Orbital Compute: 151,600 Satellites Filed, No Flight Data Yet
Two companies have now told the United States government, in writing, that they intend to put more than 150,000 satellites into orbit to run computers. Orbital Compute, Inc. filed for up to 100,000 of them in late June 2026. The filing is sized to deliver roughly 10 gigawatts of compute capacity once built. Blue Origin's Project Sunrise filed for up to 51,600 in March 2026. It drew a formal objection from the National Aeronautics and Space Administration, NASA, in May, over collision risk and an absent debris mitigation plan. Add the two filings together and the total is roughly 151,600 satellites. That is larger than every satellite currently operating in Earth orbit combined: about 16,810 as of 17 September 2026, by the independent count Jonathan McDowell's Space Report has kept for years [7].
Neither company has published an independently measured figure for what its hardware does once it reaches orbit. There is no benchmark, no demonstrated inference rate, no watt of computed output that anyone outside the company can check. What exists instead is paperwork: two applications to a regulator, each establishing a legal claim to spectrum and orbital slots. Neither establishes that the constellation works, that the economics close, or that it will be built at the filed scale.
That distinction, a regulatory filing standing in for a flight record, is not being made in most of the coverage either filing has received. It is the actual story this week. New Zealand's own position in it is unusually exact: the country that launches more small satellites than almost anywhere else in the Western world has not carried a single one of the payloads these filings describe.
Two Filings, One Objection
The two filings did not arrive together, and they read as two different bets. Orbital Compute, trading as Orbital, proposes roughly 100-kilowatt-class satellites in low Earth orbit at 500 to 850 kilometres. Each satellite is about 1.5 to 2.5 tonnes, with solar arrays and radiators spanning around 100 metres. It draws power independently of any terrestrial grid, land or water [1][3]. The company frames the pitch as solving three constraints in a single move, power, land and water, all limited on the ground and abundant, in principle, above it. Its stated focus is AI inference rather than training, on the reasoning that inference workloads can be split and distributed across separate orbital nodes in a way training cannot.
Blue Origin's Project Sunrise proposes a constellation in sun-synchronous orbits between 500 and 1,800 kilometres to host data centres in space [2]. It is not a stand-alone idea. TeraWave, a separate 5,408-satellite communications constellation Blue Origin announced in January 2026, is designed to carry connectivity for Project Sunrise and for terrestrial customers. TeraWave is not itself a data centre network.
NASA's objection, filed with the Federal Communications Commission on 5 May 2026, is the more consequential development. NASA argues that Project Sunrise's proposed altitude band overlaps crewed spaceflight paths and national science assets. Its filing cites the absence of an orbital debris mitigation plan directly [4]. Two other organisations, DarkSky International and the Center for Space Environmentalism, are reported to have lodged their own objections on astronomical interference and environmental-review grounds. Those are background to this week's finding. The Commission has not ruled on any of it, and has not fully approved an orbital data centre proposal of any kind to date.
Set against Orbital Compute and Project Sunrise are two filings already on record from earlier in the year: SpaceX's application for roughly one million satellites, and Starcloud's for 88,000. Across all four, the regulator has approved none in full.
The three threads matter now because they have not previously sat inside the same publication week. Orbital Compute's filing became public in June. Project Sunrise's application and NASA's objection to it landed in March and May respectively. Rocket Lab's own 2026 manifest has been accumulating launch by launch since January. This week, a third named filer, a formal government objection, and New Zealand's own counter-evidence about what has actually flown all became checkable against each other for the first time, inside one docket and one launch schedule.
The Governance Gap, One Order of Magnitude Wider
The pattern is not new to this series. In June 2026, Project Sunrise sat alongside SpaceX's roughly one-million-satellite application and Starcloud's 88,000-satellite filing as evidence that no framework governs jurisdiction, debris or accountability once a company decides to build at orbital scale. That gap has not closed in the three months since. NASA's objection is the first instance of a government body directly contesting one of these filings on safety grounds. But an objection is a comment inside a regulatory process, not a rule. The underlying governance question stands: who decides how much orbital capacity one company may claim before anyone has confirmed the hardware works.
Read against the pattern of corporate actors pursuing orbital-scale resource capture ahead of demonstrated capability, two additional filers arriving inside three months is the same pattern's third and fourth instance. The interesting number is not 151,600 on its own. It is 151,600 against roughly 16,810 active satellites: the filed figure is already about nine times the entire population of everything humanity currently has in orbit. That comparison does not prove either constellation is unrealistic. It shows that the claim, on paper, has already run an order of magnitude ahead of anything flying.
The clock that has actually moved this year is the regulatory one. Neither the clock that measures hardware built and flown, nor the one that measures demonstrated economics or performance, has kept pace with the clock that measures pages filed at a regulator. Three named filers and a formal government objection have advanced faster in 2026 than a single orbital-compute payload has flown anywhere these companies operate. Normally a regulator lags a technology that already exists. Here, the regulatory process is the most concrete thing that exists.
