Starcloud is an American startup developing orbital data centers, satellites that run commercial computing hardware in space. Its first satellite, Starcloud-1, carried the first NVIDIA H100 GPU to orbit on November 2, 2025.[3][4] The company had raised $450 million by August 2026, when a Series A extension valued it at $2.3 billion,[34][36] and it has asked the United States Federal Communications Commission to authorize up to 88,000 satellites.[11]
Philip Johnston, Ezra Feilden, and Adi Oltean founded the company in January 2024 as Lumen Orbit. It is based in Redmond, Washington, with Johnston as chief executive, Feilden as chief technology officer, and Oltean as chief engineer.[1][2] Starcloud-1 deployed from a SpaceX Falcon 9, and both NVIDIA and the launch record confirm the H100 milestone.[3][4] Starcloud later reported that the satellite ran Google's Gemma language model and trained NanoGPT onboard.[5][6] That supports a narrower claim about the first reported training of a language model in space. It was not the first training of any artificial-intelligence or machine-learning model in orbit: a University of Oxford-led team trained a small cloud classifier aboard D-Orbit's ION SCV004 spacecraft in autumn 2022.[7][8] AI Wiki's Starcloud article covers the company's computing payloads, business model and partners.
Everything beyond Starcloud-1 remains developmental. Starcloud plans to launch two Starcloud-2 satellites as rideshare payloads during 2027, and Johnston said in October 2026 that Starcloud-2 would launch in January 2027;[35][40][48] Starcloud-3 depends on commercial Starship service,[10] and the proposed 88,000-satellite network had not received FCC authorization as of November 14, 2026.[11][12] In November 2026 Rocket Lab announced that Starcloud had booked a dedicated launch on its Neutron rocket for 2029, without saying which spacecraft it would carry.[51] The distinctions among flown hardware, company-reported results, contracts, and projections are central to assessing the program.
Status as of November 2026
Starcloud had one satellite in space on November 1, 2026: Starcloud-1, launched November 2, 2025, was tracked in an orbit of about 500 km on October 27.[3][44] Manhattan West led a $250 million extension of the company's Series A, announced August 21, which valued Starcloud at $2.3 billion post-money and took total funding to $450 million. First-time backers included Cisco Investments and NVIDIA.[34][36] In an interview published October 31, Johnston said Starcloud-2 had just been through vibration testing and would launch in January 2027, and that a second space on the company's new manufacturing campus would open in November.[48] On September 30, Firefly Aerospace said Starcloud's SC-1L computing payload would fly on its Elytra lunar orbiter, on a mission targeted for no earlier than 2028.[37][38] On November 9, Rocket Lab said Starcloud had booked a dedicated Neutron launch for 2029.[51] The 88,000-satellite FCC application was pending review on November 14; in a letter dated October 20, Starcloud asked the FCC to apply its newly adopted Part 100 surety-bond and milestone rules to the application and grant it without a bond.[12][49]
Company history and financing
Starcloud's founders came from consulting, satellite engineering and cloud computing. Johnston had advised national space agencies on satellite projects at McKinsey & Company; Feilden spent about a decade designing satellites, with a focus on deployable solar arrays and large deployable structures, at Airbus Defence and Space's SSTL and at Oxford Space Systems; and Oltean was a principal software engineer at SpaceX, responsible for the "tracking beams" that give other spacecraft Starlink connectivity, after 20 years at Microsoft working on large production GPU clusters.[1]
Starcloud joined Y Combinator's Summer 2024 batch. Before the accelerator, the company had raised a $2.4 million pre-seed round led by Nebular. TechCrunch reported an oversubscribed seed round of more than $10 million in October 2024.[2] In February 2025, the company adopted the Starcloud name after Lumen Technologies asserted rights to "Lumen" for data-center services. At the same time, Starcloud disclosed another $10 million and described the two seed financings as a combined $21 million round.[13]
NVIDIA's early role is sometimes overstated. GeekWire reported in February 2025 that NVIDIA supplied discounted chips through NVIDIA Inception, its startup program.[13] Participation in that program is not by itself an equity investment. NVIDIA later identified Starcloud as one of the companies using its accelerated-computing platforms for space missions, but neither announcement established that NVIDIA had invested in Starcloud.[4][14] An equity stake was first disclosed in August 2026, when Starcloud's financing release listed NVIDIA among the new investors in its Series A extension. TechCrunch put NVIDIA's contribution at $25 million, citing an unnamed source; the financing release did not give individual amounts.[34][35]
