Difference between revisions of "Computer Networks"

From
Jump to: navigation, search
m (→‎Space-based Data Center)
m (→‎Space-based Data Centers)
 
(11 intermediate revisions by the same user not shown)
Line 244: Line 244:
  
 
= Space-based Data Centers =
 
= Space-based Data Centers =
−
* [https://www.huawei.com/en/giv/data-center-2030 Huawei — Data Center 2030]
+
* [[Satellite]]
−
* [https://research.google/blog/exploring-a-space-based-scalable-ai-infrastructure-system-design/ Google Research — Project Suncatcher]
+
* [[Astonomy]]
 +
* [https://research.google/blog/exploring-a-space-based-scalable-ai-infrastructure-system-design/ Google Research — Project Suncatcher] ... [[Google]]
 
* [https://www.axiomspace.com/orbital-data-center Axiom Space — Orbital Data Centers]
 
* [https://www.axiomspace.com/orbital-data-center Axiom Space — Orbital Data Centers]
 
* [https://ascend-horizon.eu/ ASCEND (Horizon Europe)]
 
* [https://ascend-horizon.eu/ ASCEND (Horizon Europe)]
 
* [https://kubeedge.io/case-studies/satellite/ KubeEdge — Cloud Native Edge Computing Satellite case study]
 
* [https://kubeedge.io/case-studies/satellite/ KubeEdge — Cloud Native Edge Computing Satellite case study]
−
* [https://nvidianews.nvidia.com/news/space-computing Nvidia — Space computing platforms]
+
* [https://nvidianews.nvidia.com/news/space-computing NVIDIA — Space computing platforms] ... [[NVIDIA]]
 +
* [https://www.huawei.com/en/giv/data-center-2030 Huawei — Data Center 2030]
 
* [https://space-compass.com/en/ Space Compass Corporation]
 
* [https://space-compass.com/en/ Space Compass Corporation]
 
* [https://spacenews.com/orbital-data-centers/ SpaceNews — Orbital Data Centers coverage]
 
* [https://spacenews.com/orbital-data-centers/ SpaceNews — Orbital Data Centers coverage]
 
* [https://www.thalesaleniaspace.com/en/press-releases/thales-alenia-space-reveals-results-ascend-feasibility-study-space-data-centers-0 Thales Alenia Space — ASCEND feasibility results]
 
* [https://www.thalesaleniaspace.com/en/press-releases/thales-alenia-space-reveals-results-ascend-feasibility-study-space-data-centers-0 Thales Alenia Space — ASCEND feasibility results]
 +
* [https://www.linkedin.com/pulse/space-based-data-centers-computings-final-frontier-david-linthicum-i3xie/ Space-Based Data Centers: Computing’s Final Frontier or Expensive Gimmick? | David Linthicum]
 +
 +
'''Space-based data centers''' (also '''orbital data centers''', '''ODCs''', or '''orbital compute''') are proposed or operational computing facilities deployed in [[low Earth orbit]] (LEO) or beyond, in which spacecraft perform data processing, storage, artificial-intelligence inference or training, or related computing functions rather than acting only as communications relays or sensors.
  
−
'''Space-based data centers''' (also '''orbital data centers''', ODCs, or '''orbital compute''') are computing facilities deployed in [[low Earth orbit]] (LEO) or beyond, in which spacecraft act as processing nodes rather than as passive sensors or communications relays. Interest in the concept accelerated sharply between 2024 and 2026 as terrestrial [[data center]] construction ran into grid interconnection queues, water constraints and permitting resistance, while the volume of data produced by Earth-observation and communications constellations began to exceed available downlink bandwidth.
+
Interest in orbital computing increased during the mid-2020s as demand for artificial-intelligence infrastructure grew, satellite operators sought alternatives to transmitting large volumes of raw data to Earth, and governments and companies investigated new approaches to energy supply, cooling, data sovereignty and distributed computing. The field includes aerospace companies, cloud and semiconductor firms, telecommunications operators, universities, government research programs and venture-backed startups.
  
−
[[Huawei]] was among the earliest large infrastructure vendors to name space data centers as a structural pattern for the coming decade, doing so in its ''Data Center 2030'' technology forecast. As of September 2026 the field spans state research programmes, hyperscalers, launch providers and a cohort of venture-funded startups, with regulatory filings on file for well over one million satellites in aggregate.
+
Approaches range from relatively small edge-computing payloads that process satellite data before downlink, to proposed constellations containing thousands or tens of thousands of dedicated computing spacecraft. As of September 2026, most large-scale orbital data-center concepts remained at the research, demonstration, regulatory-filing or early deployment stage.
  
 
== Rationale ==
 
== Rationale ==
  
−
Proponents advance four principal arguments.
+
Several arguments are commonly advanced for space-based computing.
  
 
; Power
 
; Power
−
: Solar arrays in sun-synchronous dawn–dusk orbits receive near-continuous, unattenuated sunlight, avoiding cloud cover, atmospheric losses and night. Estimates of the per-panel advantage over terrestrial installations range from roughly five-fold to eight-fold depending on the comparison baseline. Because such orbits avoid eclipse for most of the year, battery storage requirements are also reduced.
+
: Spacecraft in suitable orbits can make extensive use of solar power without atmospheric attenuation or terrestrial weather. Dawn–dusk sun-synchronous orbits can also reduce the duration of eclipse periods, potentially lowering energy-storage requirements. The practical advantage depends on orbital design, solar-array efficiency, spacecraft mass and other engineering constraints.
  
 
; Cooling
 
; Cooling
−
: Waste heat is rejected radiatively rather than by chilled air or water, eliminating the 10–30% of terrestrial facility power typically consumed by cooling plant, and eliminating water consumption entirely. Critics note that this converts an operating-cost problem into a mass-and-area problem (see [[#Criticism and open questions|below]]).
+
: Spacecraft do not require terrestrial chilled-water or air-conditioning systems, and therefore do not consume water for cooling. However, waste heat in vacuum must ultimately be rejected through thermal radiation. This requires radiator area and thermal-control hardware, making heat rejection one of the principal engineering constraints for high-power orbital computing.
 +
 
 +
; Data processing and downlink
 +
: Earth-observation and communications satellites can generate substantially more raw data than is practical to transmit continuously to the ground. Processing data on orbit allows spacecraft to return selected information such as detections, classifications, compressed products or alerts rather than transmitting all raw sensor data.
 +
 
 +
; Latency
 +
: Locating computing resources closer to satellites and other space-based sensors can reduce the time required to process information for applications such as disaster monitoring, maritime surveillance, astronomy and autonomous spacecraft operations.
 +
 
 +
; Sovereignty and resilience
 +
: Some government programs describe orbital computing as part of broader digital-sovereignty or infrastructure-resilience strategies. These proposals also raise unresolved questions concerning jurisdiction, data governance and responsibility for infrastructure operating outside national territory.
 +
 
 +
== Development approaches ==
 +
 
 +
Orbital-computing projects generally fall into several overlapping categories:
 +
 
 +
* '''Onboard edge computing''' — processing sensor data directly aboard an individual spacecraft.
 +
* '''Distributed satellite computing''' — connecting multiple spacecraft through optical or radio inter-satellite links so they can share computing tasks.
 +
* '''Dedicated orbital data-center nodes''' — spacecraft designed primarily to provide computing or storage capacity.
 +
* '''Large-scale orbital compute constellations''' — proposed fleets intended to provide data-center-scale or hyperscale computing capacity.
 +
* '''Enabling technologies''' — radiation-tolerant processors, optical networking, thermal-management systems, software orchestration and launch systems used by orbital-computing operators.
 +
 
 +
```mediawiki
 +
== Artificial intelligence and orbital computing ==
 +
 
 +
[[Artificial intelligence]] is a major proposed workload for space-based data centers, particularly for [[machine learning]] inference, Earth-observation analysis and, in some projects, model training. AI workloads are relevant to orbital computing because satellites can generate large volumes of sensor data that may be processed locally rather than transmitted in full to terrestrial data centers.
 +
 
 +
=== On-orbit inference ===
 +
 
 +
Near-term applications have focused primarily on '''AI inference''', in which previously trained models process imagery, sensor measurements or other data aboard a spacecraft.
 +
 
 +
Potential applications include:
 +
 
 +
* identifying objects or changes in Earth-observation imagery;
 +
* weather and environmental monitoring;
 +
* maritime and infrastructure surveillance;
 +
* disaster detection and response;
 +
* astronomical data processing;
 +
* autonomous spacecraft operations;
 +
* filtering or prioritizing data before transmission to Earth.
 +
 
 +
Processing data on orbit can reduce downlink requirements by transmitting classifications, detections or other derived results instead of complete raw datasets.
 +
 
 +
The '''Three-Body Computing Constellation''', developed by ADA Space and Zhejiang Lab, has reported operating multiple AI models in orbit, including large remote-sensing and astronomy models. Zhejiang Lab has also reported experiments in which tasks are distributed between multiple interconnected satellites.
 +
 
 +
The '''Tiansuan Constellation''', led by the Beijing University of Posts and Telecommunications with participation from several research and industry partners, has demonstrated collaborative inference between spacecraft and ground systems. Experiments associated with KubeEdge and related software reported reductions in the amount of data transmitted to Earth by performing preliminary processing aboard the satellite.
 +
 
 +
Other companies and research programs, including Axiom Space, [[NVIDIA]] partners and satellite edge-computing developers, have similarly investigated onboard AI inference for geospatial and autonomous applications.
 +
 
 +
=== Distributed AI computing ===
 +
 
 +
Some orbital-computing architectures propose linking multiple satellites through high-bandwidth optical inter-satellite links so that computing tasks can be distributed across several spacecraft.
 +
 
 +
This approach differs from conventional onboard processing, in which each satellite operates largely independently. A distributed orbital cluster could theoretically divide a large AI workload across multiple computing nodes in a manner conceptually similar to terrestrial distributed-computing systems.
 +
 
 +
Projects investigating this model include the Three-Body Computing Constellation, Google's '''Project Suncatcher''' and several proposed commercial orbital data-center constellations.
 +
 
 +
Such systems require:
 +
 
 +
* high-bandwidth and low-latency inter-satellite networking;
 +
* distributed workload scheduling;
 +
* synchronization between computing nodes;
 +
* fault tolerance for interrupted links or spacecraft failures;
 +
* efficient movement of model parameters and intermediate data;
 +
* coordinated thermal and power management.
 +
 
 +
The performance of distributed AI workloads in orbit remains an active area of research.
 +
 
 +
=== AI model training ===
 +
 
 +
'''AI training''' is substantially more computationally and energetically demanding than inference and is therefore a longer-term objective for most orbital data-center projects.
 +
 
 +
Starcloud reported conducting an AI model-training experiment aboard its Starcloud-1 satellite after launching an [[NVIDIA]] H100 GPU to low Earth orbit in November 2025. Several larger proposed orbital-computing systems have been designed around future GPU or accelerator clusters capable of supporting more computationally intensive workloads.
 +
 
 +
Google's Project Suncatcher is investigating whether clusters of satellites equipped with [[Tensor Processing Unit]]s could eventually operate distributed machine-learning workloads in orbit. Its research has examined processor radiation tolerance, optical interconnects and the launch-cost reductions that would be required for orbital systems to approach the economics of terrestrial AI infrastructure.
 +
 
 +
Other proposed systems from companies including SpaceX, Starcloud and Cowboy Space have described large-scale accelerator deployments, although most such architectures remained at the proposal or development stage as of 2026.
 +
 
 +
=== AI hardware ===
 +
 
 +
Artificial-intelligence workloads require processors capable of performing large numbers of parallel matrix and tensor operations.
 +
 
 +
Orbital AI projects have therefore investigated several hardware approaches:
 +
 
 +
* terrestrial data-center GPUs adapted for space operation;
 +
* purpose-built or modified AI accelerators;
 +
* radiation-tolerant edge-computing processors;
 +
* combinations of lightweight onboard processors and higher-performance ground systems.
 +
 
 +
'''[[NVIDIA]]''' has supplied or proposed hardware for several orbital-computing programs. An [[NVIDIA]] H100 was flown aboard Starcloud-1, while the company has announced the Space-1 Vera Rubin Module for future orbital data-center applications. [[NVIDIA]]'s Jetson and IGX platforms have also been associated with lower-power edge-AI applications in space.
  
−
; Downlink bottleneck
+
'''Google''' is investigating its own Tensor Processing Units for Project Suncatcher.
−
: This is the driver most often cited by operators rather than by AI-infrastructure investors. Satellites now generate more data than they can transmit; researchers at Zhejiang Lab have stated that as much as 90% of data generated on orbit is never effectively processed. Processing in orbit and downlinking only derived results — detections, classifications, alerts — collapses the bandwidth requirement. The commonly used analogy is editing video on the capture device rather than uploading raw footage.
 
  
−
; Latency and sovereignty
+
Chinese research programmes have used a mixture of commercial processors, domestic AI hardware and software frameworks including MindSpore. Huawei-associated technologies have participated primarily at the software and edge-orchestration layer through KubeEdge, Sedna and related research rather than through a dedicated publicly announced orbital AI accelerator.
−
: Placing compute adjacent to the point of data generation reduces round-trip time for time-critical applications (disaster response, maritime monitoring, defence tipping-and-cueing). Several programmes, notably in Europe and China, additionally frame orbital compute as a matter of digital sovereignty.
 
  
−
== Huawei ==
+
Other suppliers, including Ramon.Space, OrbitsEdge and Hewlett Packard Enterprise, are developing or adapting computing systems intended to operate reliably in the space radiation environment.
  
−
=== Data Center 2030 ===
+
=== Software and orchestration ===
  
−
Huawei released the ''Data Center 2030'' report at HUAWEI CONNECT 2023 in Shanghai on 20 September 2023. Presented by Michael Ma, then Vice President of Huawei and President of the ICT Product Portfolio Management & Solutions Department, the report was the product of roughly three years of consultation with more than one hundred academics, customers, partners and research institutes across more than fifty workshops.
+
AI workloads aboard satellites require software capable of remotely deploying, updating and coordinating models after launch.
  
−
The report frames the central problem of the decade as a widening gap between compute demand — growing faster than [[Moore's law]] — and the resource constraints on supply. It defines five future scenarios and six key technical characteristics of future data centers. Among these, the "new patterns" characteristic explicitly names '''underwater data centers and space data centers''' as construction patterns to be developed for varied application scenarios, alongside "big clusters" (shifting the unit of construction and operations from the server to the rack and then the whole facility) and "lightweight edges" (pushing capacity from core to edge for low-latency and data-residency reasons).
+
Traditional spacecraft software is often designed around a fixed mission and payload. Cloud-native approaches instead seek to treat spacecraft computing resources as remotely configurable infrastructure.
  
−
Huawei's framing is notable for treating orbital compute as an extension of an edge-computing continuum rather than as a replacement for terrestrial hyperscale capacity — a materially more conservative position than that later taken by [[SpaceX]] or [[Blue Origin]].
+
Technologies investigated for this purpose include:
  
−
=== Cloud-native satellites and the Tiansuan Constellation ===
+
* containers and [[Kubernetes]]-derived orchestration;
 +
* remote model deployment;
 +
* over-the-air software updates;
 +
* federated learning;
 +
* distributed inference;
 +
* workload migration between spacecraft and ground systems.
  
−
Huawei's most concrete flight heritage in this area comes through '''[[KubeEdge]]''', the [[Kubernetes]]-based edge orchestration project that Huawei Cloud initiated and open-sourced in November 2018 and later donated to the [[Cloud Native Computing Foundation]], together with its edge-AI subproject '''Sedna''' and the '''MindSpore''' deep-learning framework.
+
The KubeEdge-based Tiansuan experiments demonstrated elements of this model by allowing onboard AI workloads to be updated and coordinated with terrestrial systems.
  
−
Huawei Cloud joined as one of the first co-construction partners of the '''Tiansuan Constellation''' (天算星座), an open in-orbit research platform initiated in October 2021 by [[Beijing University of Posts and Telecommunications]] (BUPT) with the commercial satellite manufacturer Spacety. The first satellite carrying the reconstructed KubeEdge stack launched from [[Jiuquan Satellite Launch Center]] on 7 December 2021 and was described as the world's first cloud-native satellite.
+
Similar software-defined approaches are being considered by other orbital-computing projects, although there is not yet a common industry standard for managing large distributed AI clusters in space.
  
−
Reported results from the platform include:
+
=== Limitations ===
  
−
* Ground-target identification accuracy improved by more than 50% through collaborative inference between satellite and ground station, using a lightweight model on orbit and a high-precision model on the ground with confidence-based escalation;
+
AI workloads intensify several of the broader technical challenges associated with orbital data centers.
−
* A reduction of approximately 90% in the volume of data returned to Earth;
 
−
* Support for continuous over-the-air update of on-orbit AI models, incremental deep learning and federated learning.
 
  
−
The work was presented as a keynote at KubeCon + CloudNativeCon Europe 2022. The architectural claim is that containerisation breaks the traditional "one satellite, one mission, one payload" model in which software is fixed for the life of the spacecraft.
+
; Power consumption
 +
: Modern AI accelerators can consume hundreds of watts per processor, while large training clusters can require megawatts of electrical power. Providing comparable power in orbit requires large solar arrays, energy storage and power-distribution systems.
  
−
=== On-orbit hardware and stated obstacles ===
+
; Heat rejection
 +
: Almost all electrical power consumed by computing hardware ultimately becomes heat. High-performance AI processors therefore increase the radiator area and thermal-control requirements of an orbital facility.
  
−
Research associated with Huawei's technology stack has explored radiation-tolerant [[commercial off-the-shelf]] CPUs and GPUs hosting virtual machines, with neighbouring satellites linked by optical laser terminals to form cooperative computing clusters. Huawei's own AI silicon line ([[Ascend (processor)|Ascend]]) and the MindSpore framework provide the software-hardware pairing for Chinese in-orbit inference work, though Huawei has not announced a dedicated space-qualified accelerator comparable to [[Nvidia]]'s Space-1 module.
+
; Radiation
 +
: Advanced processors and high-bandwidth memory can be vulnerable to radiation-induced errors. Shielding, redundancy and fault-tolerant software can reduce this risk but add mass, cost or complexity.
  
−
Huawei's published material is candid about the obstacles: the launch cost of heavy server racks; securing continuous solar power; and rejecting heat in vacuum, where convective and evaporative cooling are unavailable. These are the same three constraints identified independently by the European ASCEND study and by sceptical analysts (see [[#Criticism and open questions|Criticism]]).
+
; Hardware replacement
 +
: AI accelerators evolve more rapidly than conventional satellite platforms. A spacecraft designed to remain in service for a decade or more may contain computing hardware that becomes commercially outdated within several years.
 +
 
 +
; Networking
 +
: Distributed AI training can require extremely high communication bandwidth between processors. Reproducing terrestrial data-center interconnect performance using optical links between moving spacecraft remains technically challenging.
 +
 
 +
; Economics
 +
: Large orbital AI clusters depend heavily on launch costs, spacecraft manufacturing costs and the ability to operate computing hardware reliably without conventional maintenance.
 +
 
 +
For these reasons, near-term orbital AI applications are generally more practical for onboard inference and data reduction than for replacing terrestrial hyperscale AI training facilities. Whether large-scale training becomes economically competitive depends on advances in launch systems, power generation, thermal management, radiation tolerance and optical networking.
 +
```
  
