GEOPOLITICA
Who Owns Position and Time — GPS, Galileo, BeiDou, the SpaceX Bet, and What Satellites Would Cost Romania
1. What we're actually buying when we say "GPS"
The vocabulary confusion hides what's at stake. "GPS" is the proper name of the American system. The category is called GNSS — global navigation satellite systems — and the service delivered is called PNT: position, navigation and, most importantly, time.
Time is the part the public ignores and the one critical infrastructure depends on. Stock market transactions are ordered by satellite-synchronized timestamps. Mobile phone networks synchronize their cells. The power grid synchronizes its phase measurements. Banking systems seal their logs. A GNSS receiver is, before anything else, a cheap atomic clock distributed for free to anyone with an antenna.
How much that's worth can be measured. A study commissioned by the American standards institute NIST and conducted by RTI International in June 2019 estimated that GPS generated approximately 1,400 billion dollars in economic benefits in the United States since the 1980s, and that a complete outage would cost the American economy about one billion dollars a day — roughly 30.3 billion for a thirty-day outage, up to 45 billion if the outage were to hit during planting season.
The service is free. However, it is not contractually guaranteed to anyone outside the American government, and the official performance standard — the fifth edition of the civil service specification, April 2020 — promises a signal-in-space error of 4 meters RMS, not legal availability. This is the textbook definition of strategic dependency: an essential resource, free of charge, controlled by someone else.
2. The four global owners and the two regional ones
Today there are four systems with global coverage and two regional ones. The differences between them are not about raw accuracy — all of them fall within a range of a few meters for the open civil service — but about resilience, additional services and, above all, who holds the key.
GPS (United States) operates 31 functional satellites plus one being brought into service (April 2026). The entire GPS III generation has been launched: the ninth satellite on January 28, 2026, the last, the tenth, on April 21, 2026, officially described as "three times more accurate and eight times more jam-resistant" than the previous generation. The military differentiator is the M-code, a higher-power encrypted signal reserved for American forces and authorized allies; M-code receiver testing does not conclude until fiscal year 2027. The next generation, GPS IIIF, is announced for "2028 at the earliest," claiming to be 60 times more jam-resistant — but the ground segment has already slipped from February to November 2026. Notably: the American Congress added 528 million dollars for two more IIIF satellites against the Space Force's request.
Galileo (European Union) showed, as verified on August 30, 2026 on the official page of the GNSS Service Centre, 34 satellites in the constellation, of which 30 usable and 4 unusable. The L14 pair was completed on July 23, 2026, the L15 pair is planned for the fourth quarter. Galileo has two things GPS does not have publicly: the High Accuracy Service (HAS), free and global, which brings the error below 20 centimeters horizontally and 40 vertically in under 300 seconds, and OSNMA — cryptographic authentication of the navigation message, operational since July 24, 2025. OSNMA is, technically, the best civil answer available today to signal spoofing. The encrypted government service PRS is approaching initial operational capability in 2026, and the first certified PRS receiver was announced on March 26, 2025. The second generation, with 12 satellites contracted in two batches — six at Thales Alenia in Italy, six at Airbus in Germany — achieved its first ground connection on July 3, 2026.
GLONASS (Russia) had, as of March 12, 2026, 28 units in orbit, of which 24 operational — exactly the nominal minimum, with no margin. The K2 generation, the first launched on August 7, 2023, has two satellites built and launched and zero operational: still in testing, three years later. A new GLONASS-K was launched from Plesetsk on August 24, 2026. The cause of the stagnation is documented: sanctions cut off access to imported electronics, and between 2020 and 2023 only 4 of the 11 contracted K satellites reached orbit. The accuracy targets of 30 and 50 centimeters were not met, and dependence on Chinese components is growing.
BeiDou (China) is the reverse story. The BDS-3 generation was completed in 2020 with 30 nominal satellites. On March 13, 2026 an in-orbit software update was announced, completed across the entire fleet on July 31, 2026: civil PPP-type accuracy dropped below 0.3 meters without a single new launch. This is proof that, at maturity, the advantage shifts from hardware into software and the ground segment. The ecosystem matches: the Chinese industry linked to BeiDou produced 1.33 trillion yuan (about 195 billion dollars) in 2025, more than 2.2 billion compatible devices are in use in China, and more than 98% of the 133 million phones shipped on the Chinese domestic market in the first half of 2026 supported BDS. The system is exported to more than 140 countries. The BDS-4 roadmap calls for three test satellites around 2027, network satellites around 2029 and completion by 2035.
The two regional systems tell the story of risk. QZSS (Japan) successfully launched QZS-7 on August 10, 2026, but QZS-5 failed to reach orbit in December 2025 due to an adhesive defect on a fuel line of the H3 rocket. The result: Japan has an operational constellation of six satellites, not seven, with no confirmed replacement.
NavIC (India) is the harshest warning in the entire industry. The system was born out of humiliation: in the Kargil war, in 1999, India requested GPS data on enemy positions and was refused by the United States. The program was approved in 2006 and cost about 525 million dollars. In March 2026 the system practically collapsed: the IRNSS-1F satellite reached its projected ten-year lifespan on March 10, and its third and last atomic clock failed on March 13. NavIC was left with three operational PNT satellites, below the minimum of four needed for a position solution — for practical purposes, offline as a standalone system. Six of the eleven NavIC satellites launched have suffered atomic clock failures. The GSLV rocket flies one to two missions a year. The lesson for any state considering space sovereignty: atomic clocks are the single point of failure, and a constellation without a replacement launch cadence is a constellation heading toward extinction.
3. The real hierarchy in August 2026: who's rising, who's falling
If we look at trajectories, not snapshots, the 2026 ranking looks different from the one in 2020.