Benchmark and EQT Ventures led a $170 million Series A round that Starcloud announced on March 30, 2026. The corrected company release and independent reporting put the valuation at $1.1 billion and total capital raised at $200 million.[15][10] Those are dated financing figures, not revenue or deployed infrastructure. The release said the money would support Starcloud-3 development, a manufacturing facility, hiring, and future launch procurement.[15]
Starcloud enlarged the round on August 21, 2026. Manhattan West led a $250 million Series A extension. Returning investors included Benchmark, EQT, NFX, Soma, and 776, and those investing for the first time included Cedar Capital, Cisco Investments, Goanna Capital, NVIDIA, and Standard Capital. The company gave a post-money valuation of $2.3 billion, roughly double the March figure, and said its funding since 2024 now totaled $450 million. It named three uses for the money: more manufacturing capacity, joint engineering with NVIDIA, and purchases of future launch capacity. Lauren Selig of Manhattan West took a board observer seat.[34][36] The announcement also said Starcloud-3 production lines were being set up in a new 100,000-square-foot plant in Woodinville, Washington.[34] TechCrunch reported that the company had 25 employees at the time.[35]
In a video tour of the company's facilities published October 31, 2026, Johnston put the headcount at about 30, all but two of them engineers. He said Starcloud was building a manufacturing campus of three spaces, with the second due to come online in November and the third available from February 2027. The room was needed partly to test the mechanisms of Starcloud-3's deployable structures, which are more than 100 m long, and Johnston said the company aimed to build up to about 100 satellites a week there. He also said part of the company's cash was going to reserve launch slots, because launch was very constrained, especially from 2028 onward.[48]
Program status
Starcloud has flown one demonstration satellite; its larger spacecraft and the filed constellation remain designs, applications, or concepts.[3][11]
| Element | Latest public specification or scope | Evidence on November 1, 2026 |
|---|---|---|
| Starcloud-1 | About 60 kg; NVIDIA H100 payload; technology demonstrator[3][4] | Launched and deployed November 2, 2025; Gemma inference and NanoGPT training reported by the company; still tracked in orbit on October 27, 2026[3][5][6][44] |
| Starcloud-2 | About 450 kg and 8 kW; GPU cluster, storage, radiator, and two Starlink Mini Laser terminals[16][9] | In testing; in October Johnston said it had just completed vibration testing and gave January 2027 for the launch, the target SpaceNews reported in May; the August plan is two satellites on 2027 rideshare flights; no flight result[9][17][35][40][48] |
| Starcloud-3 | About 3 metric tons and 200 kW in March and May reporting; FCC debris logs filed in September model 2,000 kg, and chief technology officer Ezra Feilden put it at about two tons in October; designed around Starship deployment, up to 50 per launch[10][9][41][48] | In development, with a structural manufacturing demonstrator built and production lines being set up in Woodinville; company expectation of mid- to late 2028 depends on Starship accepting customer payloads[10][9][34][48] |
| Filed constellation | Up to 88,000 satellites in five sun-synchronous shells from 600 to 850 km[11][42] | FCC application pending after Space Bureau information requests in April and August 2026; in October Starcloud asked to be held to the FCC's new Part 100 bond and milestone rules; no grant, deployment, or operating constellation[11][12][39][47][49] |
| Gigawatt-scale data center | Early white paper illustrated a 5 GW complex with an approximately 4 km by 4 km solar array[18] | Concept study and company projection, not flight hardware or an authorized configuration[18] |
Starcloud's dates and designs have changed. In December 2025, Starcloud described Starcloud-2 as an October 2026 launch and Starcloud-3 as a roughly 2 metric ton, 100 kW spacecraft.[6] By March 2026, the reported Starcloud-3 design had grown to about 3 metric tons and 200 kW.[10] In May, reporting on the Starlink laser order moved Starcloud-2 to January 2027 and put Starcloud-3 no earlier than the point when Starship could carry commercial payloads, which Johnston estimated as mid- to late 2028.[9] In August, TechCrunch described the near-term goal as two Starcloud-2 satellites on rideshare flights during 2027, without naming a month,[35] and Starcloud's September FCC filing likewise placed the first Starcloud-2 flights in 2027.[40] In an interview published October 31, Johnston said Starcloud-2 would launch in January 2027.[48] The mass figures also differ by source: the debris-assessment logs Starcloud filed in September model each spacecraft, labeled Starcloud-3, at 2,000 kg, against the roughly 3 metric tons reported in March and May.[41][10][9] In the October interview, Feilden described Starcloud-3 as about two tons fully loaded, in line with the FCC logs.[48] The current figures are therefore a dated plan, not durable specifications.
