In the comforting, laminated world of Eurospace briefings and industry webinars, OneWeb is a success story. A pan-continental comeback narrative of how a bankrupt startup turned into Europe’s LEO connectivity darling, lifted from the ashes by state wallets, Starlink fear, and a PowerPoint called “IRIS².” It’s the kind of story that looks excellent from a policy desk in Brussels.
And it’s about to hit a wall.
Not because the technology is bad. Not because there aren’t satellites in the sky. But because every satellite dies, eventually, and OneWeb’s are coming up on that particular milestone faster than anyone wants to admit. Their design life is not some abstract metric, like “affordability” or “sovereign resilience.” It’s a number, written in ink: seven years. And the first batch of those birds hit that limit in 2026. No warning light. That’s the first knock on the door.
The fleet was launched in a burst of enthusiasm between 2019 and 2024. Twenty missions, 656 satellites, sprinkled across multiple launch vehicles after Russia turned off the Soyuz tap and forced OneWeb into a strategic tour of what was available: ISRO, SpaceX, anything that could fly and wouldn’t come with sanctions attached. It worked. The constellation went up. But it went up in layers, and those layers are now aging in order. Satellites don’t suddenly drop out of the sky when their design life ends. They just start getting worse. Batteries hold less. Radiation bites a little deeper. The orbital equivalent of knees giving out on stairs.
The schedule is merciless. That first cohort of six from 2019? They don’t matter. They’re already halfway to being historical footnotes. But the 2020 group is different. 104 satellites, all aging out in 2027. After that, the 2021 cohort, a massive 284 units, reaches the same milestone in 2028. That’s over 40 percent of the constellation. You don’t need a PhD in orbital mechanics to understand that letting half your network quietly expire over a single calendar year is not a plan. It’s a countdown.
| Launch Year | Satellites Launched | Hits 7-Year Mark In | Share of Constellation | Risk If Missed |
|---|---|---|---|---|
| 2019 | 6 | 2026 | 0.9% | Negligible – warm-up round. |
| 2020 | 104 | 2027 | 15.9% | Noticeable congestion. |
| 2021 | 284 | 2028 | 43.3% | Structural degradation. |
| 2022 | 110 | 2029 | 16.8% | Regionally painful. |
| 2023 | 132 | 2030 | 20.1% | Manageable if rest is fixed. |
| 2024 | 20 | 2031 | 3.0% | Irrelevant unless you’ve failed already. |
| Total | 656 | — | 100% | — |
This is where things get awkward. Because OneWeb does, in fact, have a plan. Sort of. Airbus is contracted to build the replacements. Toulouse gets the job, not Florida, which is what happens when the French government signs your equity check and then politely suggests your industrial base should be a little more European. The number being tossed around is 100 new satellites. Production starts in 2026. First deliveries around the end of that year. That’s a good headline, until you realize it’s a third of what they actually need. 100 satellites don’t replace 284. They don’t even replace the 2020 and 2021 cohorts combined.
And those 100 new satellites? They haven’t even left the ground. Not one has flown. Not one has even been boxed up and shrink-wrapped. They are, in the most generous reading possible, a set of specifications and a promise, still invisible on Airbus’ quarterlies. Meanwhile, the replacement clock is still ticking, and the satellite funerals are on the calendar whether anyone likes it or not.
The problem is launching them.
SpaceX was happy to help finish the Gen-1 constellation, but let’s not pretend that Elon’s team is going to stop the Starlink express and wait for OneWeb to catch up. ISRO was an emergency partner, and they did the job, but LVM3 isn’t exactly rolling off the pad every two weeks. Mitsubishi’s H3 is now theoretically in the mix, starting in 2027, but it hasn’t proven anything beyond not exploding recently. The pipeline exists, but it’s narrow, bumpy, and full of scheduling roulette.