 
== United States ==
 
== United States ==
Line 308: Line 415:
 
=== Starcloud ===
 
=== Starcloud ===
  
−
'''Starcloud''' (Redmond, Washington; formerly Lumen Orbit) is the most advanced dedicated orbital data center startup by flight heritage. Founded in 2024 and a graduate of [[Y Combinator]]'s Summer 2024 cohort, it published an early white paper proposing multiple gigawatts of orbital AI compute.
+
'''Starcloud''' of Redmond, Washington, formerly known as Lumen Orbit, was founded in 2024 and participated in [[Y Combinator]]'s Summer 2024 cohort. The company has proposed dedicated satellites for artificial-intelligence computing and published concepts for eventually scaling orbital computing into the gigawatt range.
  
−
* '''Starcloud-1''' — a 60 kg satellite launched by SpaceX in November 2025, carrying the first [[Nvidia H100]] data-center GPU to operate in low Earth orbit. Starcloud reported training an AI model on orbit, a first for the sector.
+
* '''Starcloud-1''' — a 60 kg satellite launched by SpaceX in November 2025 carrying an [[Nvidia H100]] data-center GPU. Starcloud reported using the system to train an artificial-intelligence model in orbit.
−
* '''FCC filing''' — February 2026, for a constellation of up to '''88,000''' satellites in sun-synchronous dusk–dawn orbits between 600 and 850 km.
+
* '''FCC filing''' — filed in February 2026 for a constellation of up to 88,000 satellites in sun-synchronous dusk–dawn orbits between approximately 600 and 850 km.
−
* '''Funding''' — $170 million Series A at a $1.1 billion valuation in March 2026 (Benchmark, EQT); a $250 million extension announced 21 August 2026 at a $2.3 billion valuation, led by Manhattan West, with '''Nvidia''' and Cisco Investments joining as new investors. Total capital raised: approximately $450 million.
+
* '''Funding''' — the company announced a $170 million Series A financing at a reported $1.1 billion valuation in March 2026. A $250 million extension announced on 21 August 2026 reportedly valued the company at $2.3 billion and included [[NVIDIA]] and Cisco Investments among participating investors.
−
* '''Starcloud-2''' — a 450 kg spacecraft carrying Nvidia Blackwell-generation GPUs, planned for launch on a Falcon 9 in January 2027, intended to run commercial workloads for early customers including the AI infrastructure provider Crusoe.
+
* '''Starcloud-2''' — a planned 450 kg spacecraft using [[NVIDIA]] Blackwell-generation GPUs and intended to support early commercial workloads.
−
* '''Starcloud-3''' — approximately 3 tonnes and 200 kW, designed for deployment by [[SpaceX Starship|Starship]], with production lines being established at a 100,000 sq ft facility in Woodinville, Washington.
+
* '''Starcloud-3''' — a proposed approximately 3-tonne, 200 kW spacecraft intended for deployment using higher-capacity launch vehicles.
−
* '''Long-term concept''' — spacecraft with solar arrays on the order of 4 km across supporting gigawatt-class compute, and roughly 20 GW of orbital capacity at full constellation scale.
+
* '''Long-term concepts''' — Starcloud has described much larger spacecraft and constellation architectures intended to provide multi-gigawatt aggregate computing capacity.
  
−
Starcloud is working with Nvidia on the '''Space-1 Vera Rubin Module''', and hopes to fly it in late 2028. Chief executive Philip Johnston has been explicit that the company's scaling is gated on Starship availability, describing pre-Starship operations on Falcon 9 as treading water.
+
Starcloud has also worked with [[NVIDIA]] on the proposed Space-1 Vera Rubin Module. The company's larger architectures depend heavily on substantial reductions in launch cost and increases in available payload capacity.
  
 
=== Nvidia ===
 
=== Nvidia ===
  
−
Nvidia is a supplier and, since August 2026, an investor rather than an operator. At GTC in March 2026 it announced a space computing line:
+
'''[[Nvidia]]''' participates primarily as a semiconductor and computing-platform supplier rather than as an orbital data-center operator.
 +
 
 +
At GTC in March 2026, [[NVIDIA]] announced space-oriented computing products and partnerships, including:
  
−
* '''Space-1 Vera Rubin Module''' — its first purpose-built accelerator for orbital data centers, claimed to deliver up to 25× the AI compute of the H100 in orbit, with a tightly integrated CPU-GPU architecture intended to run large language and foundation models on orbit. Cooling is by large radiators rather than air or liquid loops.
+
* '''Space-1 Vera Rubin Module''' — a purpose-built computing module proposed for orbital data-center applications using [[NVIDIA]]'s Vera Rubin architecture.
−
* '''IGX Thor''' and '''Jetson Orin''' — edge-AI inferencing platforms for geospatial intelligence and autonomous space operations.
+
* '''IGX Thor''' and '''Jetson Orin''' — edge-computing platforms intended for applications including geospatial intelligence, autonomous operations and onboard inference.
  
−
Named launch-slate partners include Aetherflux (now Cowboy Space), Axiom Space, Kepler Communications, Planet Labs, Sophia Space and Starcloud. This positions Nvidia as the common silicon layer beneath several nominally competing constellations, and is the reason the phrase "Rubin servers" has become shorthand for the category.
+
Organizations publicly associated with [[NVIDIA]]'s space-computing efforts have included Axiom Space, Cowboy Space, Kepler Communications, Planet Labs, Sophia Space and Starcloud.
  
−
Nvidia chief executive Jensen Huang gave a notably measured assessment on the company's Q4 FY2026 earnings call in February 2026, saying that the economics of orbital data centers are poor at present but should improve, that space offers abundant energy and room, and that the absence of airflow makes heat rejection the binding constraint, requiring fairly large radiators.
+
[[NVIDIA]] chief executive Jensen Huang stated in February 2026 that orbital data-center economics remained unfavorable at that time but could improve as launch and infrastructure technologies developed. He also identified heat rejection as an important engineering limitation because spacecraft cannot rely on atmospheric airflow for cooling.
  
 
=== SpaceX and xAI ===
 
=== SpaceX and xAI ===
  
−
SpaceX has made orbital compute the strategic centrepiece of its corporate reorganisation.
+
'''[[SpaceX]]''' has proposed large-scale orbital computing as one possible application of future high-capacity launch systems and satellite platforms.
  
−
* '''FCC filing''' — 30 January 2026, seeking authority for up to '''one million''' orbital data center satellites at altitudes between 500 and 2,000 km. The filing projects that launching one million tonnes of satellites annually would yield on the order of 100 GW of AI compute capacity.
+
* '''FCC filing''' — on 30 January 2026, SpaceX sought authority for a proposed system containing up to one million orbital data-center satellites operating between approximately 500 and 2,000 km.
−
* '''xAI acquisition''' — closed 2 February 2026 in an all-stock transaction (one xAI share converting to 0.1433 SpaceX shares), creating a combined entity valued at approximately $1.25 trillion. Musk framed the rationale explicitly around space-based AI and orbital data centers, with xAI functioning as the implicit anchor customer.
+
* '''xAI''' — SpaceX completed an acquisition of [[xAI]] on 2 February 2026 in an all-stock transaction. Elon Musk associated the combined company's long-term strategy with space-based artificial-intelligence infrastructure.
−
* '''IPO''' — SpaceX filed for a listing targeted for mid-2026, reportedly seeking to raise upwards of $75 billion at a valuation of $1.5 trillion or more, with proceeds supporting orbital data center development.
+
* '''Manufacturing proposals''' — SpaceX, Tesla and xAI have discussed large-scale semiconductor and computing-hardware manufacturing intended to support both terrestrial and orbital applications.
−
* '''Terafab''' — announced at an Austin event on 21–22 March 2026 as a joint venture of SpaceX, [[Tesla, Inc.|Tesla]] and xAI. The proposed facility would produce one terawatt of processors annually, which Musk characterised as roughly fifty times current combined advanced-chip production. Initial cost estimates of $20–25 billion have been revised substantially upward in later filings. The site would span approximately 100 million square feet and draw more than 10 GW. Around 80% of output is earmarked for orbital deployment and 20% for terrestrial use in Tesla's robotics and autonomy programmes.
+
* '''Spacecraft concepts''' — publicly discussed designs include smaller orbital-compute satellites and larger high-power spacecraft intended to make use of future launch capacity.
−
* '''Spacecraft''' — publicly disclosed concepts include the AI Sat Mini and a larger AI1-class node with a wingspan on the order of 70 m at roughly 100–150 kW per satellite. Musk closed the Terafab presentation with a concept video of data center satellites manufactured on the Moon and launched by electromagnetic mass driver.
+
 
 +
These systems remain proposals, and their economics depend on launch cost, spacecraft manufacturing, power generation, thermal control and regulatory approval.
  
 
=== Blue Origin ===
 
=== Blue Origin ===
  
−
Blue Origin entered with two coupled programmes:
+
'''[[Blue Origin]]''' has proposed orbital-computing and communications systems associated with two programs.
  
−
* '''Project Sunrise''' — FCC application SAT-LOA-20260310-00118, filed 19 March 2026, for up to '''51,600''' data center satellites in sun-synchronous orbits between 500 and 1,800 km, relying primarily on optical inter-satellite links with limited Ka-band use for control and reliability functions.
+
; Project Sunrise
−
* '''TeraWave''' — a 5,408-satellite connectivity constellation announced in January 2026, providing the backhaul layer for Project Sunrise as well as serving terrestrial data center, enterprise and government customers, with deployment scheduled to begin in Q4 2027.
+
: An FCC application filed in March 2026 proposed up to 51,600 data-center satellites in sun-synchronous orbits between approximately 500 and 1,800 km. The architecture relies substantially on optical inter-satellite communications.
  
