New rules for new Starlink customers took effect on 14 July 2026. Existing customers move to them on 17 August¹. The update puts more than thirty European states into a single travel region so that account holders can cross borders inside it without extra registration or the standard thirty-day limit on international use. The region includes Germany, Czechia, Slovakia, Lithuania, France, Spain, Italy, Romania, and Norway, among others². Poland is not on the list. Starlink’s own documentation says clearly that Poland is not included in the European region and that accounts registered in Poland are treated as intended for use in Poland only². Polish users on the “Roam Unlimited” plan can still use the service abroad, but only for thirty days per trip. After that they must return to Poland to reset the allowance, move to a more expensive package, or register the service in another country^{1, 3}. Polish media report that users in thirty-five other European countries face none of these requirements¹. SpaceX has not explained the exclusion.

The reaction from Warsaw was immediate, and it came from three officials, including the Polish Foreign Minister. Radosław Sikorski wrote to Musk directly on X: “Hey, @elonmusk, big man, stop discriminating Polish users of Starlink or we might re-think paying you $50 million p.a. for your services.”^{2, 3} Deputy Prime Minister and Minister of Digital Affairs Krzysztof Gawkowski said he does not accept “that this is simply how it must be.” He asked SpaceX to restore equal treatment for Polish customers urgently and added that if the reason is a local regulatory requirement, he expects the specific provisions to be named because general explanations are not enough². Gawkowski also made a broader point that is worth noting for how it frames the problem: new technologies have given global companies great power over the conditions under which people work, communicate, and travel, and for this reason it is the duty of the state to respond to unequal treatment of customers and to unfair market practices². Deputy Digital Affairs Minister Michał Gramatyka, from the junior coalition party Polska 2050, supported Gawkowski’s position and criticized the restriction of a commercial service without clear justification³. The Polish technology press has been blunter. The outlet Telepolis called the new terms “mockery” and “open discrimination against Poles”¹.

What the $50 million pays for

The sum Sikorski mentioned has nothing to do with these domestic subscribers. It refers to Poland’s continued funding of the more than 29,000 Starlink terminals it has supplied to Ukraine. Poland has paid for their subscriptions since 2022, because satellite connectivity is essential wherever terrestrial networks have been destroyed or jammed, including for drone and missile guidance⁴. Ukraine’s terminals operate under separate arrangements outside the EU roaming region, so the tariff change affecting Polish accounts does not touch Ukraine’s service at all. This weakens the theories that connect the exclusion to tensions over the Ukraine relationship⁴.

A private classification, not a legal one

What happened here is not an interstate issue. It is an internal commercial decision, taken without any published process, which produces a different regulatory effect for the nationals of one EU member state than for their neighbours. No treaty, regulation, or administrative act places Poland in a lower category of service. The classification exists only inside SpaceX’s own terms, and it can be changed as unilaterally as it was introduced. Gawkowski’s own words describe the legal problem well, even if he does not use legal language: a state can regulate conduct on its territory, but without a specific legal basis it cannot compel a foreign infrastructure operator to explain or justify an internal classification of this kind, even when the practical effect looks very much like discriminatory market access.
The scale here is what makes this case unique: Starlink’s global network serves around 12 million subscribers⁴. It is also, as the Polish commentary notes, dual-use technology: the same infrastructure serves civilian customers and, as the Ukraine arrangement shows, wartime military communications⁴. Against that scale, a change in one member state’s roaming terms is a small governance question. It is still a useful one, precisely because it is small and procedurally invisible. There is no public record of a decision being taken, only its effect.

The capacity gap behind the rhetoric

The episode comes as the European Commission pursues its Technological Sovereignty Package, adopted on 3 June 2026, which targets EU dependency on non-European cloud, chip, and software providers⁵. Satellite broadband is outside that package, and the capacity gap is large. The EU’s own answer, IRIS² (Infrastructure for Resilience, Interconnectivity and Security by Satellite), is planned around roughly 300 satellites, against Starlink’s current constellation of more than 10,000⁴. Amazon’s competing Kuiper system, now called Amazon Leo, is expected to reach around 3,000 satellites. That would still leave it far behind, and it would in any case replace one commercially controlled American constellation with another⁴.

An open question for internet governance

Internet governance has developed over three decades around a clear division of work: technical coordination through multistakeholder bodies such as ICANN and the IETF, orbital and spectrum coordination through the ITU’s Radio Regulations and World Radiocommunication Conference processes, and deliberation on new public policy questions through the IGF and its national and regional initiatives, with the Forum’s mandate anchored in the 2005 WSIS Tunis Agenda. This is not a vacuum. ITU coordination is well developed on the allocation question and governs how operators and national administrations share orbital slots and frequency bands. What it does not cover, by design rather than by oversight, is the question this episode raises: the terms on which an operator that already holds a license serves the citizens of a given state and whether those terms may differ from state to state without explanation.
That is an open space rather than a failure. The IGF has been valuable precisely because it does not adjudicate or bind. Its role is to bring questions like this one into view, to gather the technical, commercial, and governmental perspectives needed to understand them properly, and to let norms develop through discussion before anyone writes rules that may not fit the technology. This episode serves that role particularly well. It is a new kind of question: connectivity access rather than content or data, private commercial classification rather than state regulation. It deserves the multistakeholder attention that has worked well for other infrastructure and access questions, rather than a rushed move toward binding treaty obligations that could turn out to be premature or badly matched to how satellite constellations actually work.

Poland’s response so far has been bilateral: public appeals from two ministers to the company and, in Sikorski’s case, to the individual who controls it. That is a reasonable first step, and probably the right one, given how recent and how local the dispute is. Whether it becomes a matter for wider multistakeholder attention at IGF 2026 in Nairobi this December, in ITU study groups on satellite service terms, or elsewhere will depend on whether other states start to report similar treatment. One excluded state is an anomaly worth watching. A pattern would be a governance question deserving exactly the kind of sustained, deliberative attention that the multistakeholder model was built to provide.