As many as half of all geostationary communications satellites reach the end of their 10 to 15 year design life with their subsystems still working. They retire for one reason: they run out of the propellant that holds them on station. That finding, from ESA's Space Debris Office, is an asset-register statement as much as an engineering one. Half the GEO fleet is written off while the hardware still functions, stranded by an empty tank rather than by failure.
Every lease on a satellite is written against a date. Life extension moves that date, and a lessor has to decide what to do with the movement.
What has actually flown
The flight history is short but real. In February 2020 Northrop Grumman's Mission Extension Vehicle, MEV-1, docked with Intelsat 901, the first time one commercial satellite took over station-keeping for another in orbit. It delivered five years of added service and undocked in April 2025, mission complete. MEV-2 docked with Intelsat 10-02 in April 2021 on the same basis. Between them the two vehicles have docked with three commercial GEO satellites and delivered more than ten years of combined extension. Both are covered in our earlier piece on in-orbit servicing.
The MEV architecture was one-to-one. The servicer docked with the client and stayed for the duration, so one servicing asset was consumed per client. That is a proof of concept and an expensive way to run a fleet.
On 22 July 2026 the architecture changed. Northrop Grumman's SpaceLogistics launched its Mission Robotic Vehicle together with three Mission Extension Pods. Over the coming year the MRV will move to geosynchronous orbit, fit a pod to a client satellite, then detach and repeat the operation on two more. The costly part, robotics and rendezvous and proximity operations, stays on orbit and is reused. What remains with the client is a pod: smaller, cheaper, and installable across a range of buses. The MRV also carries a Passive Refueling Module, a docking and refuelling interface standard, so the servicer's own life is not fixed either.
Three pods is not a fleet and the servicing operations have not yet been performed. But the architecture now flying is the one that would make life extension a repeatable service rather than a bespoke rescue.
Why this is a lease-term question
An aircraft lease works because the asset can be maintained, overhauled and redelivered in a defined condition. The lessor prices maintenance reserves, return conditions and the cost of the next heavy check into the rent, and the residual value at lease end is a function of that managed condition rather than of the airframe's original delivery date. Satellites have historically had none of this: build, launch, depreciate on a straight line to the propellant horizon, write off.
Life extension changes the inputs in three places.
First, design life and economic life separate. The write-off date shifts from the propellant budget to a commercial decision on whether extension clears its own cost. A lease can be written to the design life with an extension option, priced in advance, that the lessee exercises if the transponder market at year twelve justifies it. That is a real option with a quotable strike.
Second, life extension becomes a reservable cost. A one-off rescue negotiated with a single servicer cannot go in a residual model, because nobody can say what it will cost. A pod installed by a shared servicer on a published interface standard can be quoted, and a cost that can be quoted can be reserved against, in the same way an aircraft lessor reserves for an engine shop visit. Fleet economics appear: one servicer amortised across many clients prices very differently from one servicer per client, the difference between a shared spare-engine pool and buying a spare for every airframe.
Third, the asset can be remarketed. A serviceable satellite with propellant reserve is a satellite a second lessee can take at year fifteen. Aircraft hold value because they can be maintained, overhauled and re-marketed; orbital assets are beginning to acquire the same property, and residual value assumptions can begin, cautiously, to reflect it.
The other two levers
Propulsion is the constraint that ends most GEO missions, but it is not the only thing that changes across a satellite's life. Software-defined payloads, such as Thales Alenia Space's Space Inspire and Airbus's OneSat, can be re-tasked in orbit: frequency plan, beam shape, coverage area. The same chassis can serve a different anchor customer mid-life, which turns a single-tenant building into a multi-tenant one and widens the pool of lessees the asset can be remarketed to. We covered the economics in what flexibility really buys you.
Refuelling is the lever still in development. Astroscale's APS-R, contracted with the UK Space Agency in 2024, and propellant-depot architectures are moving servicing from demonstrator toward scheduled service. Refuelling resets the fuel budget directly rather than nursing what remains, and if it becomes routine the propellant horizon stops being a horizon at all.
What Caelum does with this today
A satellite lease written today should still be sized to design life, with residual treated as upside rather than as a component of the rent. The change is in the documentation. A sale-leaseback on an in-service satellite should ask for propellant-margin evidence the way an aircraft lease asks for maintenance records; it addresses who may contract for servicing and on whose account; and it treats an extension option as something to price rather than something to leave to a later negotiation. The refinancing structure for in-service satellites is where these terms matter most, because the asset already has a history and a remaining life that can be measured. Managed life is the property that made aircraft financeable. Orbit is building the same machinery, and the leases should be written for the fleet that machinery produces.