−
Blue Origin's ownership of New Glenn gives it launch independence that most competitors lack. The filing produced an unusual regulatory episode: [[Amazon (company)|Amazon]] — like Blue Origin, controlled by [[Jeff Bezos]] — had petitioned the FCC to deny SpaceX's million-satellite application, whereupon SpaceX asked the Commission to apply Amazon's own arguments to Blue Origin's substantially similar filing. NASA filed an objection to Project Sunrise in May 2026, as did DarkSky International on light-pollution grounds.
+
; TeraWave
 +
: A proposed 5,408-satellite communications constellation announced in January 2026. It is intended to provide connectivity for terrestrial and space-based customers and could also provide networking infrastructure for orbital-computing systems.
 +
 
 +
Blue Origin's access to the [[New Glenn]] launch vehicle provides a potential vertically integrated launch capability. Project Sunrise has also attracted regulatory objections concerning orbital congestion and effects on astronomy.
  
 
=== Google — Project Suncatcher ===
 
=== Google — Project Suncatcher ===
  
−
Announced 4 November 2025, '''Project Suncatcher''' is a research "moonshot" investigating constellations of solar-powered satellites carrying Google [[Tensor Processing Unit]]s linked by free-space optical communications.
+
'''[[Google]] Research''' announced '''Project Suncatcher''' on 4 November 2025 as a research program investigating constellations of solar-powered satellites carrying Google [[Tensor Processing Unit]]s connected through free-space optical links.
  
−
Key elements of the accompanying preprint:
+
The reference architecture described in accompanying research included:
  
−
* A reference architecture of roughly 81 satellites within a cluster radius of about 1 km, in low Earth orbit at approximately 640 km.
+
* approximately 81 satellites operating as a closely coordinated cluster;
−
* Radiation testing of the Trillium (TPU v6e) generation in a 67 MeV proton beam. High-bandwidth memory was the most sensitive subsystem but showed irregularities only after a cumulative dose of about 2 krad(Si), roughly three times the expected shielded five-year mission dose of 750 rad(Si). No hard failures attributable to total ionising dose were observed up to the maximum tested dose of 15 krad(Si).
+
* an orbit near 640 km;
−
* A conclusion that the concept is not precluded by fundamental physics or insurmountable economics, contingent on solving thermal management, high-bandwidth ground communication and on-orbit reliability.
+
* testing of Trillium-generation TPU hardware under proton radiation;
−
* Economic parity with terrestrial systems estimated to become plausible around 2035 if launch costs fall below roughly $200/kg.
+
* optical links intended to support distributed machine-learning workloads;
 +
* analysis of launch-cost thresholds required for economic competitiveness.
  
−
Google is partnering with '''[[Planet Labs]]''' to launch two prototype satellites by early 2027 to test TPU hardware in orbit and validate optical inter-satellite links for distributed machine-learning tasks.
+
The research concluded that the concept was not excluded by fundamental physical constraints but identified thermal management, high-bandwidth communications, launch economics and long-term hardware reliability as major engineering challenges.
 +
 
 +
Google has partnered with '''[[Planet Labs]]''' on two prototype satellites planned for launch in early 2027 to test TPU hardware and optical inter-satellite networking.
  
 
=== Axiom Space ===
 
=== Axiom Space ===
  
−
Axiom Space has pursued the incremental, kilowatt-scale path and holds a claim to the first commercial computing hardware operated aboard a crewed facility.
+
'''Axiom Space''' has followed a smaller-scale, incremental approach centered on computing hardware in low Earth orbit.
  
−
* '''AxDCU-1''' — a data-processing prototype powered by Red Hat Device Edge, launched to the [[International Space Station]] in August 2025 and deployed that autumn.
+
* '''AxDCU-1''' — a data-processing prototype using Red Hat Device Edge, launched to the [[International Space Station]] in August 2025.
−
* '''AxODC Node (ISS)''' — developed with Spacebilt, with an optical communication terminal from Skyloom and hardware from Phison Electronics and Microchip Technology, as a step toward 100 Gbps connectivity.
+
* '''AxODC Node''' — an orbital-computing system developed with partners including Spacebilt, Skyloom, Phison Electronics and Microchip Technology.
−
* '''ODC Nodes 1 and 2''' — the first two dedicated orbital data center nodes launched to LEO on 11 January 2026, hosted on Kepler Communications spacecraft at approximately 400 km, giving round-trip latencies in the 5–20 ms range.
+
* '''ODC Nodes 1 and 2''' — dedicated orbital data-center nodes launched to LEO in January 2026 on Kepler Communications spacecraft.
−
* '''Optical links''' — integration with Kepler Communications US and Skyloom Global relay constellations, targeting up to 10 Gbit/s and compliance with [[Space Development Agency]] interoperability standards.
+
* '''Optical communications''' — integration with Kepler Communications and Skyloom relay systems.
−
* '''ODC T1''' — a roughly half-cubic-metre server rack planned for launch by 2027, with subsequent modules on Axiom Station. Axiom has stated an intention to scale from kilowatts to megawatts and has floated lunar and Martian deployments as follow-ons.
+
* '''ODC T1''' — a larger server module proposed for launch by 2027, followed by later deployments associated with Axiom Station.
  
−
Axiom received a Space Exploration & Aeronautics Research Fund grant of up to $5.5 million from the Texas Space Commission for ODC development, and has worked with [[Amazon Web Services]] on edge hardware.
+
Axiom has described a gradual progression from kilowatt-scale systems toward larger orbital-computing installations.
  
−
=== Cowboy Space (formerly Aetherflux) ===
+
=== Cowboy Space ===
  
−
Founded in 2024 by Robinhood co-founder Baiju Bhatt as a space-based solar power company, '''Aetherflux''' rebranded as '''Cowboy Space Corporation''' in May 2026 alongside a pivot to orbital compute, following a $275 million Series B at a roughly $2 billion valuation.
+
'''Cowboy Space Corporation''', formerly '''Aetherflux''', was founded in 2024 by Robinhood co-founder Baiju Bhatt. The company initially focused on space-based solar power before expanding into orbital computing.
  
−
Its distinguishing architectural choice is vertical integration of launch and payload: the upper stage of a company-built rocket ''becomes'' the data center once in orbit. The '''Stampede''' constellation (FCC application SAT-LOA-20260323-00135, up to '''20,000''' satellites) specifies:
+
Its proposed '''Stampede''' architecture combines launch and computing infrastructure by using a launch-vehicle stage as part of the deployed orbital system.
  
−
* Dawn–dusk sun-synchronous orbits between 700 and 1,000 km;
+
The associated FCC proposal describes:
−
* Satellites of 20,000–25,000 kg each, generating approximately 1 MW of usable power;
 
−
* Just under 800 GPUs per satellite, built around the Nvidia Space-1 Vera Rubin module;
 
−
* Optical inter-satellite links as the primary data path, with first proprietary launch of a one-megawatt node targeted before the end of 2028.
 
  
−
Cowboy has signed a Space Act Agreement with NASA's Stennis Space Center for propulsion testing.
+
* dawn–dusk sun-synchronous orbits between approximately 700 and 1,000 km;
 +
* spacecraft in the approximately 20,000–25,000 kg class;
 +
* around 1 MW of usable electrical power per spacecraft;
 +
* several hundred GPU modules per spacecraft;
 +
* optical inter-satellite networking;
 +
* an initial megawatt-class orbital node targeted for the late 2020s.
  
−
=== Other US ventures ===
+
Cowboy Space has also entered into a Space Act Agreement with NASA's Stennis Space Center for propulsion-related testing.
 +
 
 +
=== Other North American and US-associated ventures ===
  
 
{| class="wikitable"
 
{| class="wikitable"
−
! Organization !! Location !! Approach !! Status (Sept 2026)
+
! Organization
 +
! Approach
 +
! Status or role
 
|-
 
|-
−
| '''Orbital Compute''' || US || 100,000 modular 100 kW satellites supporting ~10 GW; incremental deployment using third-party launch || FCC filing; first launch targeted 2027
+
| '''Orbital Compute'''
 +
| Proposed modular satellites providing distributed orbital computing capacity
 +
| FCC filing; first launch targeted for 2027
 
|-
 
|-
−
| '''Sophia Space''' || US || TILE (Thermal-Integrated LEO Edge) — tabletop-sized tiles combining solar generation with passive radiative cooling, assembled into racks; Caltech-linked patent || ~$13.9M raised incl. $10M seed (Feb 2026); orbital demo targeted late 2027/early 2028 on an Apex Space bus
+
| '''Sophia Space'''
 +
| TILE (Thermal-Integrated LEO Edge) architecture combining solar generation, computing and radiative cooling
 +
| Orbital demonstration targeted for the late 2020s
 
|-
 
|-
−
| '''Lonestar Data Holdings''' || Florida || Off-planet data ''storage'' for resilience rather than AI compute; lunar and cislunar || Operated storage hardware aboard an Intuitive Machines lunar lander, Feb 2025; $120M agreement with Sidus Space for six satellites, first with 15 PB at an Earth–Moon libration point, launching 2027–2030
+
| '''Lonestar Data Holdings'''
 +
| Off-planet data storage and infrastructure resilience, including lunar and cislunar systems
 +
| Has operated storage hardware on a lunar mission; additional spacecraft planned
 
|-
 
|-
−
| '''OrbitsEdge''' || Cocoa Beach, FL || Radiation-hardened, datacenter-grade compute enclosures sold to satellite operators; partnership with [[Hewlett Packard Enterprise|HPE]] || First orbital demonstration planned 2026
+
| '''OrbitsEdge'''
 +
| Radiation-protected enclosures intended to host commercial data-center computing hardware aboard spacecraft
 +
| Demonstration program under development
 
|-
 
|-
−
| '''Kepler Communications''' || Canada/US || Optical relay and compute backbone rather than a standalone ODC || 10 satellites launched 11 Jan 2026 (Falcon 9, Vandenberg); 300 kg each, ≥4 optical terminals, multi-GPU modules; $233M+ raised
+
| '''Kepler Communications'''
 +
| Optical relay and computing backbone
 +
| Initial optical-network spacecraft deployed
 
|-
 
|-
−
| '''Skyloom Global''' / '''Spacebilt''' || US || Optical communication terminals and platform integration for Axiom nodes || Flying
+
| '''Skyloom Global''' / '''Spacebilt'''
 +
| Optical communications and orbital-platform integration
 +
| Hardware associated with operational and planned orbital systems
 
|-
 
|-
−
| '''Ramon.Space''' || Israel/US || Radiation-tolerant space-resilient computing systems; enabler layer || Established supplier
+
| '''Ramon.Space'''
 +
| Radiation-tolerant computing systems
 +
| Established supplier
 
|-
 
|-
−
| '''HPE''' || US || Spaceborne Computer-2 aboard the ISS; consortium member in ASCEND; OrbitsEdge partner || Long-running ISS heritage
+
| '''Hewlett Packard Enterprise'''
 +
| Spaceborne computing aboard the ISS and participation in orbital-computing research programs
 +
| Long-running flight heritage
 
|}
 
|}
  
Line 413: Line 551:
 
=== Three-Body Computing Constellation ===
 
=== Three-Body Computing Constellation ===
  
−
The '''Three-Body Computing Constellation''' (三体计算星座), led by '''ADA Space''' (国星宇航, Chengdu) with '''Zhejiang Lab''', is the first dedicated orbital computing constellation to reach orbit.
+
The '''Three-Body Computing Constellation''' (三体计算星座), developed by '''ADA Space''' (国星宇航) with '''Zhejiang Lab''', is an orbital computing constellation designed to perform distributed artificial-intelligence and scientific-processing workloads.
  
−
* '''First launch''' — 12 satellites on a Long March 2D from Jiuquan on 14 May 2025, providing a combined 5 POPS (peta-operations per second) and 30 TB of onboard storage, with individual satellites rated around 744 TOPS. Inter-satellite laser links operate at 100 Gbit/s.
+
Twelve satellites were launched on a Long March 2D from Jiuquan on 14 May 2025.
−
* '''Programme scale''' — part of a wider '''Star-Compute Program''' targeting 2,800 satellites, with an ITU filing in place; 100 satellites planned by 2027 and a stated eventual capability of 1,000 POPS at 1,000-plus satellites.
+
 
−
* '''On-orbit AI''' — by February 2026, after roughly nine months of in-orbit testing, Zhejiang Lab reported ten AI models deployed on orbit, including an 8-billion-parameter remote sensing model and an 8-billion-parameter astronomical time-domain model, with inter-satellite networking validated and laser links reported at 99.99% availability over an eight-day test.
+
Reported characteristics include:
−
* '''Alibaba Qwen3''' — deployed to the constellation in January 2026; a complex task uploaded from the ground was reportedly processed jointly across satellites and returned in under two minutes.
+
 
−
* Commercial partners include Alibaba Cloud, Kepu Cloud and iSoftStone. ADA Space has pursued a Hong Kong listing and is separately advancing its own constellation split between inference and training spacecraft.
+
* approximately 5 POPS of combined computing capacity in the initial group;
 +
* approximately 30 TB of onboard storage;
 +
* individual satellite computing capacity of approximately 744 TOPS;
 +
* 100 Gbit/s inter-satellite laser links;
 +
* deployment of multiple artificial-intelligence models for remote sensing and astronomy.
 +
 
 +
The program forms part of the broader '''Star-Compute Program''', which has described plans for a substantially larger constellation.
 +
 
 +
In January 2026, Alibaba's Qwen3 model was reported to have been deployed to the constellation for distributed processing experiments. Other commercial participants have included Alibaba Cloud, Kepu Cloud and iSoftStone.
  