China is rising fastest, and not through launches, but through software optimization and ecosystem export. The United States is consolidating its military advantage — the M-code, IIIF — but with repeated delays in the ground segment and a Congress funding satellites the service did not request. Europe has closed the technical gap in the civil service and leads on authentication and free-of-charge accuracy, but is left with 30 usable satellites out of 34 and a structural launch problem we will detail below. Russia is stagnating under sanctions. India has visibly regressed. Japan has lost a satellite.
The operational conclusion for a professional user in Romania in 2026: "GPS" no longer exists as such, only a multi-constellation receiver. Any serious equipment — agricultural, surveying, aviation, maritime — uses GPS, Galileo, GLONASS and BeiDou simultaneously and degrades in a controlled way when one of them disappears. Redundancy is no longer an option, it is a procurement specification.
4. The signal war: jamming and spoofing over Romania
This is where the analysis stops being abstract. There are two distinct attacks, frequently confused in the press. Jamming (jamming) means strong noise on the frequency: the receiver loses its solution and knows it has lost it. Spoofing (spoofing) means a counterfeit, coherent signal: the receiver displays a wrong position and time and does not know it. The second is by far more dangerous, because the pilot, the ship's captain or the banking server keeps operating on false data.
The quantitative data is public. The IATA safety report from March 2026 measures a 67% increase in jamming events in 2025 compared to 2023, and a 193% increase in spoofing events. The European Union Aviation Safety Agency, EASA, published on July 3, 2026 the fourth revision of its safety bulletin on GNSS disruptions and alterations, identifying the most affected areas as the Baltic Sea, Eastern Europe, the Mediterranean, the Black Sea and the Middle East. Lloyd's List has documented 1,735 GPS interference events affecting 655 vessels. The Finnish authority Traficom recorded 421 reports in just two months. Lithuania reported an increase in Russian spoofing antennas from three to 36 since January 2025. The European Union sanctioned the 841st Electronic Warfare Centre in Kaliningrad. Thirteen European states plus Iceland issued a joint warning in January 2026.
For Romania, the most solid evidence comes from a technical report published by Spire Global based on its own reception data. In August 2024, a balloon launched from Constanța by a Romanian space engineering company detected spoofing at approximately 11 kilometers altitude, with interference across the entire L1, L2 and L5 bands — the first scientific confirmation of high-altitude GNSS spoofing in NATO airspace. In October 2024, persistent spoofing clusters were identified over east-central Romania, along the Buzău–Ploiești corridor, plus repeated events on October 8, 19 and 26 in the Tulcea–Babadag and Galați–Reni areas. In a single analysis hexagon, up to 399 aircraft were affected. In May 2025, two separate coastal events affected 441 and 385 aircraft respectively. By spring 2025 the frequency had become "almost daily" over the Black Sea, and Romania's defense chief publicly acknowledged weekly interference. The false positions frequently point toward Simferopol, in occupied Crimea, and toward Ayvazovskogo airport. On the civilian maritime side, the AIS system shows vessels appearing on land or spinning in circles. The aviation analysis group OPSGROUP estimates a 500% increase in incidents and approximately 1,500 flights affected daily worldwide.
The causal chain is direct: a war of attrition in the immediate vicinity → systematic electronic warfare as a routine tool → Romania's flight corridor and exclusive economic zone become permanent collateral damage → real costs in civil aviation, maritime transport, precision agriculture and land registry. The honest counter-hypothesis: some of the events could be unintended side effects of Russian drone defenses, not deliberate targeting of Romania. The distinction matters legally, but it does not change the practical effect on the Romanian user, nor the public policy conclusion.
5. SpaceX: from launch provider to global infrastructure
In twelve months SpaceX stopped being a launch company with a secondary internet business and became the reverse: a global infrastructure operator that owns its own launch vehicle.
Scale. By the end of August 2026, a cumulative 12,881 Starlink satellites had been launched. Jonathan McDowell's independent catalog showed 11,102 in orbit and 11,087 functional as of August 27, 2026; the 11,000 threshold was crossed on August 19. In CelesTrak's active object catalog, the 10,989 Starlink satellites represent about 67% of all of humanity's active satellites. There is no historical precedent for a private company operating two-thirds of the functional objects in low Earth orbit.
Commercial. Twelve million confirmed subscribers as of June 4, 2026 across more than 160 countries, more than 13 million announced internally on August 11, 2026, up from 9 million at the end of 2025. In the second quarter of 2026, with 12 million subscribers and an average revenue of 66 dollars per subscriber, the annualized run rate was about 17 billion dollars. The stated target for the end of 2026 is 20 million subscribers — a figure that also appears as 25 million in trade press, a discrepancy we were unable to reconcile and which we flag as such. For fiscal year 2025, SpaceX's total revenue was 18.7 billion dollars, of which Starlink accounted for about 11.39 billion (61%). At the consolidated level, however, the company reported losses in 2025: an operating result of minus 2.6 billion and a net result of minus 4.9 billion dollars — the effect of massive investment in Starship and the third generation of satellites. The distinction matters for the sovereignty discussion: Starlink is the line that brings in the money, but the group as a whole is still burning capital, and a provider that burns capital is a provider that may, at some point, need to reprice its service. The projection for 2026 is about 20 billion dollars from Starlink alone, roughly 85% recurring, plus about 5 billion from launches.
The stock market. From the company's primary investor relations source: the initial public offering price was announced on June 11, 2026, trading began on June 12 under the ticker SPCX on the Nasdaq Global Select Market and Nasdaq Texas, with 555,555,555 Class A shares at $135 — approximately $75 billion gross, the largest listing in history, at a valuation of approximately $1,770 billion. The offering closed on June 15, with an over-allotment option of 83,333,333 shares valid for 30 days. The debut closed up 19%, at $160.95; by the end of August 2026 the stock was trading around $137. The trajectory of the private valuation that preceded the listing: $210 billion in mid-2024, $350 billion in December 2024, $400 billion in July 2025, $800 billion in December 2025.