Starcloud-1
Starcloud-1 flew on SpaceX's Bandwagon-4 mission, which lifted off from Cape Canaveral Space Force Station at 1:09 a.m. Eastern Time on November 2, 2025. The mission carried 18 payloads, and SpaceX's timeline records Starcloud-1 deployment 1 hour, 13 minutes, and 38 seconds after launch.[3] NVIDIA described the spacecraft before launch as a 60 kg demonstrator carrying an H100, the first flight of that data-center GPU.[4]
Public technical descriptions of the bus and orbit are less authoritative. Gunter's Space Page identifies an Astro Digital Corvus-Micro bus and separation into an approximately 325 km orbit.[19] The builder's own account is firmer on the bus: an Astro Digital investor presentation filed with the U.S. Securities and Exchange Commission in September 2026 says Starcloud-1 was based on its Corvus-Micro+ platform and took 1.5 years from initial design to delivery.[45] Tracking data do not support the 325 km figure, however. Jonathan McDowell's satellite catalogue lists Starcloud-1 in a 496 by 500 km orbit inclined 45.4 degrees as of October 27, 2026, still in orbit almost a year after launch.[44] Published mission-duration claims conflict: Gunter's lists an expected 11 months, while a November 2025 IEEE Spectrum report called it a three-year mission.[20][19] Starcloud has not published a mission factsheet that resolves that discrepancy, so a single planned duration should not be treated as established. TechCrunch described Starcloud in August 2026 as still operating the H100 in orbit.[35] Johnston said in October 2026 that the Starcloud-1 mission had cost about $2.5 million.[48]
The two clearly described computing results are limited but real engineering demonstrations. Starcloud says it ran Gemma for inference and trained NanoGPT on a Shakespeare corpus in December 2025.[5][6] GeekWire separately reported the demonstrations after speaking with Johnston, but no peer-reviewed paper, raw telemetry, power profile, fault log, model checkpoint, or independently reproduced result has been published.[6] The correct wording is therefore "company-reported first LLM training in orbit," not "first AI model trained in orbit."
The distinction matters because earlier onboard training is documented. Oxford researchers uploaded code to ION SCV004 in 2022 and trained the classifier portion of their RaVAEn cloud-detection model onboard. Oxford's account says the training used more than 1,300 images and took about 1.5 seconds.[7] The associated 2023 paper is an arXiv preprint that was presented at IGARSS, not a peer-reviewed journal article.[8]
Starcloud also proposed using its H100 to process synthetic-aperture-radar imagery from Capella Space and return compact insights instead of raw imagery. In December 2025, Johnston described that work as something the satellite would begin in 2026.[6] TechCrunch's March 2026 report said the satellite analyzes data from Capella's radar spacecraft,[10] but the public sources cited here do not document a completed Capella result or its output by November 1, 2026. In the October 2026 interview, Johnston said Starcloud-1's processors were running test inference on synthetic-aperture-radar and hyperspectral imagery for government customers, with results still to be announced.[48] TechCrunch also reported Johnston's statement that a separate NVIDIA A6000 aboard the mission failed during launch, but no engineering failure report has been released.[10]
Starcloud-2 and commercial partners
Starcloud calls Starcloud-2 its first commercial mission. Its product page promises a GPU cluster, persistent storage, continuous access, and proprietary power and thermal systems, with full operation in sun-synchronous orbit by 2027.[16] In May 2026, SpaceNews reported a design mass of about 450 kg, roughly 8 kW of generation, and a Falcon 9 launch target of January 2027.[9] None of those values has been demonstrated in flight. By August, the company's stated priority was getting two of the 8 kW satellites up on 2027 rideshare missions, where they would do inference work for customers such as U.S. government agencies. It was also weighing the purchase of an entire Falcon 9 for more spacecraft and signing agreements with other launch providers.[35]
A video tour of the company's Washington facility published on October 31, 2026, showed Starcloud-2 in assembly and testing. In it, Johnston said the satellite had just completed vibration testing, which the company ran in-house, and that it would launch in January 2027, a flight he described as a proof point for its radiators and the other systems Starcloud had designed. Feilden said all of Starcloud's upcoming launches were with SpaceX's Falcon rockets.[48] GeekWire had reported in December 2025 that Starcloud-2 would have about 100 times Starcloud-1's power-generating capability.[6] According to the company, the satellite carries two laser terminals, deploys wings spanning about 37 m (120 ft) that include radiator segments designed and built in-house, uses liquid cooling, and has a xenon-fueled electric propulsion system.[48] Johnston said Starcloud tests its chips for radiation tolerance at particle accelerators and ran Starcloud-2's chips at full power for three months on the ground to find early-life failures before launch. He said the testing mattered because the economics of the business would fall apart if the chips failed at a rate 10 or 20 percent higher in space than on the ground, the chips being by far the most expensive part of the system.[48]
Crusoe announced a signed partnership in October 2025 to place Crusoe Cloud on Starcloud-2 and offer limited orbital GPU capacity in 2027.[21] Starcloud's March 2026 financing release also named AWS, Google Cloud, and NVIDIA as partners and described planned Blackwell and AWS hardware.[15] NVIDIA independently confirms that Starcloud is using NVIDIA platforms.[14] The cited public materials from AWS and Google do not define their commitments, so the relationship should not be described as a confirmed cloud-service deployment by either company.