 
=== Beijing Astro-Future Institute ===
 
=== Beijing Astro-Future Institute ===
  
−
The '''Beijing Astro-Future Institute of Space Technology''' (北京星辰未来空间技术研究院) is building what Beijing has publicly designated a space data center programme, announced at a municipal working conference in November 2025. It raised at least ¥140 million (approximately US$20 million) in June 2025, with reported backing from Lenovo and the Beijing municipal government, and leads a "space data center innovation consortium".
+
The '''Beijing Astro-Future Institute of Space Technology''' has proposed a dedicated space data-center program supported by a broader industry consortium.
 +
 
 +
Its published development roadmap includes:
 +
 
 +
# '''2025–2027''' — demonstration of power supply, thermal management and onboard processing;
 +
# '''2028–2030''' — development of orbital assembly and systems capable of processing data originating on Earth;
 +
# '''2031–2035''' — larger modular orbital data-center infrastructure using mass production and on-orbit assembly.
 +
 
 +
Longer-term concepts include interconnected high-power facilities in sun-synchronous orbit.
 +
 
 +
=== State and academic programs ===
 +
 
 +
Several Chinese institutions have separately pursued orbital-computing research.
 +
 
 +
* '''CASC''' — the [[China Aerospace Science and Technology Corporation]] has discussed space-based digital and intelligent infrastructure as part of planning for the 2026–2030 period.
 +
* '''Xingshu Plan''' (星枢计划) — a Shanghai initiative associated with Fudan University, announced during the 2026 World AI Conference. The proposed program progresses from experimental computing and edge satellites toward a larger on-demand orbital-computing service.
 +
* '''Tiansuan Constellation''' (天算星座) — an open in-orbit research platform initiated by the [[Beijing University of Posts and Telecommunications]] with commercial satellite partner Spacety.
 +
* '''CAICT''' — the China Academy of Information and Communications Technology has established an industry committee concerned with space-computing standards.
  
−
Its stated three-phase roadmap:
+
Research in China on space-based solar power and high-power spacecraft systems also overlaps technically with some of the power-generation and thermal-management requirements of orbital computing.
  
−
# '''2025–2027''' — solve on-orbit power supply and heat rejection; launch a demonstration satellite (Chenguang-1); build a first-phase constellation of 200 kW total power and 1,000 POPS, supporting "space data computed in space".
+
=== Huawei, KubeEdge and cloud-native satellite research ===
−
# '''2028–2030''' — master on-orbit assembly and construction to reduce build and operating cost; second-phase constellation supporting "ground data computed in space".
 
−
# '''2031–2035''' — mass production and on-orbit docking to build very large space data centers, with core compute supplied from orbit.
 
  
−
Longer term the institute has described dedicated data centers in 700–800 km sun-synchronous orbit in the first half of the 2030s: few in number but large, with concepts for a sixteen-spacecraft constellation of interlinked gigawatt-scale facilities.
+
'''[[Huawei]]''' has participated in orbital-computing research principally through data-center forecasting, edge-computing software and collaborative satellite research rather than through a publicly announced dedicated orbital data-center constellation.
  
−
=== State programmes ===
+
Huawei's ''Data Center 2030'' report, released at HUAWEI CONNECT 2023, identified underwater and space-based data centers among several possible future infrastructure patterns. The same report also discussed larger computing clusters and lightweight edge infrastructure.
  
−
* '''CASC / 15th Five-Year Plan''' — ahead of the 2026–2030 plan, the [[China Aerospace Science and Technology Corporation]] identified space infrastructure as one of four focus areas, describing gigawatt-scale space-based digital and intelligent infrastructure and a new architecture integrating cloud, edge and endpoint compute, with the goal of enabling space-data/space-compute, ground-data/space-compute and simultaneous space-ground computing.
+
Huawei Cloud initiated the '''[[KubeEdge]]''' project in 2018 as a Kubernetes-based edge-computing framework and later contributed it to the [[Cloud Native Computing Foundation]]. Related technologies have included the Sedna edge-AI project and the MindSpore machine-learning framework.
−
* '''Xingshu Plan''' (星枢计划) — Shanghai's flagship initiative, unveiled with Fudan University at the 2026 World AI Conference in July 2026. Three phases scale from a verification constellation of 2 computing plus 12 edge satellites, to 50 computing and 100 edge satellites, to a 1,000-satellite on-demand orbital computing service — effectively a rental model for governments and enterprises in weather forecasting, disaster response, maritime monitoring and grid inspection. A space computing hub opened in Songjiang on 31 August 2026.
 
−
* '''Tiansuan Constellation''' — BUPT/Spacety open research platform, with Huawei Cloud as a co-construction partner (see [[#Huawei|above]]).
 
−
* '''CAICT''' — the China Academy of Information and Communications Technology has established a space computing power professional committee to coordinate standards.
 
  
−
Chinese programmes benefit from a decade and a half of research on [[space-based solar power]], begun in 2008 with technology testing from 2013, since a space data center reuses much of the same power-generation and thermal-management hardware.
+
Huawei Cloud subsequently participated as a co-construction partner in the '''Tiansuan Constellation''', an experimental satellite research program led by the Beijing University of Posts and Telecommunications with Spacety.
 +
 
 +
A satellite launched from [[Jiuquan Satellite Launch Center]] in December 2021 carried a reconstructed KubeEdge software stack and was described by the project as a cloud-native satellite.
 +
 
 +
Experiments associated with the program reported:
 +
 
 +
* collaborative inference between satellite and ground-based computing systems;
 +
* substantial reduction in the volume of data transmitted to Earth;
 +
* over-the-air updating of onboard artificial-intelligence models;
 +
* experiments involving incremental and federated learning.
 +
 
 +
Research using related Chinese computing stacks has also examined commercial off-the-shelf CPUs and GPUs, virtualized workloads, optical inter-satellite links and distributed computing.
 +
 
 +
Huawei has not publicly announced a dedicated orbital data-center constellation or a purpose-built space-qualified accelerator directly comparable with some specialized orbital-computing hardware proposed by other vendors.
  
 
== Europe ==
 
== Europe ==
Line 446: Line 617:
 
=== ASCEND ===
 
=== ASCEND ===
  
−
'''ASCEND''' (Advanced Space Cloud for European Net zero emission and Data sovereignty) is a European Commission feasibility study funded under [[Horizon Europe]], contracted to a consortium led by '''[[Thales Alenia Space]]''' in 2022 and launched in 2023. Partners span environmental analysis (Carbone 4, VITO), cloud and IT (Orange Business, CloudFerro, Hewlett Packard Enterprise), launch (ArianeGroup) and orbital systems ([[German Aerospace Center|DLR]], [[Airbus Defence and Space]], Thales Alenia Space).
+
'''ASCEND''' ('''Advanced Space Cloud for European Net zero emission and Data sovereignty''') is a European Commission feasibility study funded through [[Horizon Europe]].
 +
 
 +
The project was contracted to a consortium led by '''[[Thales Alenia Space]]''' in 2022 and launched in 2023. Participants have included organizations working in environmental analysis, cloud computing, launch systems and spacecraft engineering, including Carbone 4, VITO, Orange Business, CloudFerro, Hewlett Packard Enterprise, ArianeGroup, the [[German Aerospace Center]], Airbus Defence and Space and Thales Alenia Space.
 +
 
 +
Results published in June 2024 concluded that space-based data centers could be technically feasible and potentially provide environmental benefits under certain assumptions.
 +
 
 +
The study identified several conditions and findings:
 +
 
 +
* launch emissions would need to decrease substantially for a large orbital data-center system to provide a net environmental benefit;
 +
* orbital computing would avoid terrestrial water consumption for data-center cooling;
 +
* modular systems could potentially be assembled robotically in orbit;
 +
* economic feasibility depends strongly on launch, manufacturing and operations costs.
  
−
Findings published in June 2024:
+
The program has discussed a long-term target of approximately 1 GW of orbital capacity by 2050.
  
−
* Space-based data centers are technically feasible and can deliver a net environmental benefit '''provided''' a launcher roughly ten times less emissive than current vehicles is developed — validated as achievable with ArianeGroup input and analysis from ESA's PROTEIN study.
+
A smaller in-orbit demonstration intended to test European orbital data-center technologies has been targeted for 2028.
−
* Orbital facilities consume no water for cooling, a material advantage under increasing drought pressure.
 
−
* A target of 1 GW of orbital capacity by 2050, with the study concluding the project economically viable. (Reported return-on-investment figures vary considerably between sources, from several billion to several hundred billion euros by 2050; the primary study documentation should be consulted.)
 
−
* Modular infrastructure would be robotically assembled on orbit using technologies from the '''EROSS IOD''' (European Robotic Orbital Support Services In-Orbit Demonstrator), led by Thales Alenia Space, with a first flight in 2026.
 
  
−
An ASCEND in-orbit demonstration deploying a small-scale orbital data center module to validate European technologies is targeted for 2028. The programme is framed principally around the [[European Green Deal]] and European digital sovereignty rather than around AI compute scaling.
+
ASCEND is framed in part around the [[European Green Deal]], environmental impact and European digital sovereignty.
  
 
== Japan ==
 
== Japan ==
Line 461: Line 640:
 
=== Space Compass ===
 
=== Space Compass ===
  
−
'''Space Compass Corporation''' is a 50/50 joint venture of '''[[Nippon Telegraph and Telephone|NTT]]''' and '''SKY Perfect JSAT''', agreed in April 2022 following a 2021 collaboration and formally established in July 2022 in Chiyoda-ku, Tokyo, with an initial investment of ¥6 billion.
+
'''Space Compass Corporation''' is a joint venture between '''[[Nippon Telegraph and Telephone|NTT]]''' and '''SKY Perfect JSAT'''. The partnership was agreed in 2022 following earlier collaboration between the companies.
 +
 
 +
Its proposed '''Space Integrated Computing Network''' combines several infrastructure layers:
 +
 
 +
* optical relay links for transferring satellite data;
 +
* orbital computing systems capable of processing data before it is returned to Earth;
 +
* NTT's IOWN all-photonics networking technology;
 +
* high-altitude platform station services.
  
−
Its '''Space Integrated Computing Network''' is a multi-orbit, optical-communication-based infrastructure combining:
+
The architecture emphasizes an integrated communications and computing network rather than a standalone hyperscale orbital data center.
  
−
* An optical data relay service carrying observation-satellite data to the ground via GEO, overcoming the capacity and contact-window limits of radio downlink;
+
Space Compass has participated in Japanese government programs for next-generation optical communications and space infrastructure.
−
* A space data center layer, progressively adding satellites with advanced computing functions so that data is analysed on orbit and only useful results returned;
 
−
* NTT's '''IOWN''' all-photonics technology as the transport substrate;
 
−
* HAPS (high-altitude platform station) services for low-latency coverage, disaster response and remote areas.
 
  
−
Space Compass has been selected under JAXA's Space Strategy Fund for next-generation optical data relay and holds a Japanese Ministry of Defense contract for a geostationary optical communication technology demonstration. The venture predates most Western ODC announcements by several years, though its schedule has slipped from the originally stated 2025 service start.
+
== Other regions and international programs ==
  
−
== Other regions ==
+
=== India ===
  
−
* '''India''' — '''SkyServe''' develops orbital edge computing software, working with NASA's [[Jet Propulsion Laboratory]] to test AI models on D-Orbit spacecraft.
+
'''SkyServe''' develops software for orbital edge computing and has worked with NASA's [[Jet Propulsion Laboratory]] on experiments involving artificial-intelligence processing aboard commercial spacecraft.
−
* '''ESA''' — beyond PROTEIN and ASCEND, ESA's earlier PhiSat-1 cubesat (with Intel and Ubotica) demonstrated onboard AI inference for Earth observation.
+
 
 +
=== European Space Agency ===
 +
 
 +
In addition to its involvement in studies associated with ASCEND and related environmental analysis, the [[European Space Agency]] has supported earlier onboard-computing demonstrations.
 +
 
 +
The PhiSat-1 CubeSat, developed with partners including Intel and Ubotica, demonstrated artificial-intelligence inference for Earth-observation imagery.
 +
 
 +
== Enabling technologies ==
 +
 
 +
=== Optical networking ===
 +
 
 +
High-bandwidth optical inter-satellite links are a central component of many orbital-computing proposals because distributed workloads can require communication rates substantially above those traditionally used for spacecraft command and telemetry.
 +
 
 +
Optical networking is being developed or supplied by organizations including Kepler Communications, Skyloom, Space Compass and multiple satellite manufacturers.
 +
 
 +
=== Radiation tolerance ===
 +
 
 +
Commercial processors used in terrestrial data centers are not normally designed for the radiation environment encountered in orbit.
 +
 
 +
Approaches include:
 +
 
 +
* radiation-hardened processors;
 +
* shielding conventional commercial processors;
 +
* error-correcting memory;
 +
* redundant computing systems;
 +
* fault-tolerant software;
 +
* replacement or refresh cycles shorter than those of conventional communications satellites.
 +
 
 +
The cost and performance penalty associated with radiation tolerance remains an important economic variable.
 +
 
 +
=== Software orchestration ===
 +
 
 +
Cloud-native and container-based software has been proposed as a means of allowing spacecraft workloads to be updated after launch rather than being fixed for the entire mission.
 +
 
 +
Projects including KubeEdge and other edge-computing frameworks have demonstrated elements of this approach.
 +
 
 +
=== Launch systems ===
 +
 
 +
Launch cost is one of the largest variables affecting orbital-computing economics.
 +
 
 +
Large proposed systems assume some combination of:
 +
 
 +
* reusable launch vehicles;
 +
* high launch cadence;
 +
* substantially lower cost per kilogram;
 +
* standardized spacecraft manufacturing;
 +
* large payload capacity.
 +
 
 +
Some orbital data-center concepts therefore depend on launch systems that had not yet demonstrated the required operating economics as of 2026.
  
 
== Regulatory landscape ==
 
== Regulatory landscape ==
  
−
Orbital data centers occupy an awkward regulatory category: spacecraft whose primary commercial function is compute capacity rather than communications, Earth observation or navigation. Existing spectrum and licensing frameworks were not written for this case, and the reliance on optical inter-satellite links means comparatively little radio spectrum is requested relative to constellation size.
+
Orbital data centers do not fit neatly into traditional regulatory categories for communications, navigation or Earth-observation spacecraft.
  
