IRIS2 is usually described as Europe’s answer to Starlink. That description is convenient and slightly wrong, and the gap between the slogan and the system explains most of the politics around the program. On 7 August 2026 the European Commission and the SpaceRISE consortium signed the implementation agreement that moves IRIS2 from planning to full-scale deployment. What was signed is a sovereign, secure communications system for governments, not a European consumer broadband network.

What was agreed in August 2026

According to the Commission, the negotiations that ran from January to August 2026 expanded the constellation to 348 satellites: 330 in a higher low Earth orbit and 18 in medium Earth orbit. The additional capacity, 66 more satellites in higher LEO than originally planned, is aimed at defense, security and emergency services. The Commission says the changes bring about 60 percent more secure governmental capacity within the EU and 54 percent more worldwide compared with the original design. The first launches are planned for 2029, with services starting that year.

SpaceRISE is the industrial consortium behind the program, with SES, Eutelsat and Hispasat among its members. The structure is a public-private concession: the original €10.6 billion budget was reported as roughly €6.5 billion in public money and about €4 billion from the private consortium. We did not find published figures for how the August expansion changes that split, so the current total is not stated here.

The push behind IRIS2 sharpened in 2025, when questions about Starlink’s role in Ukraine made European governments confront that a critical communications service could be shaped by decisions taken by one American company and one government. Eutelsat’s chief executive told Politico in 2025 that working with Starlink was a dependence that could be decided in the White House, while also acknowledging candidly that Europe’s operators could not take over Ukraine’s entire connectivity load. Both halves of that statement matter. The first is the argument for IRIS2. The second is the limit on it.

The dependence Europe is worried about is specific. It is about governments, armed forces and emergency services relying on a commercial network they do not control. A ministry cannot plan around a service whose terms, availability or geography can change through a corporate or foreign-government decision. That concern is real, and it applies whether or not Starlink ever cuts anyone off. IRIS2 is designed to remove that single point of outside influence for the users who cannot accept it, which is a narrower goal than matching Starlink’s consumer reach.

There is a bridge to the present as well. IRIS2 is paired with the EU’s GOVSATCOM program, which pools existing government satellite capacity for member states and has been reported to have begun limited initial operations. Ukraine has also asked for access to European secure connectivity, according to press reports. Those are early signals, and they show the intended customer set: governments first, with any wider access following after the constellation exists. Commissioner Andrius Kubilius has said he is confident initial services can start in 2029, which is a political commitment as much as an engineering forecast.

What IRIS2 will not do

The scale difference is large. Engadget’s coverage sets the 348-satellite plan against Starlink’s more than 10,000 active satellites, and one orbit-tracking site put Starlink’s active count near 9,839 as of mid-September 2026. Capacity matters more than satellite count, and here the criticism has been pointed. Analyst Caleb Henry of Quilty Space argued in 2025 that the full constellation as then designed would deliver only about 3.3 terabits per second in total, of which about 2 Tbps from LEO, and less than a competitor that uses fewer satellites. That analysis predates the August expansion, so the current design may differ, but it frames what the system is built for: secure, prioritized capacity for defined users, not mass-market throughput.

Schedule history is the second caution. The same analysis recorded deployment dates that had moved from 2024 to 2026, then 2028, then 2030, and possibly 2031. The August agreement now puts first launches and services in 2029. Aerospace programs slip, and a date agreed at signing is a commitment to be tested, not a fact. Even if 2029 holds, Starlink will be several years further ahead by then.

The launch angle

A sovereign constellation raises the question of what launches it. Arianespace has been reported as expecting to fly 13 Ariane 64 missions for IRIS2 in 2029 and 2030. That figure comes from secondary reporting, not from the Commission, and should be read as an operator expectation. It makes the Ariane 64 a strategic dependency in the same way Starlink is a dependency in the other direction: if the plan for a European constellation relies on a European rocket, the ramp-up of that rocket becomes a policy risk. Ariane 6 has flown four times in 2026 against an operator target of six to eight, and the IRIS2 manifest would ask it to fly a large block of missions on a specific configuration inside two years.

Reading the politics correctly

Three things can be true at once. IRIS2 is a reasonable response to a real vulnerability. It is a smaller and later system than the slogan implies. And it is a government-services program with a commercial layer added on, not a Starlink competitor in the market sense.

The test of the program will not be a launch date. It will be whether governments and defense users actually migrate their critical traffic onto it once it exists, and whether the commercial and citizen services promised for later years get built. For operators and suppliers watching from outside, the useful questions are narrower: which capacity is reserved for governmental users, how the concession pays private investors back when governments are the anchor customers, and whether the launch cadence in 2029 and 2030 arrives on time.