What comes next technically. The U.S. Federal Communications Commission approved on January 9, 2026 the second tranche of 7,500 second-generation satellites, bringing the authorized total to 15,000. V3 satellites have over 1 Tbps of downlink capacity each and require Starship. Starship's Flight 13 was aborted on July 16, 2026 after four Super Heavy booster engines failed to ignite, but succeeded on the July 24 retry, deploying 20 operational Starlink V3 satellites. On the conventional rocket side: the hundredth Falcon launch of 2026 took place on August 25, and 77 of these were for Starlink; booster B1067 reached its 37th flight, a record. In 2025, SpaceX carried out 165 orbital launches, all successful — more than all the rest of the world's rockets combined, approximately 50-51% of the global total, about 82% of the commercial launch market, 85% of satellites launched, and 2,213 metric tons to orbit, over 80% of the global mass.
Direct-to-phone. The direct-to-device service surpassed 10 million subscribers at the 2026 Mobile World Congress. Over 650 dedicated satellites were operational in April 2026, with a target of about 840. The commercial service with T-Mobile launched on July 23, 2025 at $10 per month for about 60 phone models; it is active in 22 countries and covers over 400 million people. An agreement with operator VEON from November 2025 opens up over 150 million potential customers, with Kyivstar active since the fourth quarter of 2025.
Military. On May 26, 2026, SpaceX received a $2.29 billion contract for the backbone of the Space Data Network — formerly MILNET — approximately 480 satellites, 13 in fiscal year 2026 and 21 in 2027, with a fully operational prototype by the end of 2027, operated by the Space Force. Separately, a $4.16 billion contract for aerial target tracking was reported; the two should not be confused.
The PNT ambition. On May 15, 2025 SpaceX filed a proposal with the U.S. communications authority for Starlink to function as a complementary or alternative source of position, navigation, and timing relative to GPS. Starlink already operates independently of GPS for its own synchronization. It is important to state clearly: as of August 2026 no contract has been awarded on this front. It is a proposal stage, not a capability. But the direction is clear, and if it materializes, it moves the world's most critical public utility — time — into the hands of a publicly traded company.
The risk precedent is already documented. In September 2022, Elon Musk ordered Starlink coverage cut in the Kherson area during the Ukrainian counteroffensive, with direct effects on drones and guided artillery. In March 2025 he stated that Starlink "will never shut down the terminals" and would not use them as a bargaining chip. Over 50,000 terminals had been sent to Ukraine by April 2025, and independent estimates put about 42,000 in use. The issue is not whether the statement is sincere; it is that a critical infrastructure decision for a country at war depends on the will of a single person. This is the definition of uncontracted counterparty risk.
6. Competitors: who else can build a constellation
Amazon Leo, renamed from Project Kuiper on November 13, 2025, had 396 production satellites in orbit as of July 2026, plus two prototypes, launched on Atlas V, Falcon 9, New Glenn, and Ariane 6 rockets. The U.S. authority granted it on June 5, 2026 a limited waiver from the half-constellation deadline due in July 2026, but the deadline for the full constellation — July 30, 2029 — remains in place. The estimated investment is in the range of $10-20 billion. The strategic move: Amazon announced in April 2026 the full acquisition of Globalstar, for approximately $11.57 billion, with completion expected in 2027 — that is, entry into satellite mobile spectrum. Apple, which had held a 20% stake in Globalstar since October 2024 and for which the operator reserves about 85% of network capacity for iPhone emergency messaging, retains its access: Amazon has publicly committed to continuing the service for Apple devices.
Eutelsat OneWeb operates about 648 first-generation satellites and survives through state capital injections. The €1.35 billion capital increase of June 19, 2025, anchored by the French state through the State Shareholding Agency, was extended to €1.5 billion on July 10, 2025 through the United Kingdom's entry with €163.3 million. The French state is now the largest shareholder, with nearly 30%. About €975 million in export-credit-guaranteed debt was added for 440 replacement satellites, plus a January 2026 Airbus contract for another 340 next-generation satellites. Moody's estimates a negative free cash flow of about €500 million per year between 2026 and 2028, with combined OneWeb plus IRIS² investments of about €4.2 billion. The French Ministry of National Defence signed a framework agreement with Eutelsat for LEO access specifically to reduce dependence on Starlink.
Telesat Lightspeed (Canada) is the only Western competitor with fully secured financing: $2.54 billion, cumulatively about $2.7 billion. The pioneer satellite launch is scheduled for the end of 2026, about 96 satellites by the end of 2027, and full global service around the first quarter of 2028.
China is building two parallel constellations. Guowang had about 190 satellites launched by mid-2026, with 310 planned for the current year, 900 in 2027, and 3,600 per year from 2028 onward; the planned total is approximately 13,000. Qianfan has inconsistent figures across sources — 162, 182, or 200 satellites, an unresolved discrepancy — with 324 planned for 2026, authorization from the Brazilian regulator Anatel in December 2025, and an agreement with Airbus for in-flight internet.
AST SpaceMobile is targeting about 45 BlueBird satellites in 2026; units 11-13 were launched on August 5, 2026, with antennas more than three times larger and a peak of about 200 Mbps compared to 98.9 Mbps for the first block.
Iridium reported in the second quarter of 2026 a total of 2,627,000 billable subscribers, up 6% year-over-year, and revenue of $225.2 million. It has a target of at least $100 million per year from PNT services by 2030. On June 28, 2026 it signed a definitive agreement to be acquired by Rocket Lab, with completion expected in mid-2027 — a consolidation that creates an integrated launcher-operator-PNT player.