−
=== Filings on record ===
+
Their use of optical inter-satellite links can reduce radio-spectrum requirements relative to communications constellations, but large proposed satellite populations raise issues involving orbital congestion, debris mitigation, astronomy and licensing.
 +
 
 +
=== Selected constellation filings ===
  
 
{| class="wikitable sortable"
 
{| class="wikitable sortable"
−
! Applicant !! System !! Satellites !! Filed !! Altitude / orbit
+
! Applicant
 +
! System
 +
! Proposed satellites
 +
! Filing period
 +
! Approximate altitude / orbit
 
|-
 
|-
−
| SpaceX || Orbital data center constellation || up to 1,000,000 || 30 Jan 2026 || 500–2,000 km
+
| SpaceX
 +
| Orbital data-center constellation
 +
| Up to 1,000,000
 +
| January 2026
 +
| 500–2,000 km
 
|-
 
|-
−
| Starcloud || (ODC constellation) || up to 88,000 || Feb 2026 || 600–850 km SSO dusk–dawn
+
| Starcloud
 +
| Orbital computing constellation
 +
| Up to 88,000
 +
| February 2026
 +
| 600–850 km, sun-synchronous
 
|-
 
|-
−
| Blue Origin || Project Sunrise || up to 51,600 || 19 Mar 2026 || 500–1,800 km SSO
+
| Blue Origin
 +
| Project Sunrise
 +
| Up to 51,600
 +
| March 2026
 +
| 500–1,800 km, sun-synchronous
 
|-
 
|-
−
| Blue Origin || TeraWave (connectivity backbone) || 5,408 || Jan 2026 || —
+
| Blue Origin
 +
| TeraWave
 +
| 5,408
 +
| January 2026
 +
| Multiple orbits
 
|-
 
|-
−
| Cowboy Space || Stampede Data Center System || up to 20,000 || Mar–May 2026 || 700–1,000 km dawn–dusk SSO
+
| Cowboy Space
 +
| Stampede Data Center System
 +
| Up to 20,000
 +
| 2026
 +
| 700–1,000 km, sun-synchronous
 
|-
 
|-
−
| Orbital Compute || (ODC constellation) || ~100,000 || 2026 || LEO
+
| Orbital Compute
 +
| Orbital computing constellation
 +
| Approximately 100,000
 +
| 2026
 +
| LEO
 
|}
 
|}
  
−
Filings are proposals under review, not approvals. Objections on file include Amazon's petition to deny SpaceX's application; SpaceX's reciprocal request that the same standards be applied to Blue Origin; a NASA objection to Project Sunrise in May 2026; and DarkSky International's objection on the grounds that such constellations would permanently alter the night sky.
+
Regulatory filings represent proposed systems rather than deployment approvals.
 +
 
 +
Large constellation proposals have drawn objections and requests for additional review from government agencies, competing operators and organizations concerned with orbital safety and astronomy.
  
 
== Criticism and open questions ==
 
== Criticism and open questions ==
  
−
=== Thermal ===
+
=== Thermal management ===
  
−
Heat rejection is the most frequently cited technical objection. In vacuum, radiation is the only mechanism available, and radiator area scales with dissipated power.
+
Heat rejection is one of the most frequently discussed engineering challenges for orbital data centers.
  
−
* The ISS active thermal control system rejects up to about 70 kW across roughly 422 m² of ammonia-loop radiators — approximately 166 W/m² in practice, well below theoretical maxima once solar exposure, Earth infrared and system losses are accounted for.
+
In vacuum, heat must ultimately be emitted through thermal radiation. High-power computing systems therefore require substantial radiator area and thermal-control infrastructure.
−
* Modelling published in mid-2026 puts an H100-class (~700 W) GPU at roughly 1.4 m² of radiator area and a 40 kW rack at roughly 80 m², with five-year surface degradation adding around 40% to required area. Optimised high-temperature radiator designs reach an effective PUE near 1.3 at approximately 2.5 kg/kW of thermal mass.
 
−
* Some analyses of gigawatt-class single facilities have produced radiator area figures in the millions of square feet.
 
−
* A design spiral is often described: packing accelerators densely concentrates heat faster than it can be moved to radiators, while spreading them out inflates interconnect and shielding mass. Thermal cycling roughly every 90 minutes adds fatigue loading.
 
  
−
The counter-argument from operators is that thermal management at 100 kW per spacecraft is a well-understood engineering trade with decades of LEO heritage, and that 2025–2026 flight data suggests ''power delivery'', not cooling, is the current limiter on hardware actually in orbit. The dispute is therefore less about physics than about the scale at which the trade turns unfavourable.
+
The [[International Space Station]], for example, uses large radiator systems to reject tens of kilowatts of heat.
 +
 
 +
Engineering studies of data-center-class GPUs in orbit have produced radiator requirements ranging from approximately square-metre scale for individual accelerators to much larger areas for multi-kilowatt racks.
 +
 
 +
The required area depends on factors including:
 +
 
 +
* operating temperature;
 +
* radiator orientation;
 +
* solar exposure;
 +
* infrared radiation from Earth;
 +
* surface degradation;
 +
* thermal-transfer efficiency;
 +
* spacecraft architecture.
 +
 
 +
Higher computing density can reduce spacecraft volume but makes heat transport more difficult. Conversely, spreading components across a larger structure increases mass, interconnect distance and shielding requirements.
 +
 
 +
Supporters of orbital computing argue that thermal-control engineering is well established at conventional spacecraft power levels, while critics question how efficiently it can scale to data-center or gigawatt-class systems.
  
 
=== Economics ===
 
=== Economics ===
  
−
The case rests almost entirely on launch cost. Falcon 9 costs are on the order of $2,700/kg; Starship targets approximately $200/kg; the single publicly contracted Starship price to date implies roughly $600/kg at full payload. Most published models place the inflection point for orbital compute somewhere between $200 and $500/kg.
+
Launch cost is generally considered a major determinant of whether large orbital data centers can compete with terrestrial infrastructure.
 +
 
 +
Published concepts commonly assume substantial reductions in launch cost compared with conventional expendable or partially reusable launch systems.
 +
 
 +
Additional economic issues include:
 +
 
 +
; Hardware obsolescence
 +
: Artificial-intelligence processors can become commercially outdated within several years, substantially faster than the design lifetime of many conventional satellites.
 +
 
 +
; Radiation qualification
 +
: Radiation-tolerant computing hardware can cost more or provide lower performance than equivalent terrestrial hardware.
 +
 
 +
; Maintenance
 +
: Terrestrial data centers allow technicians to replace failed components. Distributed orbital systems require redundancy, remote reconfiguration, robotic servicing or replacement spacecraft.
 +
 
 +
; Manufacturing
 +
: Constellations containing thousands of high-power computing spacecraft would require production volumes substantially beyond those of most existing satellite systems.
 +
 
 +
; Launch dependence
 +
: Many proposed business models depend on future reusable launch systems achieving cost and flight-rate targets that had not yet been demonstrated at full commercial scale as of 2026.
 +
 
 +
Economic estimates published by project developers vary widely and should generally be interpreted as projections rather than demonstrated operating costs.
 +
 
 +
=== Orbital debris ===
 +
 
 +
Large orbital-computing constellations could significantly increase the number and mass of spacecraft in low Earth orbit.
 +
 
 +
Concerns include:
 +
 
 +
* collision probability;
 +
* debris generation;
 +
* conjunction-management capacity;
 +
* spacecraft failures;
 +
* end-of-life disposal;
 +
* cascading collision scenarios associated with the [[Kessler syndrome]].
 +
 
 +
The scale of some proposals has therefore prompted calls for more detailed regulatory and environmental assessment before deployment.
 +
 
 +
=== Astronomy ===
 +
 
 +
Astronomers have raised concerns that very large constellations could affect optical and infrared observations through reflected sunlight, thermal emissions or radio interference.
 +
 
 +
Large spacecraft equipped with extensive solar arrays or radiator structures could present different observational effects from smaller communications satellites.
 +
 
 +
=== Atmospheric effects of reentry ===
  
−
Additional economic objections:
+
Large-scale satellite replacement would produce repeated atmospheric reentries.
  
−
* '''Hardware obsolescence''' — AI accelerators have a useful competitive life of roughly three to five years, against satellite platforms designed for 15–25 years. This has driven proposals for modular plug-in compute cards swappable under a long-lived bus, though on-orbit servicing at constellation scale is itself unproven.
+
Researchers have studied possible effects from aluminium oxides and other materials deposited into the upper atmosphere, although the long-term environmental consequences remain uncertain.
−
* '''Radiation premium''' — the cost and performance penalty of space-qualifying hardware is arguably the single variable determining whether the case closes.
 
−
* '''Maintenance''' — terrestrial facilities benefit from component replacement; distributed orbital assets do not.
 
−
* Operator projections should be treated as advocacy: Starcloud, for instance, has projected that operating a 40 MW orbital cluster over ten years would cost roughly $8.2 million against approximately $167 million for a terrestrial equivalent.
 
  
−
Sceptical commentary through 2026 has been substantial, including a widely circulated February 2026 video by science communicator Kyle Hill and remarks by Voyager Technologies chief executive Dylan Taylor identifying cooling as the fundamental unsolved problem. A dedicated economics session at SmallSat Europe in May 2026 marked a shift from vision-led to cost-led discussion of the category.
+
=== Material degradation ===
  
−
=== Space environment ===
+
Spacecraft surfaces experience ultraviolet radiation, atomic oxygen, charged particles and thermal cycling.
  
−
* '''Debris and collision risk''' — roughly 14,000 active satellites currently operate in LEO. A one-million-satellite constellation represents an increase of nearly two orders of magnitude, materially raising collision probability. A full [[Kessler syndrome]] cascade would be effectively irreversible on human timescales. Approximately 44,000 tracked objects larger than 10 cm are already capable of destroying a satellite.
+
These effects can reduce the performance of solar arrays, radiator coatings and other exposed materials over time.
−
* '''Astronomy''' — leading astronomers have warned that data center constellations, combined with proposed orbiting mirror projects, would severely degrade ground-based observation and the visible night sky.
 
−
* '''Reentry chemistry''' — the atmospheric effects of large-scale satellite reentry, particularly alumina and other metal oxides in the mesosphere, remain poorly characterised.
 
−
* '''Material degradation''' — unattenuated ultraviolet exposure degrades radiator surfaces, reducing performance over mission life.
 
  
 
=== Governance and sovereignty ===
 
=== Governance and sovereignty ===
  
−
Brookings and others have argued that the gap between projected capability and demonstrated capability is itself a governance risk, encouraging regulatory decisions on the basis of claims not yet validated. A distinct concern is jurisdictional: if citizen-generated data is processed in orbit, it is unclear whether sovereignty rests with the country of origin, the launching state, or the constellation operator — a question with particular force for nations that lack independent launch capability.
+
Orbital computing also presents legal and political questions.
 +
 
 +
Potential issues include:
 +
 
 +
* which jurisdiction applies to data processed aboard spacecraft;
 +
* responsibility for data originating in one country but processed by an operator registered in another;
 +
* cybersecurity and access control;
 +
* export-control restrictions;
 +
* liability for collisions or infrastructure failures;
 +
* concentration of computing capacity among countries or companies with independent launch systems.
 +
 
 +
Some analysts have also cautioned that regulators should distinguish between demonstrated capabilities and long-term capacity projections when evaluating very large proposed systems.
  
 
== Market context ==
 
== Market context ==
  
−
Space startup funding reached a record $20.3 billion in 2026, separate from SpaceX's IPO, with orbital compute emerging as a distinct investment category. Goldman Sachs published "The Second Space Age" on 13 August 2026, projecting a $1.8 trillion space economy by 2035. Earlier market estimates for the in-orbit data center segment specifically were considerably more modest, in the low billions by the end of the decade — a divergence that reflects genuine uncertainty about whether the category is a niche edge-processing market or a replacement layer for terrestrial hyperscale.
+
Orbital computing emerged during the mid-2020s as a distinct category within the broader commercial-space and artificial-intelligence infrastructure markets.
  
−
A useful analytical framing divides the field into three tiers:
+
Projects range considerably in scale.
  
−
# '''Near-term operators''' with hardware in orbit — Axiom Space, Starcloud, ADA Space/Zhejiang Lab, Kepler.
+
A useful classification is:
−
# '''Platform builders and enablers''' — Nvidia, Ramon.Space, Sophia Space, Skyloom, OrbitsEdge, HPE.
 
−
# '''Vision-led mega-architectures''' — SpaceX/xAI, Blue Origin Project Sunrise, Cowboy Space, CASC.
 
  
−
The economics may therefore resemble terrestrial cloud buildouts, in which value accrues not only to eventual orbital hyperscalers but to whoever controls bottleneck technologies such as radiation-tolerant AI compute, low-SWaP optical terminals and launch capacity.
+
# '''Operational and near-term systems''' — organizations that have flown computing hardware or deployed early orbital nodes, including Axiom Space, ADA Space/Zhejiang Lab, Starcloud and Kepler Communications.
 +
# '''Technology suppliers and infrastructure enablers''' — organizations providing processors, networking, thermal systems, radiation-tolerant hardware or software, including [[NVIDIA]], Ramon.Space, Sophia Space, Skyloom, OrbitsEdge, Hewlett Packard Enterprise and Huawei-associated open-source technologies.
 +
# '''Research programs''' — initiatives examining technical, economic or environmental feasibility, including Google Project Suncatcher, ASCEND and university-led satellite-computing programs.
 +
# '''Large proposed constellations''' — systems proposed by SpaceX, Blue Origin, Cowboy Space and other organizations that would require large numbers of dedicated spacecraft.
 +
 
 +
The eventual market structure remains uncertain. Orbital computing could develop primarily as a specialized extension of satellite edge processing, or it could expand into a larger infrastructure market if launch, thermal-control and spacecraft-manufacturing costs decline sufficiently.
  
