Analysis · Satellite Insights

The Multi-Orbit Mirage: How GEO Operators Are Retiring the Fleet They Can't Fill

SES and Eutelsat are cancelling geostationary satellites and writing down GEO assets while multi-orbit partnerships grab the headlines. An operator's read on why a LEO deal keeps the customer but never fills the transponder — and why the GEO arc may be about to empty out.

By Glenn Canales · Principal, Satellite Insights

The press releases describe reinvention. The balance sheets describe a managed retreat.

When SES used its May earnings call to confirm it had cancelled two software-defined satellites — IS-41 and IS-44, both inherited from Intelsat, both slated for 2028 — the company called it fleet rationalization. Read alongside Eutelsat scrapping its Flexsat Americas program months earlier, it's something blunter. The two largest geostationary operators in the Western world are choosing not to build the GEO capacity they're entitled to build, and quietly writing down the capacity they already have. Eutelsat took a €535 million goodwill impairment against its GEO assets in its H1 FY2024-25 (December 2024) accounts. That is not a number you book while you're reinventing a business. It's a number you book while you're shrinking one.

So how are GEO operators actually responding to the low-earth-orbit assault? Not, for the most part, by making their stranded geostationary capacity productive again. They're doing three things instead: building fewer new GEO satellites, extending the life of the ones already aloft to defer capital, and narrowing what's left toward the handful of missions where geostationary physics still wins. The multi-orbit partnerships that dominate the headlines — Viasat reselling Telesat's Lightspeed, Eutelsat fusing OneWeb with its GEO fleet — solve a different problem entirely. They keep the customer. They do not fill the transponder.

Why doesn't a LEO partnership fix idle GEO capacity?

Because a partnership is a customer-retention strategy wearing a capacity-strategy costume.

Consider the Viasat–Telesat deal, the cleanest example because the two layers sit in two different companies. Viasat signed a multi-year contract to integrate Lightspeed's Ka-band LEO capacity into its multi-orbit network, with Lightspeed targeted to enter commercial service around the end of the first quarter of 2028 — itself pushed back from late 2027 on chipset delays. Mark Dankberg framed it as a way to achieve industry-leading resource utilization and drive capital efficiency. But look at what the transaction actually is on Viasat's books: it is procurement. Viasat is buying someone else's low-latency capacity to keep an aviation or government customer who might otherwise defect to Starlink. That adds a cost line. It does nothing to light up a single underused Viasat GEO transponder. If anything, the company is now carrying the idle GEO asset and paying Telesat for the LEO layer on top of it.

The owned-fleet operators — Eutelsat with OneWeb, SES with O3b mPOWER in medium orbit — capture the LEO revenue in-house rather than paying it out, which is a better position. But the mechanism is the same. When Eutelsat sells a customer a fused GEO-LEO service, the growth shows up in the LEO line. In its half-year to December 2025, its LEO revenue rose nearly 60% year over year while GEO connectivity revenue fell 4.5%. The partnership, or the in-house equivalent, is the vehicle that carries the customer across from the declining asset to the growing one. The declining asset doesn't get more productive. It gets left behind more gracefully.

What are GEO operators actually doing with the fleet?

Three things, none of which is reinvention.

They're building less. The real news this spring wasn't a partnership announcement — it was two cancellations. SES pulled IS-41 and IS-44 because they failed its internal rate-of-return threshold. Eutelsat scrapped Flexsat Americas and redirected more than €100 million toward LEO, leaving it with no wholly owned GEO satellites on order beyond a single Thaicom joint-venture craft. The detail that should end the "flexible payloads will save GEO" argument: both cancelled SES birds were software-defined, and Flexsat was the reconfigurable-in-orbit platform that flexibility advocates point to. When even the flexible GEO satellite fails the capital test, the problem isn't the architecture. It's the demand.

They're sweating what's already up there. SES has lined up five life-extension missions between 2026 and 2029, using servicers from Northrop Grumman's SpaceLogistics, Starfish Space and Infinite Orbits. Keeping an existing satellite aloft for five more years is cheaper than replacing it, and it stretches the depreciation runway so the operator can price the marginal capacity lower against Starlink. That's real, and it's smart. But it is margin defense, not utilization. You don't extend a satellite's life because demand is straining your fleet. You extend it because it isn't.

