For the first six decades of spaceflight, the standard practice was to build an enormously complex machine, use it once, and let it fall into the ocean. The reasoning was not stupidity. It was that recovering and reusing a booster costs performance, weight, and development money, and nobody had demonstrated that the savings would outweigh the costs.
That assumption has now been tested, and the answer changed the industry.
Why reuse is hard
A rocket first stage is a thin-walled tank containing enormous quantities of propellant, structurally optimised to the edge of what will survive flight. After separation it is travelling at thousands of kilometres per hour, tens of kilometres up, with essentially no aerodynamic control and no reason to stay intact.
Bringing it back means solving several problems at once:
- Reversing its motion, which requires holding back propellant that could otherwise have gone toward payload.
- Surviving re-entry heating on a structure never designed as a re-entry vehicle.
- Controlling attitude in thin air where conventional fins do little.
- Slowing from supersonic speed to a standstill and landing on a target measured in metres.
- Doing all of this while leaving the engines in a condition that permits reflight without a rebuild.
The Space Shuttle is the cautionary example. It was reusable in principle, and it did fly its orbiters repeatedly. But the refurbishment between flights was so labour-intensive, and the system so complex, that the promised cost savings never materialised. Reusability alone is not the goal. Rapid, cheap reusability is.
What made it work
Propulsive landing turned out to be the tractable approach, and several developments converged to make it practical.
Engines had to become deeply throttleable and restartable in flight, which is a substantial demand on turbomachinery and ignition systems. Grid fins provided aerodynamic control authority during descent. Cold-gas thrusters handled attitude control outside the atmosphere. And crucially, guidance software became good enough to compute a landing trajectory in real time, correcting continuously as conditions changed.
The last point is easy to underrate. The precision required to land a booster on a small barge in the ocean is not primarily a hardware achievement. It is a control problem that was not solvable with the computing available in the 1970s.
The physics of landing a booster was never in doubt. The economics and the control software were.
What it changed
The visible effect is on price per kilogram to orbit, which has fallen substantially. The less visible effects may matter more.
Launch cadence rose dramatically. When a booster can be recovered and flown again within weeks, the bottleneck shifts from manufacturing to processing, and the number of flights per year climbs into territory that was previously implausible.
That cadence enabled business models that did not previously make sense. Large satellite constellations — hundreds or thousands of spacecraft — require a launch rate that expendable rockets could not economically supply. Rideshare missions became routine, giving small operators and universities affordable access to orbit.
It also changed how hardware is developed. When flights are cheap and frequent, testing in flight becomes a reasonable engineering strategy rather than an unacceptable risk. Iterating on real vehicles is faster than iterating on simulations.
What remains unsolved
Reuse today is mostly partial. First stages are recovered; upper stages generally are not, because an upper stage reaches orbital velocity and must survive a full re-entry to come back. That is a much harder thermal problem, and solving it is the current frontier.
Fairings — the protective nose cone — are also recovered in some systems, though they are a smaller share of vehicle cost.
Full and rapid reuse of an entire launch vehicle, with turnaround measured in hours rather than weeks, remains a goal rather than a demonstrated capability. Whether it is achievable at scale is the open question that will determine how much further launch costs can fall.
The wider shift
Reusability arrived alongside a broader change in how spaceflight is procured. Agencies increasingly buy launch services rather than build launch vehicles, competing suppliers against each other rather than funding a single national system on a cost-plus basis.
The two developments reinforce each other. Fixed-price competition creates pressure to reduce cost; reusability is the most effective available lever. Landing boosters became normal remarkably quickly — the first successful landings were genuinely startling, and within a few years they had become a routine footnote at the end of a launch broadcast.