A filing establishes standing. It does not establish capability. The same gap runs through how any organisation reads a vendor's roadmap slide, a partner's press release, or a competitor's funding announcement as though it described a shipped product. The FCC docket is simply the clearest version of that gap currently on public record, because every stage is dated, filed and searchable, and because NASA has now put a second party's name against the first party's claim in writing. The same claim mistaken for a measurement recurs wherever an evaluation happens once and gets treated as though it happens continuously.
None of this argues that Orbital Compute or Project Sunrise will fail. Companies file ambitious applications, spend years building toward them, and sometimes deliver at or near the filed scale; that is a legitimate part of how large infrastructure gets built. The argument is narrower and harder to dismiss: until one of these companies publishes a measured figure for what its hardware does once it is actually in orbit, running its own operating economics rather than an investor pitch, the honest description of either constellation is what is on file at the FCC, not what is running above the atmosphere.
What New Zealand Actually Supplies
New Zealand's role in this story is structural rather than participatory, and the difference is worth stating precisely. Rocket Lab's Māhia launch site, Launch Complex 1, is the highest-cadence small-satellite launch site operated by a Western company anywhere in the world. On 19 September 2026 it flew its 96th Electron mission overall, "Owl By The Dozen," placing the twelfth StriX synthetic-aperture-radar satellite for Japan's Synspective into a 572-kilometre orbit. Fifteen further Synspective missions are still to fly [5][6]. Every 2026 Māhia payload this article can identify by name is an Earth-observation or communications satellite. None is an orbital-compute or AI-inference payload of the kind Orbital Compute or Project Sunrise describes.
That is not a criticism of Rocket Lab's choices. Asked directly on an earnings call whether the company would build orbital data centres itself, its chief executive answered that it was not ready to yet. It is, instead, the cleanest available counter-evidence that satellite counts in the hundred thousands are close to reality. The country supplying the launches for this entire sector has not yet carried the hardware the sector's two largest current claims describe.
Rocket Lab's own commercial position makes the absence more notable, not less. This is a company that manufactures solar cells explicitly marketed for data-centre power applications, sells components into the wider space economy, and speaks fluently about orbital infrastructure ambitions in its own investor materials. It has still not put an orbital-compute payload on a rocket, from Māhia or anywhere else, in 2026. The highest-cadence Western small-satellite launcher has not carried one. The constellation counts filed at the FCC describe intentions considerably further from the launch pad than the raw satellite figures, taken alone, tend to suggest.
New Zealand's own regulatory position sits at a different layer entirely. The Outer Space and High-altitude Activities Amendment Act 2025 brought Ground-Based Space Infrastructure, GBSI, provisions into force on 29 July 2025. Existing operators' temporary authorisations lapse on 29 July 2026, after which formal GBSI activity authorisation is required, with penalties of up to NZ$250,000 for an entity and up to a year's imprisonment for an individual [8]. Those are two distinct dates a year apart, not one. The regime governs ground-segment infrastructure supporting space objects, not who may claim an orbital slot; it answers a different question from the one currently sitting with the FCC.
Two further facts describe the infrastructure New Zealand actually depends on, in contrast to the paperwork this article has spent most of its length on. New Zealand has operated a dedicated AWS Asia Pacific (New Zealand) Region since September 2025, with three availability zones: real, landed cloud infrastructure, not a filed intention. New Zealand's international connectivity, meanwhile, runs almost entirely through a small number of submarine cables, the Southern Cross Cable Network and Hawaiki among them, with Hawaiki Nui under construction. One industry estimate puts the share of international data traffic those cables carry at around 99 per cent. Satellite links offer a fraction of the capacity at higher latency, as the alternative [9].
The practical test for any organisation evaluating a claim like Orbital Compute's or Project Sunrise's, or a vendor's roadmap slide closer to home, is the same four-step check this week's filings fail to clear on their own: has it been filed, has it been funded, has it flown, and has it been measured. A filing satisfies the first test alone.
What to Watch
What happens next runs on the regulator's timetable, not the filers'. The FCC has not ruled on Project Sunrise, has not ruled on Orbital Compute, and has never fully approved an orbital data centre proposal from any applicant. Three things are worth watching. First, whether NASA's objection produces a formal condition on Project Sunrise's licence, such as a mandated debris mitigation plan, or whether it is answered and set aside. Second, whether either Orbital Compute or Project Sunrise publishes an independently verifiable performance figure before its next funding round. A measured number, not a projected one, would be the first genuine change of clock this series has seen from either company. Third, whether a fifth filer joins SpaceX, Starcloud, Blue Origin and Orbital Compute before the FCC rules on any of the first four. That would widen the gap between what is claimed and what is confirmed, rather than close it.
None of the three will resolve before this article is published, and none should be mistaken for progress simply because a filing moved through a docket. The docket is the paperwork. The satellite in orbit, actually computing something, running its own economics, is the flight data. Treating one as a stand-in for the other is exactly the confusion this week's two filings, read together, expose.
