Starlink was sold to Europe as infrastructure. This week it behaved like a lever.
According to heise, throttled Starlink capacity has pushed the question of European dependence on a single commercial satellite network back to the top of the political agenda. The detail that matters is not the bandwidth figure. It is that the throttle was a decision, not an outage — something a company can do, and undo, without asking anyone first.
That distinction is the entire argument.
Why one operator has this much leverage
A low-Earth-orbit constellation is not a neutral pipe. Someone owns the satellites, the ground stations, the terminal firmware, the peering arrangements and the contract you signed. Even in countries where the service is fully licensed and the spectrum fully coordinated, the operator decides where scarce capacity goes when demand outruns supply. Capacity on a satellite is allocated per beam, and beams are steered. A regulator can license a service; it cannot conjure a second constellation in the same afternoon.
That is why a commercial decision can arrive in a defence ministry looking like a strategic one. Military and civil communications planners have spent a decade quietly moving onto commercial capacity because it was cheaper and faster than building their own. Starlink's terminal is a flat panel you can hand to a battalion; a sovereign constellation is a decade and a budget line.
Europe already knew this
The uncomfortable precedent is Ukraine, where Starlink terminals arrived within days of the 2022 invasion and became the backbone of battlefield communications. It was also in Ukraine that the limits became visible: reporting on decisions taken around coverage in occupied and contested areas made clear that a private operator held a veto over operations it never signed up to support. Kyiv's answer was to diversify, and the Pentagon's was to buy its own managed service rather than rely on a consumer product.
Europe's answer has a name and a schedule. IRIS², the EU's multi-orbit constellation, was contracted at the end of 2024 with a design of roughly 300 satellites and a promise of initial services around the end of the decade. Govsatcom pools existing governmental capacity in the meantime. Eutelsat's OneWeb fleet offers a partial LEO alternative today, with a fraction of the capacity and a different set of owners — European, but still commercial, still able to be throttled.
Those programmes are real. They are also, in the terms that matter this week, not finished. Sovereign capability that begins operating in 2030 does not help an emergency services network deciding in October what its fallback is.
The lesson is contractual, not technical
There is a version of this debate that gets lost in Muskology — treated as a personality story about one mercurial billionaire. That framing is a distraction, and slightly convenient. The structural problem exists regardless of who runs the company: Europe has outsourced a strategic layer to a vendor with no obligation to keep it available under conditions the vendor did not choose.
The fix is unglamorous. Procurement that specifies service levels under stress, not just average throughput. Reserved capacity written into contracts and paid for, the way navies pay for sealift. Multi-orbit terminals so that switching providers is an operations decision rather than a hardware refresh. And a hard-nosed accounting of what an independent constellation actually costs per year, compared with the invoice from a supplier who can leave.
None of that is a headline. All of it is what a country does after it discovers that being a customer is not the same as being sovereign.
What to watch
Whether European procurement starts writing explicit non-throttling and priority-access clauses into satellite contracts, rather than accepting commercial terms of service. Whether IRIS² milestones hold through 2027. And whether national governments — Germany and France in particular — fund their own fallback capacity in parallel, which would be an admission that the EU timeline is too slow for the threat they perceive.
Starlink may well keep serving Europe happily for years. The point of this week is that Europe now has to think about the version of events where it does not.