Rivada deserves to be named separately, as an example of industrial bluff. The company missed the thresholds set by the International Telecommunication Union — 144 satellites by June 2026 and another 144 by September 2026 — without firm financing. It needs approximately $2.4 billion and claims market access in 33 countries and over $16 billion in service agreements. Without satellites, these figures are empty. The lesson for any government evaluating bids: in this sector, the only metric that matters is the number of functional devices in orbit.
7. What comes next: low Earth orbit PNT, quantum, and terrestrial safety nets
The fastest-moving segment is not satellite internet, but low Earth orbit navigation. The physical logic is simple: a classic GNSS satellite sits at approximately 20,000 kilometers, so its signal reaches the ground extremely weak and is easy to drown out with noise. A satellite at 500-600 kilometers delivers a signal orders of magnitude stronger and moves quickly across the sky, which speeds up convergence of the position solution.
The year's regulatory milestone is the full authorization granted on August 3, 2026 to the American company Xona Space Systems for a Pulsar constellation of 258 satellites — the first private operator authorized by the United States to transmit radionavigation signals. The initial authorization covers 16 units, with half the constellation by July 2032 and completion by July 2035. The company raised $170 million in March 2026, has the Pulsar-0 satellite launched in June 2025, and plans six satellites in October 2026 on a SpaceX rideshare mission. Competitor TrustPoint received about $4 million from the Space Force's innovation arm in May 2026 for four satellites and four C-band ground stations, with a trial launch in 2027 and up to 300 satellites in the long term. Iridium already operates STL technology, with a signal reportedly a thousand times stronger than classic GNSS.
Europe is not absent: through the Celeste program, the European Space Agency awarded two contracts of €78.4 million each — one to the Spanish company GMV, one to the French subsidiary Thales Alenia — for an eleven-satellite demonstrator. The first two units, IOD-1 and IOD-2, were launched on March 28, 2026 on an Electron rocket from Mahia, New Zealand, into a 510-kilometer orbit. China has, in parallel, CentiSpace, with about 160-190 satellites planned and ten launched in March 2026, and a 64-satellite constellation from the Geely group with a navigation augmentation layer claiming precision on the order of ten centimeters.
The second direction is navigation without an external signal. On August 27, 2026, the Australian company Q-CTRL announced the first demonstration of GPS-free quantum gravimetric navigation on a ship, in the Coral Sea, with an accuracy of about one nautical mile. A quantum gravimeter reads the map of Earth's gravitational field and determines its position without receiving anything — so it is inherently immune to jamming and spoofing. Separately, Infleqtion together with the British Royal Navy and MSubs demonstrated the first quantum optical atomic clock on an autonomous underwater vehicle. The one-nautical-mile accuracy does not replace GNSS for civil aviation, but it is sufficient for military and maritime navigation and, more importantly, it does not degrade over time the way classic inertial systems do.
The third direction is the terrestrial safety net — and here the United States has a public policy failure that Europe should study. The U.S. Space Force officially terminated the Resilient GPS program on January 19, 2026, a roughly one-billion-dollar initiative, citing funding prioritization; the program had started with four "Quick Start" agreements signed on September 23, 2024 with Astranis, Axient, L3Harris, and Sierra Space, for eight satellites with possible launch from 2028 and $100 million allocated through the 2024 defense budget law. Congress restored only $15 million for resilient GPS and another $15 million for a demonstration of commercial PNT services. As for eLoran — the terrestrial long-wave navigation system mandated by the U.S. Timing Resilience law of 2018 — the most recent verifiable status, from November 2021, showed it had still not been deployed. Sources from 2023-2024 report that China has completed its national eLoran network. This is the most important information gap remaining in this analysis, and we flag it as such.
8. The market in numbers
The European Union Agency for the Space Programme published on May 26, 2026 its first report bringing together navigation, Earth observation, secure communications, and space surveillance across sixteen market segments. Key figures: global GNSS revenues grow from €300 billion in 2024 to €580 billion in 2034; GNSS devices, from 5.8 billion units to about 10 billion; Earth observation, from €3.5 to €7.9 billion, with agriculture as the dominant segment; secure satellite communications, from over €200 million in service revenue in the Union in 2025 to about €1.2 billion in 2040.
The 29th annual report of the satellite industry association, published in mid-May 2026 with data for 2025, measures the global space economy at $429 billion, up 3%, of which the commercial satellite sector accounts for $303 billion (71%). Commercial launch revenue grew 33%, to $12.4 billion. 2025 was an absolute record year, with 4,434 satellites launched, 65% more than the previous year.
The correct interpretation of these figures: the market is not dominated by satellite sales, but by downstream applications. Of the €580 billion projected for GNSS in 2034, the satellite share is marginal; the value lies in chips, receivers, agricultural services, logistics, insurance, and synchronization. This is exactly why China built an ecosystem, not just a constellation, and why a national strategy that boils down to "we want a satellite" misses 95% of the value.
9. Europe: independence on paper, dependence in practice
Europe has the best civilian navigation system in the world and cannot launch it on its own on time. This paradox is the entire story of European space sovereignty.
The concrete fact: four full operational capability Galileo satellites were launched in two pairs, in April and September 2024, on SpaceX Falcon 9 rockets, for approximately €180 million — because Ariane 6 was not ready and access to Soyuz had been cut off. The symbol of European technological sovereignty reached orbit on a private American rocket. There is no clearer illustration of the gap between ambition and capacity.