Connectivity has a more specific contractual record. On May 26, 2026, Starcloud said it had signed a contract for more than 50 of SpaceX's Starlink Mini Laser terminals, to be spread over at least 25 future satellites. It said each spacecraft would carry two terminals, with up to 25 Gbit/s per terminal at distances up to 4,000 km.[17] That is a supplier announcement, not an operating Starcloud mesh. The first hardware was expected in orbit within one year of the announcement, beginning with Starcloud-2.[17][9] Starcloud's September 2026 FCC response described the Starlink order as covering only the first tranche of the filed constellation, with further optical-relay agreements still to be signed before later deployments.[40]
Starcloud's newest announced agreement is for a payload in lunar orbit. On September 30, 2026, Firefly Aerospace announced that Starcloud's SC-1L computing payload would fly aboard an Elytra orbiter on Firefly's third mission to the Moon, targeted for launch no earlier than 2028. Elytra is first to deliver a Blue Ghost lander to the Gruithuisen Domes under NASA's Commercial Lunar Payload Services program and relay its data. The Starcloud demonstration begins after the lander mission ends: SC-1L is to take in data from Elytra's Solux vision system, process it in lunar orbit, and send the results to Earth.[37][38] Firefly expects the orbiter to remain in lunar orbit for five years. The lunar payload is a contracted demonstration, not a flight result.[37]
Starcloud-3 and scaling assumptions
Press descriptions of Starcloud-3 from March and May 2026 give a roughly 3 metric ton, 200 kW spacecraft shaped for Starship's satellite-deployment system.[10][9] Starcloud's September 2026 FCC filings add physical detail without settling the mass. The response describes the satellites in the 88,000-satellite application as having a gross footprint of about 6 by 220 m, most of it two deployable solar arrays, each about 5 by 107 m and 535 kg.[40] The attached debris-assessment logs model a 2,000 kg spacecraft labeled Starcloud-3, using 2028 as the start year.[41] In the October 2026 interview, Feilden described Starcloud-3 as about two tons fully loaded and sized to pass through the door of Starship's payload dispenser, with up to 50 fitting in one Starship, stacked two to a layer. He showed a manufacturing demonstrator built to give the team experience with a primary structure at Starcloud-3's scale. Johnston called it the 200-kilowatt satellite and said 50 per Starship would add close to 10 MW of new computing capacity per launch.[48] Starcloud has not published a detailed Starcloud-3 specification sheet, environmental qualification record, launch contract, or fixed flight date. Johnston's mid- to late 2028 estimate was explicitly tied to when Starship might be ready for customer payloads.[9]
Starcloud also intends to fly the NVIDIA Space-1 Vera Rubin Module on future spacecraft,[34][36] a computing module for use in orbit that NVIDIA announced in March 2026 and said would be available at a later date.[14] NVIDIA rates the module's Rubin GPU at up to 25 times the AI compute of an H100 for inference in space, a manufacturer claim rather than a measured flight result.[14] TechCrunch reported that Starcloud hoped to put the module in orbit in late 2028, although the chip had not yet been built.[35] Johnston told TechCrunch that launch capacity was tight and that Starcloud wanted Starship launch contracts as soon as it could get them.[35]
The economics depend on that launch system. Johnston told TechCrunch that Starcloud-3 could approach a power cost near $0.05 per kilowatt-hour if commercial launch prices reached about $500 per kilogram. He also acknowledged that Starcloud would not be competitive on energy cost until Starship flew frequently.[10] Johnston gave the same threshold in October 2026, saying that below about $500 per kilogram computing in orbit would be much cheaper than on the ground, against roughly $5,000 per kilogram on Falcon 9. In his comparison, orbit removes the land, permitting and battery-storage costs of a solar-powered data center on Earth and needs about one-eighth of the solar-cell area, leaving launch as the main added cost.[48] Starcloud's September 2026 FCC response nonetheless gives each satellite a battery subsystem and expects it to be in sunlight for about 97 percent of its operational life.[40] Johnston said the company's current business, processing data for other satellite operators, was already economic at current launch prices, and that Starcloud hoped SpaceX would sell Starship capacity to commercial customers from 2028 and had booked launches on other vehicles in case it did not.[48] Neither the launch price, flight rate, nor resulting compute price has been achieved.