A fourth thing is worth watching quietly rather than urgently: whether coverage of either filing starts to distinguish the two kinds of evidence on its own. That would spare the reader the work of checking a docket against a press release. The change would not be in the orbital compute sector itself, but in how confidently the public record gets read.
The orbital compute sector may yet build what it has filed for. Large infrastructure often takes a decade to move from application to operation. What a reader, a board or a regulator can reasonably ask for in the meantime is a plain distinction between the paperwork that exists today and the flight data that does not yet exist for anyone to check.
The same discipline that separates a filing from a flight applies to how a satellite's position gets verified, and the open-source layer behind that verification is worth naming directly. Orekit, a free space flight dynamics library released under the Apache License in 2008 by the French firm CS Group, is not a hobbyist tool. The Centre National d'Études Spatiales adopted it as the basis for its next-generation flight dynamics systems. It has tracked real rendezvous operations between Europe's Automated Transfer Vehicle and the International Space Station [10]. Nobody has to take a single company's telemetry on trust to check whether a claimed orbit is real. The propagation code doing the checking is published, inspectable and reused across agencies. That openness does not yet exist for the harder claim: whether a satellite actually computes what its filing says it will.
That gap between what is filed and what is flown is not only a market question. It is why NASA's formal objection to Project Sunrise matters more than a single regulatory comment. A United States government agency used the Federal Communications Commission's own public filing process to contest a private company's claim on shared orbital space, citing collision risk to crewed spaceflight and the absence of a debris mitigation plan, rather than settling the question by fiat. The Commission has not ruled. New Zealand's own Ground-Based Space Infrastructure regime works a different layer entirely, ground-segment authorisation rather than orbital-slot allocation, so the two are not directly comparable. What both share is the same underlying test: a claim on orbital capacity is not a demonstrated capability. The process for adjudicating that difference is still being written in public, one filing at a time.
So the question worth putting to any organisation watching this sector, in New Zealand or anywhere else: if the vendor in front of you cannot show a measured result, only a filed one, what exactly are you being asked to believe?
This is one of seven weekly series in The Hamberger Report. Subscribe on LinkedIn and the next one arrives in your feed.
• • •
The views expressed in this article are entirely my own, informed by more than 30 years of professional experience in architecture, security, and technology leadership in New Zealand. I write as director of Te Pono Limited; the views are personal and do not represent the position of any client, any government agency, or the New Zealand government. My commentary on legislation and policy is analytical, drawing on publicly available sources and my professional expertise in architecture, security, and AI governance, and it is politically neutral.
• • •
Andreas Hamberger is a New Zealand leader in Architecture & Security and Associate Member of the Institute of Directors. Space Mafia examines the sovereignty implications of orbital compute infrastructure.
• • •
This article was produced with AI assistance under my direction. Research, drafting and images pass through a pipeline I built and govern: automated gates for source verification, forbidden language and political neutrality, and my own review before anything is published. The tools include Claude, Gemini and Openart. The frameworks, arguments and editorial judgements are mine and are the same discipline I apply to the AI systems I audit for clients. AI accelerated the work; the thinking, and the responsibility for it, are mine.
• • •
[1] Broadband Breakfast. "Orbital Seeks FCC Approval for 100,000-Satellite Data Center Constellation." 2026. https://broadbandbreakfast.com/orbital-seeks-fcc-approval-for-100-000-satellite-data-center-constellation/
[2] GeekWire. "Blue Origin Joins the Data Center Space Race With Project Sunrise." 2026. https://www.geekwire.com/2026/blue-origin-data-center-space-race-project-sunrise/
[3] Orbital, Inc. "Orbital Files for FCC Approval of 100,000-Satellite Compute Constellation." 2026. https://orbital.inc/press/orbital-100000-constellation.html
[4] SatNews. "NASA Objects to Blue Origin's Project Sunrise." 5 May 2026. https://satnews.com/2026/05/05/nasa-objects-to-blue-origins-project-sunrise/
[5] Rocket Lab. "Mission Success: Rocket Lab Launches 96th Electron Mission." 19 September 2026. https://investors.rocketlabcorp.com/news-releases/news-release-details/mission-success-rocket-lab-launches-96th-electron-mission
[6] Space.com. "Rocket Lab Launches Electron 'Owl By The Dozen' Mission." 19 September 2026. https://www.space.com/space-exploration/launches-spacecraft/rocket-lab-electron-owl-by-the-dozen-launch
[7] Jonathan McDowell. "Space Report: Active Satellites." Planet4589, 17 September 2026. https://planet4589.org/space/stats/active.html
[8] DLA Piper. "New Zealand's Regulatory Regime Is Preparing for Liftoff." August 2025. https://www.dlapiper.com/en/insights/publications/2025/08/new-zealands-regulatory-regime-is-preparing-for-liftoff
[9] Bill Bennett. "Submarine Cables in New Zealand." billbennett.co.nz. https://billbennett.co.nz/submarine-cables-in-new-zealand/
[10] Orekit. "Orekit: An Open-Source Space Flight Dynamics Library." CS Group. https://www.orekit.org/