Launchers. Ariane 6 first flew on July 9, 2024, with a partial failure, and had accumulated nine launches by August 30, 2026, of which eight were complete successes. The target for 2026 is six to eight launches; four had flown by the end of August. The list price is around 100 million euros for the Ariane 62 variant and 115 million for the Ariane 64, versus 74 million dollars for a basic commercial Falcon 9. The European Space Agency paid around 410 million dollars in subsidies for four Ariane 6 launches in 2024. Vega C returned to flight on December 3, 2024, after the failure in December 2022, and has seven flights with six successes; Avio received 350 million euros for Vega C and Vega E.
For the European Launcher Challenge, 902.16 million euros were subscribed at the 2025 ministerial conference, and the contracts were signed in August 2026: 197.8 million for Isar Aerospace, 186.9 for Rocket Factory Augsburg, 158.9 for PLD Space, with MaiaSpace pending. The reality of execution is more sobering: Isar's Spectrum rocket failed about 30 seconds after liftoff, on its first flight, on March 30, 2025, and the second flight, scheduled for August 31, 2026, had not taken place; RFA was postponed to September 30, 2026, MaiaSpace to April or the second half of 2027; PLD Space raised 180 million euros in March 2026 and is targeting the first Miura 5 flight "by the end of 2026" from Kourou, with 540 kilograms to low Earth orbit. And the British company Orbex has dropped out of the race: it entered administration proceedings in February 2026.
IRIS², the secure European constellation, got more expensive before it even existed. The baseline approved in December 2024 was 10.6 billion euros — 6 billion from the Union budget, over 4 billion from the SpaceRISE industrial consortium and 550 million from the European Space Agency — for 290 satellites. The revised implementation agreement of August 7, 2026 raises the total investment to over 15.6 billion euros and the constellation to 348 satellites: 330 in low Earth orbit and 18 in medium orbit. Public funding now exceeds 11.6 billion, drawn from the 2028-2034 European budget. The first launches were moved from 2030 to 2029. Industrial breakdown: Eutelsat 2.23 billion, SES 1.35 billion as leader of the medium-orbit segment, Hispasat 600 million.
The arithmetic is merciless. IRIS² will have 348 satellites around 2030-2031. Starlink has 11,100 today and authorization for 15,000. IRIS² is not a commercial competitor to Starlink, nor does it claim to be: it is a secure governmental capability, with European-controlled cryptography. Judged as such, it makes sense. Judged as the "European Starlink", it is marketing.
Consolidation. SES completed the acquisition of Intelsat on July 17, 2025 for 2.6 billion dollars in cash, with a total enterprise value of 5 billion, creating an operator with around 120 satellites in geostationary and medium orbit, and synergies estimated at 2.4 billion dollars in present value. Viasat had absorbed Inmarsat in May 2023 for 7.3 billion dollars. Consolidation is the classic defensive response to a new entrant that redefines unit cost.
What already works. GOVSATCOM, the European mechanism for pooling governmental satellite communications capacity, became operational in January 2026 through an interim hub operated by the European agency, aggregating capacity from five member states — France, Greece, Italy, Luxembourg and Spain — with permanent sites in Cologne and Athens; Cyprus achieved the first operational use. On the navigation side, Galileo's encrypted governmental PRS service guarantees member states unlimited access, worldwide and by right — provided each state establishes a National PRS Competent Authority.
The latter is the cheapest form of sovereignty available to a European state, and it is worth emphasizing: it does not cost a satellite, it costs an administrative structure and certified receivers.
10. Romania: three cubes, a teleport and an eight-year program on paper
The factual, verified record looks like this.
What flies. Goliat, Romania's first satellite — a 10-centimeter, one-kilogram cube built by students — was launched on February 13, 2012 on the inaugural flight of the Vega rocket. It cost at least 1.5 million lei in public money, around 330,000 euros. Telemetry was received on February 18, then contact was lost for good; the agency's leadership attributed the failure to the fact that Vega placed it at 1,500-1,800 kilometers instead of 400-450, exposing unhardened components to radiation. It reentered the atmosphere uncontrolled on December 31, 2014. The second, ROM-2 or "Space Sparrow", a 5-centimeter, 250-gram cube with a 2-megapixel camera built by a group of nine high school students, was launched in June 2023 on a rideshare mission. The third, EMISAR, a 1U cube for demonstrating store-and-forward digital messaging between ground stations, was launched on March 30, 2026 on the Transporter-16 mission of a Falcon 9 from Vandenberg, built by the Institute of Space Science with Romanian partners and with ground stations in Măgurele and Constanța.
This is Romania's entire orbital inventory: three cubes, together worth a few hundred thousand euros. The RoBiSAT program, with two 2U cubes for the QB50 constellation, was postponed indefinitely and never flew.
What exists on the ground and actually works. ROMPOS, the national network of permanent GNSS stations, launched by the National Cadastre Agency in September 2008 and administered since March 2018 by the National Center for Cartography, has between 75 and 86 permanent stations on national territory, with 97 included in processing, and offers 2-3 centimeter accuracy in real time. It is part of the Class A National Spatial Geodetic Network, in the ETRS89 system. It is real, functional infrastructure that is strategically underused — a network of reference stations is exactly the kind of sensor that can detect and characterize GNSS spoofing at national scale.
The Cheia satellite communications center, inaugurated in October 1976 and operated by the state-owned radiocommunications company, is described as the largest teleport in Central and Southeastern Europe, with seven ground stations in the C and Ku bands. The two 32-meter antennas that have been operating there since 1977 are being converted by RARTEL S.A. into a quasi-monostatic radar for space object surveillance and tracking. RARTEL is a Romanian-Italian joint venture from 1996, with Telespazio holding 62% and the state-owned radiocommunications company 38%, which supplies the satellite communications of the Ministry of Defense and carried out the preparatory activities for Romania's participation in GOVSATCOM. Romania also operates the national mirror of the Copernicus collaborative ground segment, operational since November 30, 2018.
