In the first half of 2026, SpaceX quietly vaporized 260 of its own satellites. No malfunction, no mission failure — just planned decommissioning of first-generation hardware that had reached the end of its operational life. The spacecraft were commanded to fire their Hall-effect thrusters, lower their perigee, and let the atmosphere finish the job.

It was, by SpaceX’s own characterization, routine maintenance.

And yet the semi-annual FCC compliance filing that disclosed this fact is arguably the most revealing document in satellite broadband this year. Not because burning satellites is alarming — it is not — but because it makes visible a cost that business plans and analyst models have quietly discounted: a mega-constellation doesn’t just need to be built. It needs to be continuously rebuilt.

A Clock Built Into Every Satellite

Starlink’s earliest v1.0 spacecraft launched starting in May 2019. The v1.5 generation followed in 2021. Both cohorts were designed for a roughly five-year operational life — a deliberate tradeoff. Longer-lived satellites require more propellant for altitude maintenance, more radiation hardening, and heavier components. They also risk becoming technologically obsolete in a network that is upgrading rapidly. Short design lives keep the hardware lean and ensure that the constellation improves with each replenishment cycle.

The empirical record largely confirms the design intention. An analysis of Starlink orbital dynamics published earlier this year found operational lifespans clustering in the four-to-six-year range, with a measurable daily failure probability. By late 2025, the first cohorts were approaching — or had already passed — that threshold.

Between December 1, 2025 and May 31, 2026, SpaceX deorbited 260 satellites via controlled atmospheric reentry. According to the FCC filing, the decommissioned hardware consisted primarily of v1.0 and v1.5 units exhibiting battery degradation or telemetry anomalies consistent with age. SpaceX reported exceeding the FCC’s 95% disposal reliability threshold, with over 99% of commanded satellites completing deorbit maneuvers as planned. An additional 349 retired spacecraft were in the disposal queue at the time of filing — bringing the total being turned over to roughly 609 units, or about 5.7% of the active fleet of approximately 10,413 satellites.

Solar Cycle 25 Threw the Math Off

The generational clock was always part of the plan. What wasn’t part of the plan — for SpaceX or anyone else — was Solar Cycle 25.

The cycle, which peaked around 2024–2025, ran substantially stronger than the consensus forecast issued by the NOAA/NASA/ISES prediction panel in 2019. Atmospheric densities in low Earth orbit from 2022 through 2026 held at roughly two to three times the predicted levels, dramatically increasing drag on all low-orbiting objects. A research paper published in June 2026 quantified the damage across the broader LEO satellite population: against the nominal solar forecast, operators collectively lost an estimated 2,472 cumulative mission years of satellite service, representing approximately $2.77 billion in lost mission value. Even measured against the conservative two-sigma engineering design target, the shortfall reached 688 cumulative mission years and roughly $880 million.

For satellites that rely on propulsion to hold their orbital altitude — which is most operational Starlink units — the elevated drag environment burned through propellant budgets faster than planned, effectively shortening individual satellite lifespans. Some satellites that might have run five or six years on-nominal solar conditions were done in four.

The broader implication is subtle but important: operational planning built around historical solar cycles systematically underestimates replacement frequency near solar maximum. For operators now designing constellations meant to operate into the next solar cycle (forecast peak around 2035), this is a live engineering and financial variable, not an edge case.

The Real Cost of Running the Treadmill

Turning over a fleet of ten thousand satellites every five years is not cheap, even when it is expected. Independent financial analysis of Starlink’s cost structure estimates an annual fleet replacement burden in the range of $2.4 to $4.6 billion for a constellation at that scale, depending on unit costs and launch pricing assumptions.

Starlink generated $11.4 billion in connectivity revenue in 2025 and produced $4.4 billion in operating profit — making it SpaceX’s only profitable business segment and the financial engine that funds Starship development. That operating profit exists net of replenishment costs. The replenishment treadmill is already priced in, and the constellation is running it sustainably.

This is where SpaceX’s vertical integration becomes structurally decisive. Starlink satellites are manufactured internally, and they ride to orbit almost exclusively on SpaceX’s own Falcon 9. The marginal cost of a replenishment launch is effectively the internal transfer price. No third-party launch provider’s margin is extracted. No spot-market volatility applies. For a business measured in thousands of replacement satellites per year, that matters enormously.

Any competitor without that stack — integrated manufacturing, captive launch — faces a structurally higher replenishment burden that eats directly into the economics of a connectivity business with thin margins to begin with.

What This Means for Amazon

Amazon’s Leo constellation, still in early deployment, had approximately 331 satellites in orbit when the FCC waived its July 2026 milestone deadline in June. The milestone had required half the licensed constellation — 1,616 satellites — to be operational by that date. The full-constellation deadline remains July 30, 2029.

The replenishment arithmetic compounds the challenge. Amazon’s constellation will be operating under a five-year replacement clock from the moment of launch. By the time it has completed initial deployment, the earliest satellites may have only a few years of remaining life. Unlike SpaceX — which is already running a mature replenishment operation funded by a profitable network — Amazon will be simultaneously completing initial build-out and beginning to plan for the first wave of end-of-life disposal. The business must be in continuous replacement mode before it has finished its initial construction mode.

That is a harder position than it appears.

The Environmental Footnote Worth Watching

A quieter concern has been gathering around the vaporization of aluminum-chassis satellites at scale. When several hundred spacecraft per year burn up in the upper atmosphere, the combustion produces aluminum oxide particulates — alumina — that accumulate in the stratosphere. The potential effects on stratospheric albedo and ozone chemistry are not fully characterized, but atmospheric scientists have flagged the cumulative burden as a genuine unknown. Pressure on the FCC to require proper environmental impact assessments — rather than relying on categorical NEPA exclusions — is growing, and the issue is unlikely to stay technical for long.

The Regulatory Design Was Intentional

None of this is accidental. The FCC adopted its five-year deorbit rule in September 2022, replacing the old 25-year guideline with an explicit compliance obligation. The Commission’s stated rationale was direct: the old rule had no teeth, defunct satellites lingered for decades, and the debris environment was accumulating risk faster than voluntary guidelines could address. Requiring disposal within five years of end-of-mission forces operators to treat replenishment as an ongoing operational commitment rather than a future problem.

The rule has worked as intended. SpaceX’s FCC filing exists precisely because the five-year mandate requires reporting. The debris that isn’t accumulating — the 260 spacecraft that came down rather than drifting into crossing orbits — is a direct product of that enforcement mechanism.

A Moat Made of Launches

The replenishment treadmill has been running for a while. What is new is that it is now visible in regulatory filings, auditable at scale, and carrying real financial weight. The constellation that was still being built three years ago is now also being continuously decommissioned.

For SpaceX, this is infrastructure management — tedious, expensive, and thoroughly absorbed into a profitable operating model. For every other would-be competitor, it is a preview of a structural cost that has no elegant solution except building a Starlink.

Sources