Starcloud's public concept has also shifted. Its September 2024 white paper centered on training and illustrated a modular 5 GW complex with a solar array about 4 km by 4 km. Its cost table assumed a $5 million launch carrying a 40 MW cluster, very low solar-cell cost, a ten-year service period, and no backup power.[18] In a May 2026 interview, Johnston instead emphasized distributed inference satellites and described the filed constellation as potentially delivering on the order of 20 GW.[22] The August 2026 financing release repeated that goal of 20 GW from 88,000 satellites.[34] Johnston called those 88,000 satellites a first step in October 2026 and estimated that dawn-dusk sun-synchronous orbit could hold about 10 TW of computing.[48] These figures are company scenarios. They are not the output of an authorized system, and the FCC filing allows different computing payloads rather than committing every satellite to one power level.[23]
FCC application
Starcloud filed application SAT-LOA-20260202-00073, call sign S00803, with the Federal Communications Commission on February 4, 2026. The FCC accepted it for filing on March 13.[11] "Accepted for filing" means that the agency opened the application to review and public comment; it is not an authorization. The ICFS record still listed the application as "Pending Review" on November 1, 2026, with no grant date and no recorded action.[12]
The FCC's Space Bureau has since asked twice for more information. Its April 29, 2026 letter requested, among other items, the nominal shell altitudes, the number of planes, and missing debris-mitigation data.[47] Starcloud's May 29 reply named five shells, each a single orbital plane of up to 17,600 satellites, with nominal altitudes of 620, 660, 710, 740, and 830 km, and said each satellite would hold back 50 kg of propellant for disposal.[42] A second letter on August 28 asked how many satellites would use the backup Ka-band links at one time, whether Starcloud had contracts for optical relay service, and for more debris data, including updated Debris Assessment Software runs that counted the battery system.[39] In a response dated September 16, Starcloud estimated that no more than 200 of its satellites would transmit in Ka-band at once, apart from a narrowly defined emergency, and expected zero in normal operations.[40] Both letters warned that failing to respond could lead to dismissal. They are requests for information, not decisions on the application.[39][47]
Starcloud has since asked to be treated under the FCC's new licensing rules. On July 22, 2026, the Commission adopted an order replacing its Part 25 satellite licensing rules with a new Part 100; on October 7 the Space Bureau said the new rules were not yet in effect and that it would announce their effective date later. Under Part 100, systems licensed outside a processing round post no surety bond; they must launch and operate at least one satellite for 90 continuous days within seven years of authorization, then have 10 percent of the authorized satellites operating within nine years, half within 12 and all within 14.[50] In a letter dated October 20, 2026, Starcloud asked the FCC to apply those bond and milestone rules to its pending application ahead of the general effective date, citing the Space Bureau's October 7 transition guidance, and to grant the application without a bond, noting that it was not seeking processing-round status. It withdrew its requests to waive the Part 25 milestone and bond rules on condition that the Part 100 rules are applied, kept them in place otherwise, and asked for a grant at the earliest permissible opportunity.[49]
The application requests authority for up to 88,000 satellites operating as a distributed data center. Its main parameters are:
| Parameter | What the filing says | Evidentiary limit |
|---|---|---|
| Orbits | Sun-synchronous inclinations, nominal 06:00 crossing, between 600 and 850 km | The narrative describes narrow shells up to 50 km wide; a May 2026 supplement lists five single-plane shells of up to 17,600 satellites each but gives no deployment schedule[11][23][42] |
| Spacecraft commonality | Same size, mass, maneuverability, communications, and control systems | Computing hardware may vary; a September 2026 response gives a gross footprint of about 6 by 220 m and its debris logs use 2,000 kg, but no compute rating is given for each satellite[23][40][41] |
| Primary communications | Optical links routed through third-party satellite backhaul networks | The narrative names Starlink, Amazon Leo, TeraWave, and Kepler subject to commercial arrangements; only the Starlink terminal purchase is separately documented here[17][23] |