They're narrowing to what GEO still does well. This is the only mechanism that genuinely turns idle inventory productive again, and it works case by case rather than fleet-wide. Eutelsat took its Konnect VHTS satellite — 500 Gbps, originally built to sell consumer broadband across Europe, North Africa and the Middle East — and repointed it at mobility markets after its original demand never materialized and Italy's TIM declined to migrate onto it. That's a genuine second life. But notice it's a scramble: one satellite whose first market evaporated, redeployed to a second. It's not a repeatable model for a fleet, and it depends on there being an adjacent market with unmet demand — which is exactly what's getting scarcer.

Where does GEO capacity still win?

In the missions where low-earth orbit is structurally bad: one-to-many distribution, and assured connectivity in contested environments.

Broadcast is the obvious one, and it's declining — but slowly, and it still generates cash. SES's media business remains roughly a third of revenue and secured close to €100 million in renewals and new business in a single quarter even against the structural headwind. One geostationary beam reaching millions of fixed receivers is a job a LEO constellation does badly and expensively. The economics of one-to-many haven't inverted the way the economics of point-to-point broadband have.

The live growth pocket is government. Eutelsat now markets itself as the operator that can fuse GEO, LEO and terrestrial layers into a single resilient architecture — primary, alternative, contingency and emergency pathways for users who need connectivity to survive jamming and interference. In the quarter to March 2026, its government revenue rose nearly 12%, lifted by Ukraine and by non-US government demand. This is where geostationary capacity commands a premium that commoditization can't reach, because the buyer isn't paying for bandwidth — they're paying for assurance, sovereignty, and the guarantee that a known beam will be over a known place at a known time. Starlink's consumer economics — which Novaspace pegs near $0.20 per gigabyte — are irrelevant to a defense ministry that needs a hardened, sovereign link it controls.

So the productive remainder of the GEO fleet concentrates around these two poles: the declining-but-durable broadcast core, and the growing, premium, sovereignty-driven government layer. Everything in the middle — the generic wide-beam data capacity that was the industry's bread and butter for thirty years — is the part that's stranded, and it's the part that's quietly being cancelled, written down, or extended into a cheaper twilight.

Is GEO running out of room, or running out of demand?

Here's the inversion almost nobody is pricing in. For three decades, the working assumption of the entire industry was that geostationary slots are scarce — the Manhattan real estate of space, perpetually contested, reliably appreciating. The data now points the other way.

By one widely-tracked count, the number of active commercial GEO satellites grew from 541 in February 2022 to just 573 in January 2026 — a compound growth rate of about 1.5%, effectively flat. Replacement orders are slowing across the sector, a trajectory London Economics now reads as an emerging surplus of orbital slots. And the demand-side pressure is about to intensify: with Amazon Leo — the constellation formerly known as Project Kuiper — targeting commercial service in mid-2026, Novaspace expects a second mega-constellation competing for the same users to trigger another downward lurch in capacity pricing. As operators rationalize fleets and defer deployments, orbital positions that were once fought over may increasingly fall vacant. For the first time in decades, the credible question isn't how operators will squeeze into a crowded arc. It's what happens when the arc starts to empty out.

That reframes the geostationary slot itself. It stops being a permanent address for long-lived infrastructure and starts looking like a time-limited lease for flexible capacity — held while it's worth holding, surrendered when it isn't. And it opens an angle that ties directly to the sovereignty premium above: with the arc less congested, satellite manufacturers could begin pairing software-defined platforms with managed regulatory filings, offering turnkey packages to governments that hold dormant orbital filings they've never had the means to deploy before the regulatory deadline expires. The scarce asset in that world isn't the slot. It's the ability to put something credible into it on time.

The strategic read is uncomfortable for anyone still modeling GEO as a scarcity business. Idle geostationary inventory doesn't get rescued. It gets retired, written down, or narrowed to the few missions where the physics still pays. The multi-orbit partnership is simply how the operator keeps the customer while it does the narrowing — a graceful exit dressed as a growth strategy. Watch the order manifests, not the partnership announcements. The manifests are where the truth is, and right now they're getting shorter.