Novaspace's count, published on 21/09/2026, puts commercial GEO satellite orders at 14 for 2026 so far, against 10 across the whole of 2025 and the highest annual total since the 18 booked in 2020. Three of the 14 exist because of the upper C-band auction scheduled for 2027, and SES alone is planning seven GEO spacecraft to carry its broadcast and data customers through the spectrum transition. The prime contractors building them describe the surge as targeted fleet replacement and regulatory compliance rather than a return of structural growth, and expect procurement to settle at moderate rates through 2027.
As a market signal, that reads as a caution. As a credit it reads as the most financeable order book the GEO segment has produced in six years. A replacement satellite goes to an operator that already holds the orbital slot, already serves the customers and already books the revenue against them. The only genuinely new variable in the transaction is the spacecraft. Readers coming to these asset classes for the first time will find the layers set out in Space Assets 101.
What replacement removes from the underwriting
Financing a greenfield GEO programme means taking a view on three questions at once: whether the capacity finds customers, whether the slot and the licence hold, and whether the operator can execute against either. Replacement answers all three before the contract is signed. The capacity is already sold, the slot is already occupied, and the operator has flown the predecessor for the better part of two decades.
| Greenfield GEO programme | Replacement GEO satellite | |
|---|---|---|
| Demand risk | Capacity yet to be sold | Contracted on the incumbent spacecraft |
| Orbital slot and licence | To be secured and brought into use | Held and in use |
| Operator record on the asset | None specific to it | Full service history on the predecessor |
| Revenue at service entry | Builds from zero | Transfers from the retiring satellite |
| What is genuinely new | Market, licence and hardware | Hardware |
Read the bottom row. Everything a lessor would normally have to price on a GEO transaction has been priced already by the operator's existing business, and what remains is an equipment risk on a known bus from a known manufacturer. That is the narrowest underwriting question this asset class offers.
Aviation drew the same line a long time ago. An airline replacing a 737-800 on a route it has flown for fifteen years borrows on different terms from a start-up carrier ordering its first aircraft, although the airframe is identical and the manufacturer is the same. The difference sits in the revenue the aircraft flies into. Satellite replacement is that trade, and it has not yet been priced that way, because the sale-leaseback market that would price it is only now forming.
The replacement curve behind the orders
Orders are the visible half. The other half is how many satellites are reaching the end of their design life. We counted the 573 active GEO satellites in public tracking data at the start of 2026 and took out those outside a commercial lessor's reach: government and military systems, and Chinese and Indian state fleets. That leaves 321. Assuming replacement 15 years after launch, the 126 already past that age clearing evenly over five years, and seven in ten replaced one for one:
| In-orbit delivery year | Satellites reaching 15 years | Backlog clearing | Replacement demand |
|---|---|---|---|
| 2027 | 19 | 25 | 31 |
| 2028 | 14 | 25 | 27 |
| 2029 | 16 | 25 | 29 |
| 2030 | 18 | 25 | 30 |
| 2031 | 20 | 25 | 32 |
| 2032 | 24 | — | 17 |
| 2033 | 15 | — | 11 |
| 2034 | 11 | — | 8 |
| 2035 | 8 | — | 6 |
| 2036 | 8 | — | 6 |
That is about 30 replacements a year to 2031, against 14 orders so far in 2026 and the 15 to 18 a year Airbus expects. Two readings follow. Orders are running below the curve, so some operators are extending old satellites rather than replacing them, which is where life extension comes in. And from 2032 the curve thins sharply, so the replacement wave that financiers can underwrite is concentrated in the next five years.
Caelum analysis of public tracking data as at 01/2026. Indicative, not a forecast. Demand is counted in the year of in-orbit delivery, about three years after order. Rows rounded.
Why spectrum-driven capex belongs off the operator's balance sheet
Three of this year's orders exist because of a regulatory deadline. That separates them from the rest of the book in a way that matters to a treasurer. Discretionary growth capex can be deferred a year when the cost of capital moves against it. Compliance capex cannot, because the satellite has to be built, launched and in service by a date the regulator sets, whatever the financing market happens to be doing when the construction contract is signed.
It is also the capex that earns nothing incremental. It defends revenue the operator already has, against spectrum it is being required to vacate, and the value of what is being defended is the subject of the most valuable real estate you can't see. SES's award to Intuitive Machines, reported on 19/08/2026 at an anticipated value of more than $600m and covering a reported five operational spacecraft plus two spares, is defensive spending of exactly that kind. Intuitive Machines has said the remaining programme details stay confidential at the customer's request.
A sale-leaseback at in-orbit delivery answers the problem without touching the network. The operator funds construction, sells the spacecraft to a S.110 DAC on hand-over and leases it back on a term matched to useful life, a calculation that in-orbit servicing has begun to change and which is worked through in satellite life extension and the lease term. The capital returns to the balance sheet at the moment the asset enters service, the capacity never leaves the network, and the debt headroom the operator has preserved goes to the orbits where it is actually taking risk. Every euro of compliance capex carried on the corporate balance sheet is a euro unavailable to the LEO or MEO programme competing for the same credit.
The conversation has to happen at order, not at delivery
None of the 14 is in orbit. Every one of them is a construction contract today, and the three tied to the auction have to become working spacecraft before the deadline bites. The lease itself cannot start until in-orbit delivery, when title can pass and the asset exists to be owned. The commercial decision about who takes that title has to be made now, while the construction contracts are still being negotiated.
Aviation handles this at order. A lessor commits to the sale-leaseback when the airline signs the purchase agreement, takes an assignment of the delivery position, and funds the pre-delivery payments in the interim. The airline gets certainty on its capital years before it needs it, and the lessor gets the same years to underwrite an asset it will eventually own. Nothing in that sequence depends on the asset being an aircraft. It depends on a lessor being at the factory gate when the contract is signed, which is an arrangement the satellite market has not yet made.
An operator ordering a replacement spacecraft today is therefore making two decisions and only recognising one of them. The procurement decision is visible and gets a board paper. The ownership decision is being taken by default, in favour of the balance sheet, because nobody has put the alternative in front of it in time. Our view of how that capital should be structured is set out for institutional investors.
Order counts and the 2027 outlook per Novaspace as reported on 21/09/2026. The SES award as reported by Advanced Television on 19/08/2026 and described by Intuitive Machines as anticipated; satellite count and contract value are unconfirmed. Nothing here is investment advice or a forecast of returns.