only because launch fell far enough that continuous replacement is affordable.
The shared environment
Orbit is finite in two senses, and both now bind. Debris persists, and at orbital velocities even small fragments carry enough energy to destroy a satellite while generating more fragments. The concern is not a sudden event but that the fragment population reaches a level where it grows through collisions faster than drag removes it, making certain altitude bands progressively less usable.
Regulatory responses focus on prompt deorbit at end of life and on collision avoidance, and the operational burden rises with the number of objects.
The second constraint is regulatory. Spectrum and orbital positions are coordinated internationally and the process rewards filing early, so a large share of usable geostationary positions and desirable allocations are already claimed. Coordinating a new system against existing rights has little to do with engineering and is as significant a barrier as the technology.
What is commercial and what is not
Launch is a market, with multiple providers competing for commercial payloads at published prices. Satellite communications has been a market for decades, Earth observation imagery is one, and small satellite manufacturing has become genuinely competitive.
Human spaceflight is not, in any meaningful sense. Crewed missions are almost entirely purchased by governments, the vehicles were developed under public programmes even where privately owned, and private passenger flights are a small activity at a price reflecting cost rather than a functioning market. Deep space science is not commercial at all, because there is no customer other than a public agency, and proposals to extract resources from asteroids or the Moon remain at a stage where the cost of access exceeds any plausible value returned, which is arithmetic rather than ambition.
The honest summary is that a genuine market has formed around access to low orbit and services delivered from it, while everything beyond remains publicly funded exploration carried out by commercial contractors. Those two are frequently discussed as one and behave completely differently. If launch cost keeps falling the binding constraint simply moves, to manufacturing throughput, to ground segment and spectrum capacity, to orbital congestion, and for anything beyond low orbit to the ability to move mass once already in space. None of those are solved by a cheaper rocket.
Note: the energy needed to reach orbital velocity is roughly comparable to the energy content of the propellant that carries a similar mass across an ocean by aircraft. Space is expensive not because the energy is extraordinary, but because the vehicle carries its own oxidiser and, until recently, was discarded on arrival.