To pull this off, OneWeb needs a constant beat of launches from 2026 to 2030. Three to six missions a year, every year, on time, with payloads that actually arrive, satellites that actually work, and orbits that don’t just fill gaps. No launch delays. No propellant shortfalls. No gateway issues. No customs snafus. No IRIS² integration bugs. In short, it needs to be the most boring and reliable manufacturing-and-logistics operation in modern satellite history. The kind of boring that Europe is famously good at imagining but rarely delivers without drama.
Speaking of IRIS²
This is where the political oxygen lives. OneWeb is being quietly wrapped into Europe’s next-gen sovereign connectivity constellation. The narrative is elegant: LEO + MEO + GEO. Secure. Assured. Multi-orbit. Multi-layered. On paper, it’s a neat trick. OneWeb replaces its aging fleet not just to survive, but to become the LEO layer of IRIS². Some of those 100 new satellites are described as “IRIS² compatible,” which means no one knows exactly what they need to be yet, but they’ll try to meet whatever the spec becomes.
That sounds nice. It also buys them time. But it doesn’t change the physics. IRIS² services aren’t expected before 2030. That’s two full years after OneWeb’s 2021 cohort hits retirement. If those satellites aren’t replaced in time, OneWeb becomes a constellation with holes, and IRIS² gets to start its sovereign rollout by patching someone else’s coverage map.
Then there’s the little A-PNT elephant. Assured Positioning, Navigation, and Timing. The favorite acronym in every defense sales pitch involving satellites and vaguely hostile environments. OneWeb’s AstraPNT offering claims to provide timing and positioning when GPS is jammed or spoofed. What it actually provides is timing. Position is another matter. It’s “assured” the way a backup clock is assured. It helps the system stay in sync. It helps the terminal think it knows what time it is. But it does not, by itself, allow a moving user to know where they are without GPS. For that, you need inertial sensors. Gyros. Accelerometers. Dead reckoning. External help.
The datasheets imply this without saying it. You can feed in NMEA data. You can override position manually. The terminal doesn’t care, as long as someone gives it coordinates. But don’t buy the brochure thinking OneWeb replaces GPS for troops on the move in a denied environment. It doesn’t. It’s a timing aid. A good one. But not magic. The moment you need navigation in a GNSS blackout, you’re back to trusting the same black box you yanked from a decommissioned submarine and hoping your ITAR paperwork was filed correctly.
And what about the money?
Building satellites isn’t cheap, but OneWeb at least benefits from industrial scale. The original production model aimed for around one million dollars per unit. Launches cost somewhere between sixty and seventy million depending on provider and luck. That means the cost to build and deploy 100 new satellites probably sits in the $300 to $450 million range.
That’s before you factor in IRIS² bells and whistles or operational headaches. France helped. In 2025, they subscribed to more than half of Eutelsat’s €1.35 billion capital raise, becoming the largest shareholder in the process. That was triage. Strategic triage, dressed up as industrial policy.
It worked. But now the leash is short. There is no more room for poetic failure. OneWeb has exactly one job: replace what’s aging on time, integrate into IRIS², and keep the network from blinking under pressure. It doesn’t need to outperform Starlink. (It won’t). It doesn’t need to match Kuiper’s promises. (It won’t). But it does need to keep every minister, every system integrator, every middle-tier telco CTO from calling up someone else and asking whether they can get guaranteed service delivery next year.
If OneWeb nails the handover, it stays quietly essential. Not flashy. Not headline-grabbing. Just necessary. A sovereign-aligned constellation with stable throughput, stable contracts, and an upgrade path that leads cleanly into IRIS². But if they let 2027 or 2028 slip, the illusion of continuity breaks. The service degrades. The gaps widen. And the story shifts from “European LEO resilience” to “a polite reminder of why replacement cycles aren’t optional.”
OneWeb’s biggest threat isn’t Starlink. It’s entropy.
And entropy never delays its launch.
SES: Sovereignty With Stock Tickers