 
== Timeline ==
 
== Timeline ==
  
 
{| class="wikitable"
 
{| class="wikitable"
−
! Date !! Event
+
! Date
−
|-
+
! Event
−
| 2008 || China begins research into space-based solar power, later relevant to orbital data center power and thermal systems
 
−
|-
 
−
| Nov 2018 || Huawei Cloud initiates and open-sources KubeEdge
 
 
|-
 
|-
−
| Dec 2021 || First Tiansuan Constellation satellite launched with KubeEdge; described as the world's first cloud-native satellite
+
| 2008
 +
| China begins research into space-based solar-power technologies later applicable to high-power orbital infrastructure
 
|-
 
|-
−
| Apr–Jul 2022 || NTT and SKY Perfect JSAT establish Space Compass
+
| November 2018
 +
| Huawei Cloud initiates and open-sources KubeEdge
 
|-
 
|-
−
| Nov 2022 || European Commission contracts Thales Alenia Space to lead the ASCEND study
+
| December 2021
 +
| Tiansuan Constellation satellite launched with a KubeEdge-based cloud-native computing stack
 
|-
 
|-
−
| Sept 2023 || Huawei releases ''Data Center 2030'' at HUAWEI CONNECT, naming space data centers a key structural pattern
+
| April–July 2022
 +
| NTT and SKY Perfect JSAT establish Space Compass
 
|-
 
|-
−
| Jun 2024 || ASCEND feasibility results published
+
| November 2022
 +
| European Commission contracts the Thales Alenia Space-led ASCEND consortium
 
|-
 
|-
−
| Sept 2024 || Starcloud (as Lumen Orbit) publishes white paper on gigawatt-scale orbital AI compute
+
| September 2023
 +
| Huawei's ''Data Center 2030'' report lists space-based data centers among several potential future infrastructure patterns
 
|-
 
|-
−
| Feb 2025 || Lonestar operates data storage hardware en route to the Moon aboard an Intuitive Machines lander
+
| June 2024
 +
| ASCEND feasibility results published
 
|-
 
|-
−
| 14 May 2025 || First 12 satellites of the Three-Body Computing Constellation launched
+
| September 2024
 +
| Lumen Orbit, later renamed Starcloud, publishes an orbital-computing white paper
 
|-
 
|-
−
| Aug–Sept 2025 || Axiom AxDCU-1 deployed aboard the ISS
+
| February 2025
 +
| Lonestar operates off-Earth data-storage hardware aboard an Intuitive Machines lunar mission
 
|-
 
|-
−
| 4 Nov 2025 || Google announces Project Suncatcher
+
| 14 May 2025
 +
| First 12 satellites of the Three-Body Computing Constellation launched
 
|-
 
|-
−
| Nov 2025 || Starcloud-1 launched; first Nvidia H100 in low Earth orbit
+
| August–September 2025
 +
| Axiom Space deploys AxDCU-1 aboard the International Space Station
 
|-
 
|-
−
| Nov 2025 || Beijing announces its space data center construction programme
+
| 4 November 2025
 +
| Google announces Project Suncatcher
 
|-
 
|-
−
| 11 Jan 2026 || Axiom's first two dedicated ODC nodes and Kepler's 10 optical relay satellites launched
+
| November 2025
 +
| Starcloud-1 launches carrying an [[NVIDIA]] H100 GPU
 
|-
 
|-
−
| Jan 2026 || Alibaba Qwen3 deployed to the Three-Body constellation; Blue Origin announces TeraWave
+
| November 2025
 +
| Beijing announces a dedicated space data-center development program
 
|-
 
|-
−
| 30 Jan 2026 || SpaceX files with the FCC for up to one million ODC satellites
+
| 11 January 2026
 +
| Axiom orbital-computing nodes and Kepler optical-network satellites launched
 
|-
 
|-
−
| 2 Feb 2026 || SpaceX completes acquisition of xAI at a combined ~$1.25 trillion valuation
+
| January 2026
 +
| Alibaba Qwen3 reported deployed to the Three-Body constellation
 
|-
 
|-
−
| Feb 2026 || Starcloud files for 88,000 satellites; Zhejiang Lab reports ten AI models running on orbit; Sophia Space raises $10M
+
| January 2026
 +
| Blue Origin announces TeraWave
 
|-
 
|-
−
| Feb 2026 || Jensen Huang publicly characterises orbital data center economics as poor today but improving
+
| 30 January 2026
 +
| SpaceX files for a proposed orbital-computing constellation of up to one million satellites
 
|-
 
|-
−
| Mar 2026 || Nvidia unveils Space-1 Vera Rubin Module at GTC; Starcloud raises $170M at $1.1B
+
| February 2026
 +
| Starcloud files for a proposed constellation of up to 88,000 satellites
 
|-
 
|-
−
| 19 Mar 2026 || Blue Origin files Project Sunrise (51,600 satellites)
+
| February 2026
 +
| Zhejiang Lab reports multiple artificial-intelligence models operating on the Three-Body constellation
 
|-
 
|-
−
| 21–22 Mar 2026 || Musk unveils Terafab; SpaceX details orbital data center spacecraft
+
| March 2026
 +
| [[NVIDIA]] announces the Space-1 Vera Rubin Module
 
|-
 
|-
−
| May 2026 || Aetherflux becomes Cowboy Space, raises $275M, files for 20,000-satellite Stampede; NASA objects to Project Sunrise
+
| 19 March 2026
 +
| Blue Origin files Project Sunrise
 
|-
 
|-
−
| Jul 2026 || Shanghai unveils the Xingshu Plan at the World AI Conference
+
| 2026
 +
| Cowboy Space develops its Stampede orbital-computing proposal
 
|-
 
|-
−
| 13 Aug 2026 || Goldman Sachs publishes "The Second Space Age"
+
| July 2026
 +
| Shanghai announces the Xingshu Plan
 
|-
 
|-
−
| 21 Aug 2026 || Starcloud raises $250M at $2.3B with Nvidia and Cisco participating
+
| August 2026
 +
| Additional private investment announced for Starcloud
 
|-
 
|-
−
| 31 Aug 2026 || Shanghai opens its Songjiang space computing hub
+
| 31 August 2026
 +
| Shanghai opens the Songjiang space-computing hub
 
|-
 
|-
−
| Early 2027 || ''Planned:'' Google/Planet Suncatcher prototype satellites; Starcloud-2
+
| Early 2027
 +
| ''Planned:'' Google/Planet Labs Suncatcher prototypes and additional commercial orbital-computing demonstrations
 
|-
 
|-
−
| 2028 || ''Planned:'' ASCEND in-orbit demonstration; Cowboy Space first 1 MW node; Starcloud Space-1 Vera Rubin flight
+
| 2028
 +
| ''Planned:'' European ASCEND demonstration and additional megawatt-class orbital-computing demonstrations
 
|}
 
|}

Latest revision as of 16:16, 5 September 2026

Youtube search... ...Google search


Transport Agnostic - capability is achieved by splitting the control/management ‘planes’ from the data ‘plane’, using software defined networking (SDN) mechanisms to dynamically redirect traffic to the best transport, as opposed to having control/management/data on the same ‘plane’. The users/edge devices delegate (are agnostic) the tasks of assigning transport addresses/routes/protocols/mechanisms are used. From a security point of view, users/edge never access the control/management ‘planes’!

Contents

Cognitive Network (CN)

In communication networks, cognitive network is a new type of data network that makes use of cutting edge technology from several research areas to solve some problems current networks are faced with. Cognitive network is different from cognitive radio as it covers all the layers of the OSI model. Cognitive Network | Wikipedia

Cognitive-Networking-1-Image.png

Instant-Network-Blog-1-768x432.png

Instant-Network-Blog-3-768x432.png

Intent-Based Networking (IBN)

  • Intelligent Automation
  • Intelligent Assurance
  • Understanding what's on the Network
  • Detecting Threats in encrypted traffic

Multiprotocol Label Switching (MPLS)

Multiprotocol Label Switching (MPLS) is a networking technology that enhances the efficiency and speed of data transmission across networks by using labels to route packets instead of traditional IP addresses. This technology is particularly useful in large networks where the traditional routing methods can become complex and inefficient. MPLS operates independently of the underlying IP addressing and routing protocols, allowing for more flexible and efficient routing of traffic.

MPLS works by assigning labels to packets, which are then used to determine the path the packet should take through the network. This label-switched path (LSP) is determined by the first device (usually a router) that processes the packet, which then forwards the packet along the LSP to its destination. This process is much faster and more efficient than traditional routing methods, which require each device in the path to perform a routing lookup for each packet.

MPLS supports a variety of protocols and technologies, including IP, ATM, and Frame Relay, and it interfaces with existing routing protocols such as RSVP and OSPF. It also provides mechanisms for traffic engineering, quality of service (QoS), and the creation of virtual private networks (VPNs) both at Layer 2 and Layer 3.

In addition to improving network performance, MPLS also offers features like traffic engineering, which allows for the optimization of network paths based on various constraints such as bandwidth availability, and the creation of VPNs that can transport different types of traffic over the same network infrastructure.

MPLS is widely used in enterprise and service provider networks to deliver advanced, value-added services over a single infrastructure. It can be integrated seamlessly with existing infrastructure and supports a wide range of platforms, making it a versatile solution for both service providers and enterprises.

For example, in a service provider network, MPLS can be used to aggregate subscribers with differing access links on an MPLS edge without changing their current environments. This allows for the delivery of a wide variety of services over a single infrastructure, including Layer 3 VPNs, Layer 2 VPNs, Traffic Engineering, QoS, GMPLS, and IPv6.

In summary, MPLS is a powerful technology that enhances network efficiency and performance by using labels to route packets, supporting a wide range of protocols and technologies, and offering advanced features like traffic engineering and VPN creation.

Software-Defined Enterprise (SDE) / Software-Defined Networking (SDN) / Software-defined Wide Area Network (SD-WAN)

Youtube search... ...Google search

Virtual network architecture that allows enterprises to leverage any combination of transport services to securely connect users to applications. SD-WAN simplifies the management and operation of a WAN by decoupling the networking hardware from its control mechanism. This concept is similar to how software-defined networking implements virtualization technology to improve data center management and operation. Wikipedia

  • WAN Optimization
  • Fault Prediction
  • Network Management
  • Security


4TH Estate Network Optimization - Defense Enclave Services (DES)
USDISA The Fourth Estate Network Optimization will modernize the DOD IT architecture, consolidate networks, reduce costs, improve business practices and mitigate operational and cyber risks. The 14 organizations onboarding between fiscal years 2020-2025 onto DoDNet.

ENCOR - SD-WAN Principles
It's time for the ENCOR 1.4 blueprint - The Principles of SD-WAN! In this video, we explore the problems with traditional WANs, why SD-WAN is a game-changer, and how Cisco's SD-WAN solution is architected.

Artificial Intelligence (AI) on Software-Defined Network (SDN)
See how DCConnect makes use of Artificial Intelligence (AI) to apply on Software-Defined Networking (SDN) so as to improve network planning and facilitate efficient network utilization. Let’s see how DCConnect helps our customers to drive their network cost down with our solution on multi-point Carrier Ethernet Orchestration.

Cisco's Intent-Based Networking and the Journey to Software Defined Networks
Expert Cisco instructor Chris Olsen explores Cisco's Intent-Based Networking and the Journey to Software Defined Networks. As organizations migrate to all-digital solutions in areas like Cloud, Mobility, and IoT, the strength of network will be more critical than ever. Cisco's Intent-Based Networks and SDN solutions will enable your organization to meet the demands of tomorrow's networks including programmability and automation. To learn more about Cisco Software-Defined Networking (SDN) Training, visit https://ter.li/okom270

Towards application aware networking
Featuring: Beth Cohen, NFV/SDN Network Product Strategy, Verizon SD-WAN is the first step towards application aware networking, but intelligent networking is still very much in its infancy. Not only do telecoms networks need to respond to the applications they run, but the applications themselves need to be better at responding to the network. Meanwhile, work continues at a rapid rate to minimize the number of NFVi architectures available and make their compatibility with VNFs more straightforward and faster for CSPs to implement. Filmed at: ONS Europe, Antwerp, Belgium, September 2019

Evolution of SDN in Google’s Network Infrastructure- Vijoy Pandey
OpenDaylight Project Evolution of Software Defined Networking in Google’s Network Infrastructure - Vijoy Pandey, Google https://sched.co/7j8X Google has long been a pioneer in distributed computing and data processing, and we’ve known that great computing infrastructure like this requires great networking technology. For the past decade we have been building our own network hardware and software to connect all of the servers in our datacenters together, and also to connect our datacenters with each other, with Software Defined Networking principles in mind. This talk walks through the evolution of Google’s SDN-based networking infrastructure, from building an SDN-based WAN (B4), to allocating wide area bandwidth amongst thousands of individual applications based on centralized policy management (BwE), to creating building-scale data center fabrics (Jupiter). We will discuss how technologies were cross leveraged in building these networks, the operational challenges faced, and the lessons learned. Vijoy Pandey is Head of Engineering ​for Data Center Fabrics and Inter Data Center Backbone Networks at Google. He ​works on SDN and data center, cloud and backbone network architectures looking out 18 months+​, and​ leads ​the ​team responsible for the design, engineering​,​ deployment ​& operations of ​these networks. Prior to Google, he was the CTO of Networking​ at IBM where he led the technical vision & system architecture for IBM System Networking. He was previously the CTO and Director Engineering of a startup, Blade Networking Technologies (BNT) which was acquired by IBM. Before that has held various leadership and management roles in switching, security, and application delivery controller companies & startups. Vijoy holds an undergraduate degree in Computer Science from the Indian Institute of Technology, and a Ph.D. in Computer Science from the University of California, Davis.

Large Scale Overlay Networks with OVN Problems and Solutions
OVN is the SDN solution provided by OVS community. In this presentation, we will have a deep dive on how we scale OVN in eBays private cloud environment to support VMs and nested workloads (k8s containers) on overlay networks. The presentation will focus on the scaling pain points of SDN and how we solved the problems by improving and tuning OVN.

Software Defined Network Virtual Lab
NEC America Many networking professionals are interested in evaluating network controllers to become familiar with the power and benefits of SDN. This video demonstrates how to setup configure and deploy a SDN virtual lab on a laptop. The SDN virtual lab consists of the NEC ProgrammableFlow Controller and the Mininet network simulation tool. With this lab, users can simulate an SDN network of various topologies and test a variety of use cases .

Explained

Software-Defined WAN (SD-WAN) - Explained
Learn what SD-WAN is, how it works, and its benefits. If you were ever curious about what SD-WAN is or how it works, then this explainer video will teach you the basics. See how it is more effective and efficient compared to older technology managed by Command Line Interfaces (CLI). SD-WAN offers greater flexibility, control, agility, and performance than traditional means of managing Wide Area Networks (WANs). Learn more: https://rvbd.ly/2zGYiel

What is software-defined networking (SDN)?
IDG TECHtalk A graphical look at the technology behind software-defined networking (SDN)

Network Functions Virtualization (NFV)

Youtube search... ...Google search

NFV or Virtual Network Function (VNF) allows network operators to manage and expand their network capabilities on demand using virtual, software based applications where physical boxes once stood in the network architecture. This makes it easier to load-balance, scale up and down, and move functions across distributed hardware resources. With continual updates, operators can keep things running on the latest software without interruption to their customers. On the road to NFV deployment | Ericsson

For example, a virtual session border controller could be deployed to protect a network without the typical cost and complexity of obtaining and installing physical network protection units. Other examples of NFV include virtualized load balancers, firewalls, intrusion detection devices and WAN accelerators. ...The NFV framework consists of three main components:

  1. Virtualized network functions (VNFs) are software implementations of network functions that can be deployed on a network functions virtualization infrastructure (NFVI).
  2. Network functions virtualization infrastructure (NFVI) is the totality of all hardware and software components that build the environment where NFVs are deployed. The NFV infrastructure can span several locations. The network providing connectivity between these locations is considered as part of the NFV infrastructure.
  3. Network functions virtualization management and orchestration architectural framework (NFV-MANO Architectural Framework) is the collection of all functional blocks, data repositories used by these blocks, and reference points and interfaces through which these functional blocks exchange information for the purpose of managing and orchestrating NFVI and VNFs.