What does not exist. Romania became the nineteenth member state of the European Space Agency on December 22, 2011, following the agreement signed in Bucharest on January 20, 2011 and after the cooperating state status of 2006-2007. At the November 2025 ministerial conference in Bremen, member states subscribed a record 22.3 billion euros. Romania's aggregate contribution is not published as a single figure — there are only separate lines per program — and the trade press reports, without independent confirmation, that Romania has reduced its subscription compared to previous levels, with other states covering the difference. We flag this information as unconfirmed.
We could not identify any Romanian financial commitment to IRIS², while Poland committed 656 million euros, Hungary 500 million, and Spain between 1.6 and 2 billion. There is no Romanian contract of the Poland-ICEYE or Greece-Open Cosmos type for observation satellites. There is no standalone national space strategy document — unlike, for example, Greece's new national strategy through 2035. Romania's participation in NATO's persistent space surveillance program, APSS, is reported by the press based on the memorandum from the July 2024 Washington summit, but we could not confirm it in a primary NATO document, and we flag it as such.
The program that has existed on paper for eight years. The Military Satellite Telecommunications System — SSMT — started through a collaboration protocol between the Ministry of Defense, the Romanian space agency and the European one, signed on November 22, 2018, for a telecommunications satellite in geostationary orbit intended for the national security system. The justification published at the time: between 2009 and 2018 Romania had spent around 7.5 million euros leasing satellite communications services from third parties, and the program was meant to "strengthen Romania's role in NATO and the EU and reduce dependence on external suppliers". The target for the first phase was the first quarter of 2022.
What has happened since then, according to a specialist analysis from February 15, 2026 that compiles ANCOM reports from 2019-2024, reports from the Supreme Council of National Defense, parliamentary inquiries and communications from the Ministry of Defense: in 2019 the regulatory authority started the procedure for activating the orbital positions; in 2020 the Supreme Council approved the continuation of the program; in January 2023 no funding had been allocated for the production and launch of the satellite; in April 2025 the program was still at the substantiation stage; and on November 24, 2025 the Supreme Council approved the financing of SSMT through the European SAFE instrument. Romania has five allocated geostationary orbital positions, between 26.5 and 53.5 degrees east. The estimated contract value is approximately 420 million dollars excluding VAT, with delivery within 36 months of signing; exceeding the threshold of 100 million euros, the acquisition requires Parliament's approval. The minimum guaranteed lifespan is 15 years, and the projected operational cost is around 26.6 million euros per year.
Eight years, five reserved orbital positions, zero signed contracts. The causal chain behind the deadlock is not technical, but budgetary and institutional: without a dedicated multiannual funding line, a 420-million-dollar program that goes through parliamentary approval competes in every budget cycle with acquisitions that have immediate political effect. The counter-hypothesis worth mentioning: the SAFE instrument changes exactly this equation, because it offers long-maturity loans dedicated to defense, and the November 2025 approval could be the unlocking moment. The exact amount allocated to Romania through SAFE could not be verified against an accessible primary source, and we do not assert it.
11. What it would cost to have satellites: real prices, by tier
The question "how much does a satellite cost" does not have one answer, but six. Here are the real prices, with contracted examples, not brochure estimates.
| Tier | Cost (build + launch) | Actual contracted example | What you get |
|---|---|---|---|
| CubeSat 1U-3U, minimal mission | 150,000 - 300,000 € | Goliat, Romania, 2012 (≈330,000 €); forestry nanosatellite planned for 2027 (150-200,000 € + ≈100,000 € launch) | Technology demonstration, personnel training. No operational value. |
| CubeSat 6U-12U, commercial quality | 0.7 - 2.5 mil. € | Illustrative industry models; no Romanian example at this tier | Medium-resolution imagery, IoT, messaging. Niche utility. |
| Optical or radar EO satellite, single unit | 5 - 70 mil. € / satellite | Greece: 4 thermal satellites 20 mil. € (≈5 mil./unit); 2 radar satellites 33 mil. € (≈16.5 mil./unit). Poland PIAST ≈70 mil. PLN. | Real observation capability, but rare revisit with a single unit. |
| National EO micro-constellation (3-13 satellites) | 25 - 350 mil. € | Portugal 24.3 mil. € (3 satellites); Poland MikroGlob 556.7 mil. PLN ≈ 130 mil. € (4 satellites, signed Dec. 20, 2024); Greece 130 mil. € (13 satellites) → 350 mil. € successor program announced Jun. 29, 2026 | Useful revisit, national industry, data independence. Best value/cost ratio for a state the size of Romania. |
| Military-grade radar constellation | 165 - 200 mil. € | Poland - ICEYE: 200 mil. € for 3+3 radar satellites, signed May 14, 2025, delivered within 12 months; Luxembourg NAOS: 168.2 mil. € for a single top-tier optical satellite | All-weather, day-and-night observation. Direct relevance for the eastern flank. |
| Military GEO communications satellite | ≈420 mil. $ - 2 mld. € | Romania SSMT ≈420 mil. $ (estimate, uncontracted); Astranis MicroGEO - Chunghwa Telecom 115 mil. $ (one satellite with launch); Spain SpainSat NG ≈2 mld. € (multi-satellite program + 15 years operation) | Sovereign communications for deployed forces. Operational cost ≈26.6 mil. €/year. |
| Purchase of imagery as a sovereign service | over 90 - 460 mil. € multiannual | Sweden - Planet Labs: over 100 mil. $ multiannual, announced Jan. 12, 2026, part of a total of over 500 mil. $ with Sweden, Japan and Germany | Immediate capability, zero build risk. Zero national industry. |
| Joining a multinational constellation | 500 mil. - 2.3 mld. € (commitment level) | IRIS²: Poland 656 mil. €, Hungary 500 mil. €, Spain 1.6-2 mld. €. NATO APSS: over 1 mld. $ allied contributions, initial operational capability Dec. 4, 2025 | Access to capability far beyond what a medium-sized state can build alone. No exclusive control. |
A few launch benchmarks, for anyone who wants to do the math component by component. A slot on SpaceX rideshare missions was quoted in February 2026 at 350,000 dollars for up to 50 kilograms, plus 7,000 dollars per kilogram above this threshold. A dedicated Falcon 9 has a list price of 74 million dollars in 2026, versus 70 in 2025 and 67 in 2022. A Vega C rocket costs around 48 million euros as a reference price; the European Space Agency actually paid 51.65 million for the Sentinel-1C mission, and a package of five Copernicus launches was contracted at 232 million, i.e. around 46.4 million per flight. A dedicated Electron rocket costs around 7.5 million dollars for about 300 kilograms. The ground segment is paid by the minute: 5-15 dollars per minute of contact, which means 20,000-60,000 dollars per year for a satellite with four to six daily passes, plus 50,000-200,000 dollars per year for mission operations software.