| Backup radio | 18.8-19.3 GHz space-to-Earth and 28.6-29.1 GHz Earth-to-space | Requested only for telemetry, tracking, and command on a non-interference, unprotected basis[11][23] |
| Operational lifetime | Five years from launch | Schedule S figure, not an observed fleet lifetime[24] |
| Disposal | Active orbit lowering followed by atmospheric reentry | Applicant design representation; full-scale performance is not demonstrated[25] |
Schedule S appears to list two satellites in two orbital planes at 600 and 850 km. Starcloud's waiver request explains that these are representative boundary cases used because the form could not encode the full range of orbital planes and beam configurations.[24][26] They should not be read as replacing the narrative request for up to 88,000 spacecraft. Starcloud also seeks waivers from the FCC's ordinary processing-round, buildout-milestone, and surety-bond rules, so the filing does not establish a binding rollout schedule.[11][26] Its October 2026 letter would drop the milestone and bond waivers if the FCC applies the Part 100 rules instead.[49]
The orbital-debris assessment describes triple-redundant propulsion, automated collision avoidance, lower-altitude checkout, and active end-of-life lowering. It says a failed spacecraft at the worst-case 850 km altitude would passively decay in about 16.2 years, according to the applicant's model.[25] The same document calls for uncontrolled atmospheric reentry after active lowering, with controlled reentry attempted but not relied upon. It says the satellites will be designed for demisability and that a final assessment will be conducted, rather than presenting a completed fleet-wide demise demonstration.[25] Starcloud's September 2026 response confirmed uncontrolled reentry as the disposal method it seeks a license for. Each satellite would use its remaining propellant to descend below 390 km, after which drag would typically bring it down in less than 30 days, according to the company. For a five-year mission at 850 km, the updated debris runs put the combined probability of small debris penetrating a satellite's critical systems at about 2.6 in a million, below the 0.01 limit in FCC rules.[40][41]
The scale drew formal objections. The American Astronomical Society asked the FCC to deny the application in its current form or delay action until Starcloud supplied independently verified aggregate analyses of optical, infrared, radio, atmospheric, and debris effects.[27] Secure World Foundation did not oppose orbital computing as a category, but argued that a constellation of this size needed coordinated, system-level review across agencies.[28] On July 8, 2026, DarkSky International, Environment America, and Public Employees for Environmental Responsibility petitioned the FCC to prepare a programmatic environmental impact statement covering the orbital data center applications, Starcloud's among them, and to pause licensing until it was finished.[43] After Starcloud's September answers, the Center for Space Environmentalism asked on October 5 that the application be dismissed or treated as a major amendment.[46] These are comments in a pending proceeding, not FCC findings.
Technical and environmental evidence
Starcloud's near-term edge-computing case is narrower than its terrestrial-cloud case. Processing imagery beside the sensor can reduce a large raw dataset to a small result, avoiding a downlink bottleneck.[6] Sending workloads from Earth to orbit and returning their outputs adds network, ground-station, storage, security, and availability requirements.[29] The broader engineering trade-offs are covered in the orbital data centers article.
Heat rejection is the most visible scaling constraint. Vacuum prevents convective cooling, so almost all electrical power used by the computers must ultimately leave through radiators as thermal radiation. An independent IEEE Spectrum analysis estimated about 1.4 square meters of radiator for one 700 W H100 held near 60 C under idealized conditions, about 80 square meters for a 40 kW rack, and a roughly 40 percent area increase after modeled coating degradation. Its rough cost model still put a GPU-year in orbit at least ten times above a terrestrial equivalent even while assuming an optimistic $44 per kilogram Starship launch cost.[29] Those are analyst calculations, not peer-reviewed measurements or Starcloud flight data, but they show why a working H100 does not validate megawatt-scale cooling or economics.