The building block for both the NFVI and the NFV-MANO is the NFV platform. In the NFVI role, it consists of both virtual and physical processing and storage resources, and virtualization software. Network function virtualization | Wikipedia

Beyond SDN and NFV: augmenting AI to transform the network
TelecomTV Five years into NFV, and we’ve reached a point where autonomous operations must be realized: IBM has been doing just that. Featuring: Steven Teitzel, Telco Global Solution Executive, Network & OSS Transformation and Security, IBM and Anil Rao, Principal Analyst, Analysys Mason Filmed at: SDN NFV World Congress 2017, The Hague, Netherlands

Difference between SDN Vs NFV : FavouriteBlog.com
Top and Best Blog about AI/Machine Learning Deep Learning - FavouriteBlog.com Check SDN Video on other Channel.

2. Introduction to NFV Network function Virtualization Basics - NFV Architecture and ETSI - NFV MANO
https://telecomtutorial.info Covering Introduction & Tutorial for Virtualization & NFV in Telco Networks . Covering Difference between NFV & SDN or How they work . Both architectures use network abstraction, they do so differently. While NFV covers Softwarization , Virtualization and makes building blocks ready , SDN forwards data packets from one network device to another. At the same time, SDN's networking control functions for routing & policy definition NFV : Network function Virtualization : Introduction & Basics Why we need NFV & Features of NFV ETSI Framework & Specs • NFV Architecture Need of SDN Connect Me @ Linkedin : www.linkedin.com/in/vikas-shokeen Music by Joakim Karud https://youtube.com/joakimkarud Free HD Stock-Footage and Motion Graphics by CyberWebFX : https://www.youtube.com/c/CyberWebFX

Dynamic Service Chaining for SDN NFV
Kishore Inampudi, A10 Networks In talk focus on framework for dynamic L4-L7 services in NFV/SDN environments. The modern service provider and data center networks demand cloud delivery model for agile and cost-effective rollout of services for revenue generation. There are some proposals to incorporate service insertion in the emerging SDN and NFV environments. However, the current methods are sub-optimal, complex and inflexible in delivering automated end-to-end service delivery. In a virtualized cloud environment, service delivery requires intelligence in the network for policy-based traffic handling and differentiated services. Dynamic service chaining is a fundamental component in building an on-demand and scalable model for policy enforcement. This session will cover following topics: 1. NFV & Service chaining use cases 2. Challenges with existing solutions 3. Opportunities with dynamic service chaining 4. Policy Enforcement model in SDN/NFV environment 5. Dynamic Service Chaining Architecture & Design considerations 6. Benefits of Dynamic Service Chaining

Space-based Data Centers

Space-based data centers (also orbital data centers, ODCs, or orbital compute) are proposed or operational computing facilities deployed in low Earth orbit (LEO) or beyond, in which spacecraft perform data processing, storage, artificial-intelligence inference or training, or related computing functions rather than acting only as communications relays or sensors.

Interest in orbital computing increased during the mid-2020s as demand for artificial-intelligence infrastructure grew, satellite operators sought alternatives to transmitting large volumes of raw data to Earth, and governments and companies investigated new approaches to energy supply, cooling, data sovereignty and distributed computing. The field includes aerospace companies, cloud and semiconductor firms, telecommunications operators, universities, government research programs and venture-backed startups.

Approaches range from relatively small edge-computing payloads that process satellite data before downlink, to proposed constellations containing thousands or tens of thousands of dedicated computing spacecraft. As of September 2026, most large-scale orbital data-center concepts remained at the research, demonstration, regulatory-filing or early deployment stage.

Rationale

Several arguments are commonly advanced for space-based computing.

Power
Spacecraft in suitable orbits can make extensive use of solar power without atmospheric attenuation or terrestrial weather. Dawn–dusk sun-synchronous orbits can also reduce the duration of eclipse periods, potentially lowering energy-storage requirements. The practical advantage depends on orbital design, solar-array efficiency, spacecraft mass and other engineering constraints.
Cooling
Spacecraft do not require terrestrial chilled-water or air-conditioning systems, and therefore do not consume water for cooling. However, waste heat in vacuum must ultimately be rejected through thermal radiation. This requires radiator area and thermal-control hardware, making heat rejection one of the principal engineering constraints for high-power orbital computing.
Data processing and downlink
Earth-observation and communications satellites can generate substantially more raw data than is practical to transmit continuously to the ground. Processing data on orbit allows spacecraft to return selected information such as detections, classifications, compressed products or alerts rather than transmitting all raw sensor data.
Latency
Locating computing resources closer to satellites and other space-based sensors can reduce the time required to process information for applications such as disaster monitoring, maritime surveillance, astronomy and autonomous spacecraft operations.
Sovereignty and resilience
Some government programs describe orbital computing as part of broader digital-sovereignty or infrastructure-resilience strategies. These proposals also raise unresolved questions concerning jurisdiction, data governance and responsibility for infrastructure operating outside national territory.

Development approaches

Orbital-computing projects generally fall into several overlapping categories:

