ESA put two satellites into low Earth orbit last Saturday and called it a navigation demonstrator. The launch was real. The hardware is up there. But the actual story is the frequency clock ticking in the background — and the commercial company that is moving fastest toward a product while Europe's demonstration phase is still, well, demonstrating.
The European Space Agency's Celeste mission lofted its first two In-Orbit Demonstrator satellites on a Rocket Lab Electron rocket from New Zealand on March 28, 2026, with both spacecraft confirmed to have separated into a 510-kilometer quasi-polar orbit about an hour after liftoff SpaceNews. IOD-1 weighs roughly 20 kilograms and was built by a consortium led by GMV of Spain with OHB Germany as a core partner. IOD-2 is heavier at about 30 kilograms and comes from Thales Alenia Space France as prime contractor, with Thales Alenia Space Italy handling the space segment OHB. The full constellation, once complete, would total 12 satellites — 11 in orbit plus one ground spare — with the next nine spacecraft scheduled for launch by the end of 2027, targeting a 560-kilometer orbit SpaceNews.
Here is what the frequency filing reveals. Under International Telecommunication Union rules, ESA had until May 2026 to bring its L-band and S-band frequency allocations for LEO-based positioning, navigation, and timing (PNT) into actual use, or priority rights to those bands would lapse. Vega-C, the European workhorse rocket, was fully booked at the relevant window. So ESA went to Rocket Lab — a commercial provider, not a European one — and bought a ride on an Electron. "We had an obligation to use the frequencies by May 2026," Roberto Prieto-Cerderia, Celeste program manager at ESA, said at a pre-launch briefing. "In Europe, the main possibility was Vega-C, which was fully booked at the time, so we had to find alternative solutions." The launch was not a demonstration of European launch autonomy. It was a spectrum reservation.
This context matters because the case for LEO-PNT is genuine — and the problem it solves is urgent. GPS and Europe's Galileo constellation both operate in medium Earth orbit, roughly 23,222 kilometers up, where satellite signals arrive weak, with high latency, and are increasingly vulnerable to jamming and spoofing OHB. LEO satellites at 500 to 600 kilometers offer lower signal latency, higher received power at the ground, and a smaller geographic footprint that makes them harder to jam at scale. The tradeoff is coverage: because LEO satellites see less of Earth's surface than MEO birds, you need roughly 10 times as many — 200 to 300 satellites for global coverage, compared to 20 to 30 for GPS or Galileo GPS World.
The jamming problem is no longer theoretical. In the first four months of 2025, nearly 123,000 commercial flights in Europe were disrupted by GNSS jamming — a number that reflects both deliberate interference in conflict zones and the sheer fragility of a system designed in the 1970s GPS World. The U.S. Federal Communications Commission launched a formal Notice of Inquiry in March 2025 to explore resilient PNT alternatives to GPS, acknowledging that dependence on a single navigation system carries national security implications Inside GNSS. SpaceX, in its response to that inquiry, noted that existing Starlink terminals can already deliver nanosecond-level timing accuracy and meter-level positioning using time-of-arrival measurements from its satellites — without relying on GPS at all GPS World.
But Starlink's Ku-band timing is currently too irregular for pseudorange-based PNT, according to Todd Humphreys, a UT Austin researcher who studies satellite navigation security. "At present, Starlink's timing in the Ku band is so irregular that accurate pseudorange-based PNT is not possible," Humphreys said in a filing response, per Inside GNSS. SpaceX has filed for its own PNT service and has the spectrum argument largely figured out — the engineering question is whether the clocks can be disciplined sufficiently.
Meanwhile, a U.S. company called Xona Space Systems is further along than ESA in this race, according to trade reporting. Xona's Pulsar-0 satellite has been in orbit since June 2025 and has achieved a user-equivalent range error of 43 millimeters — more than 10 times the accuracy of GPS — across tracking in more than six countries with 12 third-party receiver prototypes GPS World. The company reported its first precision-timing customer agreements in August 2025 and holds a $4.6 million U.S. defense contract Space Capital. The Pulsar signal is 100 times stronger than standard GPS without causing harmful interference to existing GNSS signals. Xona has a partnership with Trimble, the precision positioning company, to integrate Trimble's correction services with the Pulsar service Trimble press release. The company has raised $92 million in new funding Data Center Dynamics. Its planned 258-satellite constellation has hardware demonstrated in orbit — a pathfinder that puts Xona ahead of any other LEO PNT operator currently filing, per trade reporting.
Xona is not alone. More than 10 entities are now working toward dedicated PNT functions in LEO, with announced constellations totaling over 2,500 satellites if every filing were fulfilled GPS World. SpaceX, Amazon's Project Kuiper, and several defense-oriented ventures are all on that list. The ITU frequency filings are real and the regulatory urgency is real. Trimble's March 2025 press release on the partnership says Xona's service is targeting commercial launch starting in 2027, with active field deployments planned from late 2026 per GPS World. What the record shows is that Xona has hardware demonstrated in orbit, customer agreements signed, and a path to service — the most advanced position of any LEO PNT operator currently filing.
This is the gap between ESA's Celeste mission and where the market is heading. Celeste will validate core technologies and test new signal designs across L- and S-band frequencies — legitimate engineering work. The IOD phase will inform the design of an operational fleet. But operational deployment, according to ESA's own timeline, does not begin until 2028 at the earliest, with the full 12-satellite demonstrator constellation not complete until the end of 2027 ESA. Xona is targeting defense and precision agriculture customers, has a U.S. defense contract and a Trimble partnership covering those use cases, and is working toward commercial timing customers — with service starting in 2027 per Trimble. The gap between "demonstrator" and "product" here is not semantic — it is the difference between filing a frequency priority and being the frequency that commercial and defense customers actually use.
Marco Fuchs, CEO of OHB Germany — one of the two prime contractors on the Celeste consortia — called the launch "a significant step toward a fully functional constellation" that plays "an important role in strengthening Europe's resilience and sovereignty." ESA director Josef Aschbacher framed Celeste as among the first ESA missions to embrace "a New Space-inspired development approach." Both statements are accurate. They also describe a program still in its early phases.
What Celeste does accomplish is establish European priority in a frequency band that is now the subject of an active global land grab. ITU rules reward actual use, not planned use. ESA needed hardware in orbit by May 2026, and hardware is now in orbit. The remaining question is whether Europe's follow-on procurement — the operational constellation that would actually compete with Xona, Starlink PNT, and whoever else enters the market — can move fast enough to matter.
We covered Xona's $170 million fundraise and its Trimble partnership here. The short version: GPS is 50 years old and showing it. LEO-PNT is the proposed upgrade. ESA just arrived at the airport. Xona has been orbiting since June 2025.