And a comparison that shifts the perspective on the entire discussion: the US Government Accountability Office documented that the CHIRP program, a military payload hosted on a commercial satellite, saved about $300 million compared to a dedicated satellite. A NASA case study compared an owned satellite worth about $500 million to a hosted alternative worth about $65 million that met approximately 80% of requirements. The ratio is nearly eight to one for four-fifths of the capability.
12. The five realistic alternatives for Romania
Starting from the data above, here are the options, ordered by the ratio between the strategic effect achieved and the money spent.
1. National Competent Authority for PRS and encrypted Galileo receivers. Cost: under €10 million. Effect: high. Romania already has, by treaty, unlimited access to Galileo's encrypted government service. What is missing is the national structure that issues, distributes and manages the keys, plus certified receivers. It is the only measure on this list that offers spoofing-resistant navigation for the armed forces, border police and critical infrastructure without a single satellite. It should have been done already.
2. Transforming ROMPOS into a national GNSS integrity sensor. Cost: under €5 million. Effect: high. The 75-86 reference stations already exist, are powered, connected and geographically distributed. Adding a layer of interference and spoofing monitoring turns a cadastral network into an early-warning system that can issue alerts to aviation, maritime transport and agriculture. Romania would become a safety-information provider for the entire eastern flank, in a domain where it is the most affected NATO state and therefore best placed to gather data.
3. Imagery acquisition as a sovereign service, on the Swedish model. Cost: €30-100 million over five years. Effect: medium-high, immediate. Sweden's agreement with Planet Labs, worth over $100 million, allocates dedicated satellites to a state without the state owning them. The capability begins in months, not years. The drawback, which must be accepted explicitly: zero national industry and dependence on a foreign commercial provider — that is, exactly the counterparty risk demonstrated by the Kherson episode, shifted from communications to imagery.
4. National observation micro-constellation, on the Polish or Greek model. Cost: €130-200 million. Effect: high, in 2-4 years. This is the option with the best value/cost ratio at the ambition level of a state the size of Romania. Poland contracted four optical microsatellites with ground segment for about €130 million and, separately, six radar satellites for €200 million delivered in twelve months. Greece obtained thirteen satellites for €130 million, using the European recovery fund, with the program managed together with the European Space Agency, and announced a €350 million successor. Both were achieved by states comparable in economic size to Romania, in the last twenty months, with European money. The precondition is exactly what Romania lacks: a strategy document and a program structure capable of contracting.
5. SSMT, the geostationary military satellite. Cost: ≈$420 million plus ≈€26.6 million per year. Effect: high on communications, zero on observation and navigation. It is the only large program actually on the table and has approval for funding through the European SAFE instrument as of 24 November 2025. However, it must be said clearly what it does not solve: a geostationary communications satellite does not provide imagery, does not provide navigation and does nothing against GNSS spoofing over the Delta. It solves a single problem — dependence on leased satcom capacity — and it solves it expensively, with a single unit, therefore without redundancy, in an orbit where a single satellite is by definition a single point of failure.
A sixth option deserves mention because it is the most underrated: the hosted payload. The eight-to-one ratio documented by US agencies suggests that a Romanian instrument — for example an interference-monitoring receiver, or a camera — placed on an already-planned commercial satellite would deliver a large part of the value at a fraction of the cost. We found no European or Romanian case study along this line, which is itself an indication that the option has not been evaluated.
What the data does not recommend: repeating the CubeSat model. Three cubes in fourteen years, of which the first died immediately, does not constitute a capability. They have real educational value and have trained engineers who work today in European programs, but confusing them with space sovereignty is the most frequent public-communication error in this field.
13. What would invalidate this analysis
Intellectual honesty requires listing the conditions under which the reasoning above breaks down.
If SpaceX actually obtains a PNT supply contract and operationalizes it, the entire sovereignty calculation changes: navigation becomes a commercial service with a contract and penalties, which is, paradoxically, more legally secure than today's free GPS signal. If quantum navigation drops below the cost and size of standard onboard equipment within the next five years, the entire discussion about backup constellations becomes irrelevant for military applications. If IRIS² slips again and exceeds €20 billion, the argument to "join the European program" weakens significantly compared to "buy commercial capacity." If the SAFE instrument proves more flexible than it appears, the priority order from point 12 could reverse, because long-maturity money changes what "expensive" means.
We also flag the limits of the factual base: we were unable to verify on primary sources Romania's aggregate contribution to the European Space Agency, the amount allocated to Romania through the SAFE instrument, Romania's confirmed participation in the NATO APSS program, the current status of the American eLoran network, the generational breakdown of the roughly 50 operational BeiDou satellites, and the exact number of Qianfan satellites, where sources indicate 162, 182 or 200. None of these gaps change the main conclusions, but all are marked as such in the text.