Academic work reaches conditional, not universal, conclusions. A 2025 peer-reviewed Nature Electronics perspective presents orbital edge and cloud data centers as technically plausible frameworks and supplies a lifecycle-carbon method.[30] A 2025 HotCarbon workshop paper modeled launch, operation, and reentry and found that its orbital cases produced up to an order of magnitude more carbon emissions than terrestrial computing, mainly through embodied launch and reentry costs.[31] The different results depend on system lifetime, launch vehicle, electricity mix, utilization, hardware replacement, and what emissions are counted. Starcloud's claim of tenfold carbon savings remains a projection rather than a measured outcome.[4]
Astronomy and atmospheric effects also depend on hardware that has not yet been specified in enough detail. A peer-reviewed 2026 MNRAS paper analyzed the early 5 GW, 4 km-array concept and estimated that such a structure could span about 0.4 degrees and reach magnitude -5 to -7 during twilight.[32] That model does not predict the brightness of Starcloud-1 or prove that every satellite in the FCC filing would have that geometry. The satellites described in Starcloud's September 2026 FCC response are far smaller, with a gross footprint of about 6 by 220 m.[40] A separate peer-reviewed Journal of Geophysical Research study modeled a generic 10 gigagram-per-year aluminum-oxide reentry scenario and found potentially significant middle-atmosphere changes, while emphasizing uncertainty in actual reentry aerosols.[33] It is relevant to cumulative megaconstellation review, not a forecast of Starcloud's emissions without satellite mass, materials, launch cadence, and measured ablation products.
What is demonstrated and what is not
As of November 1, 2026, Starcloud's demonstrated record is one deployed spacecraft carrying an H100 and company-reported execution of two small language-model workloads. Its August 2026 financing release added a claim of inference and fine-tuning runs on the flight hardware without giving details.[34] The result establishes that a modern terrestrial GPU can survive launch, power on, and execute selected code in orbit. It does not establish continuous commercial availability, long-duration radiation tolerance, useful customer throughput, a profitable service, a large deployable radiator, inter-satellite cloud networking, or cost and carbon advantages over terrestrial data centers.[3][5][6][11]
The next meaningful evidence would be Starcloud-2 telemetry: sustained input power, radiator temperature and degradation, corrected and uncorrected compute errors, storage integrity, link availability, completed customer workloads, and the cost of delivering them.[16][9][29] Until those measurements exist, Starcloud-2, Starcloud-3, the 20 GW figure, and the 88,000-satellite system remain plans at very different levels of technical and regulatory maturity.[9][10][11][12][22][23]
References
- Starcloud team - Starcloud, accessed November 1, 2026.
- Lumen Orbit closed one of the biggest rounds from Y Combinator's last cohort - TechCrunch, October 24, 2024.
- Bandwagon-4 mission - SpaceX, November 2, 2025.
- How Starcloud Is Bringing Data Centers to Outer Space - NVIDIA, October 15, 2025.
- Starcloud-1 - Starcloud, accessed August 12, 2026.
- Starcloud plans its next power plays after training first AI model in space - GeekWire, December 22, 2025.
- Researchers successfully train a machine learning model in outer space for the first time - University of Oxford, July 28, 2023.
- Fast model inference and training on-board of satellites - arXiv preprint and IGARSS 2023 conference paper, July 17, 2023.
- Starcloud orders Starlink lasers for orbital data center network - SpaceNews, May 26, 2026.
- Starcloud raises $170 million Series A to build data centers in space - TechCrunch, March 30, 2026.
- Applications accepted for filing, Report SAT-01982 - Federal Communications Commission, March 13, 2026.
- ICFS application summary for SAT-LOA-20260202-00073 - Federal Communications Commission ICFS portal, accessed November 14, 2026.
- Lumen Orbit changes its name to Starcloud and raises $10M for space data centers - GeekWire, February 26, 2025.
- NVIDIA launches space computing - NVIDIA Newsroom, March 16, 2026.
- Correcting and replacing: Starcloud raises $170M Series A at $1.1bn valuation - Starcloud press release via Business Wire, March 30, 2026.
- Starcloud-2 - Starcloud, accessed November 1, 2026.
- Starcloud to integrate SpaceX's Starlink Mini Lasers into its orbital data center constellation - Starcloud press release via Business Wire, May 26, 2026.
- Why we should train AI in space - Lumen Orbit, now Starcloud, white paper version 1.03, September 2024.
- Starcloud 1 (Lumen 1) - Gunter's Space Page, updated August 24, 2026.
- NVIDIA sends a powerful GPU to space - IEEE Spectrum, November 3, 2025.
- Crusoe to become first cloud operator in space through partnership with Starcloud - Crusoe, October 22, 2025.
- The case for data centers in space - McKinsey & Company interview with Philip Johnston, May 8, 2026.
- Application for launch and operating authority for the Starcloud Orbital Datacenter System - Starcloud filing with the Federal Communications Commission, February 4, 2026.
- FCC Form 312 Schedule S for the Starcloud Orbital Datacenter System - Starcloud filing with the Federal Communications Commission, February 2026.
- Starcloud Orbital Debris Assessment Report - Starcloud filing with the Federal Communications Commission, February 2026.
- Starcloud waiver requests - Starcloud filing with the Federal Communications Commission, February 2026.
- AAS comments on Starcloud FCC application to operate 88,000 orbital data centers - American Astronomical Society, April 13, 2026.
- Comments on Starcloud's orbital data center application - Secure World Foundation, April 10, 2026.
- Why orbital data centers are harder than Silicon Valley thinks - IEEE Spectrum, June 11, 2026.
- The development of carbon-neutral data centres in space - Nature Electronics, peer-reviewed perspective, October 27, 2025.
- Dirty Bits in Low-Earth Orbit: The Carbon Footprint of Launching Computers - HotCarbon 2025 workshop paper and ACM SIGENERGY Energy Informatics Review, July 10, 2025.
- The impact on astronomy of data centres orbiting Earth - Monthly Notices of the Royal Astronomical Society, peer-reviewed article, May 1, 2026.
- Investigating the Potential Atmospheric Accumulation and Radiative Impact of the Coming Increase in Satellite Reentry Frequency - Journal of Geophysical Research: Atmospheres, peer-reviewed article, 2025.
- Starcloud raises $250 million at $2.3 billion valuation to scale AI with orbital data centers - Starcloud press release via Business Wire, August 21, 2026.
- Starcloud raises $250 million for orbital data centers as launch options dry up - TechCrunch, August 21, 2026.
- Starcloud raises $250M to support the creation of data center satellite network in league with Nvidia - GeekWire, August 21, 2026.
- Firefly Aerospace signs Starcloud as commercial customer, demonstrating data center capabilities around the Moon - Firefly Aerospace, September 30, 2026.
- Starcloud's AI computing payload will get a ride to lunar orbit on Firefly's space vehicle - GeekWire, September 30, 2026.
- Space Bureau information request of August 28, 2026, ICFS File No. SAT-LOA-20260202-00073 - Federal Communications Commission, August 28, 2026.
- Starcloud response to the Space Bureau's August 28, 2026 information request - Starcloud filing with the Federal Communications Commission, September 16, 2026.
- Updated Debris Assessment Software logs for Starcloud-3 - Starcloud filing with the Federal Communications Commission, September 16, 2026.
- Supplemental response to April 29, 2026 Space Bureau information request - Starcloud filing with the Federal Communications Commission, May 29, 2026.
- Petition to initiate programmatic environmental impact statement on orbital data center applications - DarkSky International, Environment America, and Public Employees for Environmental Responsibility filing with the Federal Communications Commission, July 8, 2026.
- GCAT current catalog - Jonathan McDowell, General Catalog of Artificial Space Objects, updated October 29, 2026.
- Space infrastructure from concept to constellation: investor presentation - Astro Digital, exhibit 99.2 (dated September 25, 2026) to a Proem Acquisition Corp. I Form 425 filed with the U.S. Securities and Exchange Commission, September 28, 2026.
- Response to ex parte presentation and supplemental comments - Center for Space Environmentalism filing with the Federal Communications Commission, October 5, 2026.
- Space Bureau information request of April 29, 2026, ICFS File No. SAT-LOA-20260202-00073 - Federal Communications Commission, April 29, 2026.
- Nvidia GPUs in Orbit: Inside Starcloud's $2B Factory - Sourcery with Molly O'Shea (YouTube), October 31, 2026.
- Part-100 Applicability Supplement, ICFS File No. SAT-LOA-20260202-00073 - Starcloud filing with the Federal Communications Commission, October 20, 2026.
- Guidance on Transition of Part 25 Milestone and Surety Bond Rules and Requirements to Part 100, DA 26-1080 - Federal Communications Commission Space Bureau, October 7, 2026.
- Rocket Lab Lands Over $515M in Neutron Launch Deals from Constellation Operators in Q3 - Rocket Lab, November 9, 2026.