14. Falsifiable predictions
Ten predictions with probabilities calibrated on the ICD-203 scale, an explicit horizon and a public verification criterion. They will be scored at resolution using the Brier score.
| # | Prediction | Horizon | Probability | How it is verified | Verification source |
|---|---|---|---|---|---|
| F1 | Starlink will publicly announce at least 14 million subscribers | 31 Dec 2026 | likely (70%) — from 9 mil. in December 2025 to over 13 mil. in August 2026, pace ≈0.5 mil./month | Official press release, post-listing quarterly reporting, or published company statement | SpaceX investor relations; SEC filings |
| F2 | IRIS² will not have a single operational satellite in orbit | 31 Dec 2028 | very likely (90%) — the implementation agreement of 7 Aug 2026 fixes the first launches for 2029 | Absence of any confirmed IRIS² launch in orbital catalogs | European Commission; ESA; CelesTrak catalog |
| F3 | Romania will not sign the SSMT procurement contract | 30 Jun 2027 | likely (65%) — eight years without a contract, parliamentary approval threshold, feasibility phase in Apr 2025 | Absence of a contract-award announcement and of parliamentary approval | Ministry of National Defence; SEAP; parliamentary transcripts |
| F4 | Galileo will have at least 30 satellites marked "usable" simultaneously | 30 Jun 2027 | likely (70%) — 30 usable as of 30 Aug 2026, the L15 pair in Q4 2026, the first G2 in 2026-2027 | Reading the official constellation status page on the deadline date | European GNSS Service Centre |
| F5 | NavIC will remain below 4 operational PNT satellites | 31 Dec 2026 | very likely (85%) — 3 operational as of March 2026, GSLV flies 1-2 missions per year | Constellation status published by the Indian space agency | ISRO; NavIC bulletins |
| F6 | The EASA safety bulletin on GNSS interference will remain active for the Black Sea region | 31 Dec 2027 | very likely (90%) — four successive revisions, near-daily frequency, root cause unresolved | EASA domain page on GNSS outages and alterations | EASA |
| F7 | Xona Space Systems will have at least 6 Pulsar satellites in orbit | 30 Jun 2027 | likely (70%) — full authorization 3 Aug 2026, $170 mil. raised, 6 satellites scheduled Oct 2026 | Count in public orbital catalogs | CelesTrak; Xona press releases |
| F8 | Amazon Leo will have under 1,000 satellites in orbit | 31 Dec 2027 | likely (70%) — 396 in Jul 2026, pace ≈26/month, waiver already granted for the 2026 threshold | Count in public orbital catalogs | CelesTrak; FCC filings |
| F9 | No Romanian financial commitment for IRIS² will be publicly announced | 31 Dec 2027 | likely (65%) — no commitment identified as of Aug 2026, unlike Poland, Hungary and Spain | Absence of a government or Commission statement naming an amount for Romania | European Commission; Government of Romania |
| F10 | Ariane 6 will carry out at most 8 launches in calendar year 2026 | 31 Dec 2026 | very likely (85%) — 4 flights by the end of August, the announced target being 6-8 | Count of Ariane 6 launches in 2026 in public registries | Arianespace; ESA; launch registries |
15. Methodology and sources
Anchor data (public, verified as of 30 August 2026): the official Galileo constellation status page of the European GNSS Service Centre (34 satellites, 30 usable, 4 not usable); the SpaceX investor relations page on the initial public offering price (11 June 2026, 555,555,555 shares at $135); the EUSPA report on the European space market, published on 26 May 2026; the 29th annual report of the satellite industry association, May 2026, with data for 2025; the Spire Global report on GNSS interference in the airspace of the Black Sea and Romania; the EASA page on GNSS outages and alterations, revision 4 of 3 July 2026; the IATA safety report from March 2026; the RTI International study for NIST on the economic value of GPS, June 2019; the CelesTrak active-object catalog and the independent launch record maintained by Jonathan McDowell; OHB's ad-hoc announcement on the NAOS contract in Luxembourg (€168.2 million); the ESA press release on the record subscriptions from the November 2025 ministerial conference (€22.3 billion); the Copernicus page on the collaborative ground segment for Romania; the RARTEL programs page; the ANCOM regulatory reports 2019-2024 and the CSAT reports, via the compilation published on 15 February 2026; the defenseromania.ro press release of 22 November 2018 on the SSMT protocol; the primary documents of the 2022 and 2025 ESA ministerial conferences; the ROMPOS page of the National Centre for Cartography.
Method: aggregation of primary sources where they exist (space agencies, regulatory authorities, company press releases, ministerial documents) and explicit flagging of every piece of information originating from secondary press or industry syntheses. Every contradictory figure found in sources is flagged as such in the text, not silently reconciled — the main cases being the Starlink 2026 subscriber target (20 vs. 25 million), the number of Qianfan satellites (162, 182 or 200), the generational breakdown of the BeiDou fleet, and the SSMT contract values. Prediction probabilities are expressed on the standardized ICD-203 scale and will be scored via Brier score at the resolution date, through a separate record, not by modifying this article. Cost comparisons use exclusively publicly announced contracts, not proprietary estimates; where only industry estimates exist, this is stated explicitly.
Disclaimer: This material is for informational and analytical purposes. It does not constitute investment advice, a public procurement recommendation, or a technical system evaluation. The satellite cost figures come from contracts concluded in other jurisdictions and under other market conditions; they indicate orders of magnitude, not prices directly applicable to a Romanian procurement. The $420 million estimate for SSMT is a reported value, not a contracted price. The analysis reflects public information available as of 30 August 2026 and may be superseded by subsequent events.