  • Onboard edge computing — processing sensor data directly aboard an individual spacecraft.
  • Distributed satellite computing — connecting multiple spacecraft through optical or radio inter-satellite links so they can share computing tasks.
  • Dedicated orbital data-center nodes — spacecraft designed primarily to provide computing or storage capacity.
  • Large-scale orbital compute constellations — proposed fleets intended to provide data-center-scale or hyperscale computing capacity.
  • Enabling technologies — radiation-tolerant processors, optical networking, thermal-management systems, software orchestration and launch systems used by orbital-computing operators.

```mediawiki

Artificial intelligence and orbital computing

Artificial intelligence is a major proposed workload for space-based data centers, particularly for machine learning inference, Earth-observation analysis and, in some projects, model training. AI workloads are relevant to orbital computing because satellites can generate large volumes of sensor data that may be processed locally rather than transmitted in full to terrestrial data centers.

On-orbit inference

Near-term applications have focused primarily on AI inference, in which previously trained models process imagery, sensor measurements or other data aboard a spacecraft.

Potential applications include:

  • identifying objects or changes in Earth-observation imagery;
  • weather and environmental monitoring;
  • maritime and infrastructure surveillance;
  • disaster detection and response;
  • astronomical data processing;
  • autonomous spacecraft operations;
  • filtering or prioritizing data before transmission to Earth.

Processing data on orbit can reduce downlink requirements by transmitting classifications, detections or other derived results instead of complete raw datasets.

The Three-Body Computing Constellation, developed by ADA Space and Zhejiang Lab, has reported operating multiple AI models in orbit, including large remote-sensing and astronomy models. Zhejiang Lab has also reported experiments in which tasks are distributed between multiple interconnected satellites.

The Tiansuan Constellation, led by the Beijing University of Posts and Telecommunications with participation from several research and industry partners, has demonstrated collaborative inference between spacecraft and ground systems. Experiments associated with KubeEdge and related software reported reductions in the amount of data transmitted to Earth by performing preliminary processing aboard the satellite.

Other companies and research programs, including Axiom Space, NVIDIA partners and satellite edge-computing developers, have similarly investigated onboard AI inference for geospatial and autonomous applications.

Distributed AI computing

Some orbital-computing architectures propose linking multiple satellites through high-bandwidth optical inter-satellite links so that computing tasks can be distributed across several spacecraft.

This approach differs from conventional onboard processing, in which each satellite operates largely independently. A distributed orbital cluster could theoretically divide a large AI workload across multiple computing nodes in a manner conceptually similar to terrestrial distributed-computing systems.

Projects investigating this model include the Three-Body Computing Constellation, Google's Project Suncatcher and several proposed commercial orbital data-center constellations.

Such systems require:

  • high-bandwidth and low-latency inter-satellite networking;
  • distributed workload scheduling;
  • synchronization between computing nodes;
  • fault tolerance for interrupted links or spacecraft failures;
  • efficient movement of model parameters and intermediate data;
  • coordinated thermal and power management.

The performance of distributed AI workloads in orbit remains an active area of research.

AI model training

AI training is substantially more computationally and energetically demanding than inference and is therefore a longer-term objective for most orbital data-center projects.

Starcloud reported conducting an AI model-training experiment aboard its Starcloud-1 satellite after launching an NVIDIA H100 GPU to low Earth orbit in November 2025. Several larger proposed orbital-computing systems have been designed around future GPU or accelerator clusters capable of supporting more computationally intensive workloads.

Google's Project Suncatcher is investigating whether clusters of satellites equipped with Tensor Processing Units could eventually operate distributed machine-learning workloads in orbit. Its research has examined processor radiation tolerance, optical interconnects and the launch-cost reductions that would be required for orbital systems to approach the economics of terrestrial AI infrastructure.

Other proposed systems from companies including SpaceX, Starcloud and Cowboy Space have described large-scale accelerator deployments, although most such architectures remained at the proposal or development stage as of 2026.

AI hardware

Artificial-intelligence workloads require processors capable of performing large numbers of parallel matrix and tensor operations.

Orbital AI projects have therefore investigated several hardware approaches:

  • terrestrial data-center GPUs adapted for space operation;
  • purpose-built or modified AI accelerators;
  • radiation-tolerant edge-computing processors;
  • combinations of lightweight onboard processors and higher-performance ground systems.

NVIDIA has supplied or proposed hardware for several orbital-computing programs. An NVIDIA H100 was flown aboard Starcloud-1, while the company has announced the Space-1 Vera Rubin Module for future orbital data-center applications. NVIDIA's Jetson and IGX platforms have also been associated with lower-power edge-AI applications in space.

Google is investigating its own Tensor Processing Units for Project Suncatcher.

Chinese research programmes have used a mixture of commercial processors, domestic AI hardware and software frameworks including MindSpore. Huawei-associated technologies have participated primarily at the software and edge-orchestration layer through KubeEdge, Sedna and related research rather than through a dedicated publicly announced orbital AI accelerator.

Other suppliers, including Ramon.Space, OrbitsEdge and Hewlett Packard Enterprise, are developing or adapting computing systems intended to operate reliably in the space radiation environment.

Software and orchestration

AI workloads aboard satellites require software capable of remotely deploying, updating and coordinating models after launch.

Traditional spacecraft software is often designed around a fixed mission and payload. Cloud-native approaches instead seek to treat spacecraft computing resources as remotely configurable infrastructure.

Technologies investigated for this purpose include:

  • containers and Kubernetes-derived orchestration;
  • remote model deployment;
  • over-the-air software updates;
  • federated learning;
  • distributed inference;
  • workload migration between spacecraft and ground systems.

The KubeEdge-based Tiansuan experiments demonstrated elements of this model by allowing onboard AI workloads to be updated and coordinated with terrestrial systems.

Similar software-defined approaches are being considered by other orbital-computing projects, although there is not yet a common industry standard for managing large distributed AI clusters in space.

Limitations

AI workloads intensify several of the broader technical challenges associated with orbital data centers.

Power consumption
Modern AI accelerators can consume hundreds of watts per processor, while large training clusters can require megawatts of electrical power. Providing comparable power in orbit requires large solar arrays, energy storage and power-distribution systems.
Heat rejection
Almost all electrical power consumed by computing hardware ultimately becomes heat. High-performance AI processors therefore increase the radiator area and thermal-control requirements of an orbital facility.
Radiation
Advanced processors and high-bandwidth memory can be vulnerable to radiation-induced errors. Shielding, redundancy and fault-tolerant software can reduce this risk but add mass, cost or complexity.
Hardware replacement
AI accelerators evolve more rapidly than conventional satellite platforms. A spacecraft designed to remain in service for a decade or more may contain computing hardware that becomes commercially outdated within several years.
Networking
Distributed AI training can require extremely high communication bandwidth between processors. Reproducing terrestrial data-center interconnect performance using optical links between moving spacecraft remains technically challenging.
Economics
Large orbital AI clusters depend heavily on launch costs, spacecraft manufacturing costs and the ability to operate computing hardware reliably without conventional maintenance.

For these reasons, near-term orbital AI applications are generally more practical for onboard inference and data reduction than for replacing terrestrial hyperscale AI training facilities. Whether large-scale training becomes economically competitive depends on advances in launch systems, power generation, thermal management, radiation tolerance and optical networking. ```

United States

Starcloud

Starcloud of Redmond, Washington, formerly known as Lumen Orbit, was founded in 2024 and participated in Y Combinator's Summer 2024 cohort. The company has proposed dedicated satellites for artificial-intelligence computing and published concepts for eventually scaling orbital computing into the gigawatt range.

  • Starcloud-1 — a 60 kg satellite launched by SpaceX in November 2025 carrying an Nvidia H100 data-center GPU. Starcloud reported using the system to train an artificial-intelligence model in orbit.
  • FCC filing — filed in February 2026 for a constellation of up to 88,000 satellites in sun-synchronous dusk–dawn orbits between approximately 600 and 850 km.
  • Funding — the company announced a $170 million Series A financing at a reported $1.1 billion valuation in March 2026. A $250 million extension announced on 21 August 2026 reportedly valued the company at $2.3 billion and included NVIDIA and Cisco Investments among participating investors.
  • Starcloud-2 — a planned 450 kg spacecraft using NVIDIA Blackwell-generation GPUs and intended to support early commercial workloads.
  • Starcloud-3 — a proposed approximately 3-tonne, 200 kW spacecraft intended for deployment using higher-capacity launch vehicles.
  • Long-term concepts — Starcloud has described much larger spacecraft and constellation architectures intended to provide multi-gigawatt aggregate computing capacity.

Starcloud has also worked with NVIDIA on the proposed Space-1 Vera Rubin Module. The company's larger architectures depend heavily on substantial reductions in launch cost and increases in available payload capacity.

Nvidia

Nvidia participates primarily as a semiconductor and computing-platform supplier rather than as an orbital data-center operator.

At GTC in March 2026, NVIDIA announced space-oriented computing products and partnerships, including:

  • Space-1 Vera Rubin Module — a purpose-built computing module proposed for orbital data-center applications using NVIDIA's Vera Rubin architecture.
  • IGX Thor and Jetson Orin — edge-computing platforms intended for applications including geospatial intelligence, autonomous operations and onboard inference.

Organizations publicly associated with NVIDIA's space-computing efforts have included Axiom Space, Cowboy Space, Kepler Communications, Planet Labs, Sophia Space and Starcloud.

NVIDIA chief executive Jensen Huang stated in February 2026 that orbital data-center economics remained unfavorable at that time but could improve as launch and infrastructure technologies developed. He also identified heat rejection as an important engineering limitation because spacecraft cannot rely on atmospheric airflow for cooling.

SpaceX and xAI

SpaceX has proposed large-scale orbital computing as one possible application of future high-capacity launch systems and satellite platforms.

  • FCC filing — on 30 January 2026, SpaceX sought authority for a proposed system containing up to one million orbital data-center satellites operating between approximately 500 and 2,000 km.
  • xAI — SpaceX completed an acquisition of xAI on 2 February 2026 in an all-stock transaction. Elon Musk associated the combined company's long-term strategy with space-based artificial-intelligence infrastructure.
  • Manufacturing proposals — SpaceX, Tesla and xAI have discussed large-scale semiconductor and computing-hardware manufacturing intended to support both terrestrial and orbital applications.
  • Spacecraft concepts — publicly discussed designs include smaller orbital-compute satellites and larger high-power spacecraft intended to make use of future launch capacity.

These systems remain proposals, and their economics depend on launch cost, spacecraft manufacturing, power generation, thermal control and regulatory approval.

Blue Origin

Blue Origin has proposed orbital-computing and communications systems associated with two programs.

Project Sunrise
An FCC application filed in March 2026 proposed up to 51,600 data-center satellites in sun-synchronous orbits between approximately 500 and 1,800 km. The architecture relies substantially on optical inter-satellite communications.
TeraWave
A proposed 5,408-satellite communications constellation announced in January 2026. It is intended to provide connectivity for terrestrial and space-based customers and could also provide networking infrastructure for orbital-computing systems.

Blue Origin's access to the New Glenn launch vehicle provides a potential vertically integrated launch capability. Project Sunrise has also attracted regulatory objections concerning orbital congestion and effects on astronomy.

Google — Project Suncatcher

Google Research announced Project Suncatcher on 4 November 2025 as a research program investigating constellations of solar-powered satellites carrying Google Tensor Processing Units connected through free-space optical links.

The reference architecture described in accompanying research included:

  • approximately 81 satellites operating as a closely coordinated cluster;
  • an orbit near 640 km;
  • testing of Trillium-generation TPU hardware under proton radiation;
  • optical links intended to support distributed machine-learning workloads;
  • analysis of launch-cost thresholds required for economic competitiveness.

The research concluded that the concept was not excluded by fundamental physical constraints but identified thermal management, high-bandwidth communications, launch economics and long-term hardware reliability as major engineering challenges.

Google has partnered with Planet Labs on two prototype satellites planned for launch in early 2027 to test TPU hardware and optical inter-satellite networking.

Axiom Space

Axiom Space has followed a smaller-scale, incremental approach centered on computing hardware in low Earth orbit.

  • AxDCU-1 — a data-processing prototype using Red Hat Device Edge, launched to the International Space Station in August 2025.
  • AxODC Node — an orbital-computing system developed with partners including Spacebilt, Skyloom, Phison Electronics and Microchip Technology.
  • ODC Nodes 1 and 2 — dedicated orbital data-center nodes launched to LEO in January 2026 on Kepler Communications spacecraft.
  • Optical communications — integration with Kepler Communications and Skyloom relay systems.
  • ODC T1 — a larger server module proposed for launch by 2027, followed by later deployments associated with Axiom Station.

Axiom has described a gradual progression from kilowatt-scale systems toward larger orbital-computing installations.

Cowboy Space

Cowboy Space Corporation, formerly Aetherflux, was founded in 2024 by Robinhood co-founder Baiju Bhatt. The company initially focused on space-based solar power before expanding into orbital computing.

Its proposed Stampede architecture combines launch and computing infrastructure by using a launch-vehicle stage as part of the deployed orbital system.

The associated FCC proposal describes:

  • dawn–dusk sun-synchronous orbits between approximately 700 and 1,000 km;
  • spacecraft in the approximately 20,000–25,000 kg class;
  • around 1 MW of usable electrical power per spacecraft;
  • several hundred GPU modules per spacecraft;
  • optical inter-satellite networking;
  • an initial megawatt-class orbital node targeted for the late 2020s.

Cowboy Space has also entered into a Space Act Agreement with NASA's Stennis Space Center for propulsion-related testing.

Other North American and US-associated ventures

Organization Approach Status or role
Orbital Compute Proposed modular satellites providing distributed orbital computing capacity FCC filing; first launch targeted for 2027
Sophia Space TILE (Thermal-Integrated LEO Edge) architecture combining solar generation, computing and radiative cooling Orbital demonstration targeted for the late 2020s
Lonestar Data Holdings Off-planet data storage and infrastructure resilience, including lunar and cislunar systems Has operated storage hardware on a lunar mission; additional spacecraft planned
OrbitsEdge Radiation-protected enclosures intended to host commercial data-center computing hardware aboard spacecraft Demonstration program under development
Kepler Communications Optical relay and computing backbone Initial optical-network spacecraft deployed
Skyloom Global / Spacebilt Optical communications and orbital-platform integration Hardware associated with operational and planned orbital systems
Ramon.Space Radiation-tolerant computing systems Established supplier
Hewlett Packard Enterprise Spaceborne computing aboard the ISS and participation in orbital-computing research programs Long-running flight heritage

China

Three-Body Computing Constellation

The Three-Body Computing Constellation (三体计算星座), developed by ADA Space (国星宇航) with Zhejiang Lab, is an orbital computing constellation designed to perform distributed artificial-intelligence and scientific-processing workloads.

Twelve satellites were launched on a Long March 2D from Jiuquan on 14 May 2025.

Reported characteristics include:

  • approximately 5 POPS of combined computing capacity in the initial group;
  • approximately 30 TB of onboard storage;
  • individual satellite computing capacity of approximately 744 TOPS;
  • 100 Gbit/s inter-satellite laser links;
  • deployment of multiple artificial-intelligence models for remote sensing and astronomy.

The program forms part of the broader Star-Compute Program, which has described plans for a substantially larger constellation.

In January 2026, Alibaba's Qwen3 model was reported to have been deployed to the constellation for distributed processing experiments. Other commercial participants have included Alibaba Cloud, Kepu Cloud and iSoftStone.

Beijing Astro-Future Institute

The Beijing Astro-Future Institute of Space Technology has proposed a dedicated space data-center program supported by a broader industry consortium.

Its published development roadmap includes:

  1. 2025–2027 — demonstration of power supply, thermal management and onboard processing;
  2. 2028–2030 — development of orbital assembly and systems capable of processing data originating on Earth;
  3. 2031–2035 — larger modular orbital data-center infrastructure using mass production and on-orbit assembly.

Longer-term concepts include interconnected high-power facilities in sun-synchronous orbit.

State and academic programs

Several Chinese institutions have separately pursued orbital-computing research.

  • CASC — the China Aerospace Science and Technology Corporation has discussed space-based digital and intelligent infrastructure as part of planning for the 2026–2030 period.
  • Xingshu Plan (星枢计划) — a Shanghai initiative associated with Fudan University, announced during the 2026 World AI Conference. The proposed program progresses from experimental computing and edge satellites toward a larger on-demand orbital-computing service.
  • Tiansuan Constellation (天算星座) — an open in-orbit research platform initiated by the Beijing University of Posts and Telecommunications with commercial satellite partner Spacety.
  • CAICT — the China Academy of Information and Communications Technology has established an industry committee concerned with space-computing standards.

Research in China on space-based solar power and high-power spacecraft systems also overlaps technically with some of the power-generation and thermal-management requirements of orbital computing.

Huawei, KubeEdge and cloud-native satellite research

Huawei has participated in orbital-computing research principally through data-center forecasting, edge-computing software and collaborative satellite research rather than through a publicly announced dedicated orbital data-center constellation.

Huawei's Data Center 2030 report, released at HUAWEI CONNECT 2023, identified underwater and space-based data centers among several possible future infrastructure patterns. The same report also discussed larger computing clusters and lightweight edge infrastructure.

Huawei Cloud initiated the KubeEdge project in 2018 as a Kubernetes-based edge-computing framework and later contributed it to the Cloud Native Computing Foundation. Related technologies have included the Sedna edge-AI project and the MindSpore machine-learning framework.

Huawei Cloud subsequently participated as a co-construction partner in the Tiansuan Constellation, an experimental satellite research program led by the Beijing University of Posts and Telecommunications with Spacety.

A satellite launched from Jiuquan Satellite Launch Center in December 2021 carried a reconstructed KubeEdge software stack and was described by the project as a cloud-native satellite.

Experiments associated with the program reported:

  • collaborative inference between satellite and ground-based computing systems;
  • substantial reduction in the volume of data transmitted to Earth;
  • over-the-air updating of onboard artificial-intelligence models;
  • experiments involving incremental and federated learning.

Research using related Chinese computing stacks has also examined commercial off-the-shelf CPUs and GPUs, virtualized workloads, optical inter-satellite links and distributed computing.

Huawei has not publicly announced a dedicated orbital data-center constellation or a purpose-built space-qualified accelerator directly comparable with some specialized orbital-computing hardware proposed by other vendors.

Europe

ASCEND

ASCEND (Advanced Space Cloud for European Net zero emission and Data sovereignty) is a European Commission feasibility study funded through Horizon Europe.

The project was contracted to a consortium led by Thales Alenia Space in 2022 and launched in 2023. Participants have included organizations working in environmental analysis, cloud computing, launch systems and spacecraft engineering, including Carbone 4, VITO, Orange Business, CloudFerro, Hewlett Packard Enterprise, ArianeGroup, the German Aerospace Center, Airbus Defence and Space and Thales Alenia Space.

Results published in June 2024 concluded that space-based data centers could be technically feasible and potentially provide environmental benefits under certain assumptions.

The study identified several conditions and findings:

  • launch emissions would need to decrease substantially for a large orbital data-center system to provide a net environmental benefit;
  • orbital computing would avoid terrestrial water consumption for data-center cooling;
  • modular systems could potentially be assembled robotically in orbit;
  • economic feasibility depends strongly on launch, manufacturing and operations costs.

The program has discussed a long-term target of approximately 1 GW of orbital capacity by 2050.

A smaller in-orbit demonstration intended to test European orbital data-center technologies has been targeted for 2028.

ASCEND is framed in part around the European Green Deal, environmental impact and European digital sovereignty.

Japan

Space Compass

Space Compass Corporation is a joint venture between NTT and SKY Perfect JSAT. The partnership was agreed in 2022 following earlier collaboration between the companies.

Its proposed Space Integrated Computing Network combines several infrastructure layers:

  • optical relay links for transferring satellite data;
  • orbital computing systems capable of processing data before it is returned to Earth;
  • NTT's IOWN all-photonics networking technology;
  • high-altitude platform station services.

The architecture emphasizes an integrated communications and computing network rather than a standalone hyperscale orbital data center.

Space Compass has participated in Japanese government programs for next-generation optical communications and space infrastructure.

Other regions and international programs

India

SkyServe develops software for orbital edge computing and has worked with NASA's Jet Propulsion Laboratory on experiments involving artificial-intelligence processing aboard commercial spacecraft.

European Space Agency

In addition to its involvement in studies associated with ASCEND and related environmental analysis, the European Space Agency has supported earlier onboard-computing demonstrations.

The PhiSat-1 CubeSat, developed with partners including Intel and Ubotica, demonstrated artificial-intelligence inference for Earth-observation imagery.

Enabling technologies

Optical networking

High-bandwidth optical inter-satellite links are a central component of many orbital-computing proposals because distributed workloads can require communication rates substantially above those traditionally used for spacecraft command and telemetry.

Optical networking is being developed or supplied by organizations including Kepler Communications, Skyloom, Space Compass and multiple satellite manufacturers.

Radiation tolerance

Commercial processors used in terrestrial data centers are not normally designed for the radiation environment encountered in orbit.

Approaches include:

  • radiation-hardened processors;
  • shielding conventional commercial processors;
  • error-correcting memory;
  • redundant computing systems;
  • fault-tolerant software;
  • replacement or refresh cycles shorter than those of conventional communications satellites.

The cost and performance penalty associated with radiation tolerance remains an important economic variable.

Software orchestration

Cloud-native and container-based software has been proposed as a means of allowing spacecraft workloads to be updated after launch rather than being fixed for the entire mission.

Projects including KubeEdge and other edge-computing frameworks have demonstrated elements of this approach.

Launch systems

Launch cost is one of the largest variables affecting orbital-computing economics.

Large proposed systems assume some combination of:

  • reusable launch vehicles;
  • high launch cadence;
  • substantially lower cost per kilogram;
  • standardized spacecraft manufacturing;
  • large payload capacity.

Some orbital data-center concepts therefore depend on launch systems that had not yet demonstrated the required operating economics as of 2026.

Regulatory landscape

Orbital data centers do not fit neatly into traditional regulatory categories for communications, navigation or Earth-observation spacecraft.

Their use of optical inter-satellite links can reduce radio-spectrum requirements relative to communications constellations, but large proposed satellite populations raise issues involving orbital congestion, debris mitigation, astronomy and licensing.

Selected constellation filings

Applicant System Proposed satellites Filing period Approximate altitude / orbit
SpaceX Orbital data-center constellation Up to 1,000,000 January 2026 500–2,000 km
Starcloud Orbital computing constellation Up to 88,000 February 2026 600–850 km, sun-synchronous
Blue Origin Project Sunrise Up to 51,600 March 2026 500–1,800 km, sun-synchronous
Blue Origin TeraWave 5,408 January 2026 Multiple orbits
Cowboy Space Stampede Data Center System Up to 20,000 2026 700–1,000 km, sun-synchronous
Orbital Compute Orbital computing constellation Approximately 100,000 2026 LEO

Regulatory filings represent proposed systems rather than deployment approvals.

Large constellation proposals have drawn objections and requests for additional review from government agencies, competing operators and organizations concerned with orbital safety and astronomy.

Criticism and open questions

Thermal management

Heat rejection is one of the most frequently discussed engineering challenges for orbital data centers.

In vacuum, heat must ultimately be emitted through thermal radiation. High-power computing systems therefore require substantial radiator area and thermal-control infrastructure.

The International Space Station, for example, uses large radiator systems to reject tens of kilowatts of heat.

Engineering studies of data-center-class GPUs in orbit have produced radiator requirements ranging from approximately square-metre scale for individual accelerators to much larger areas for multi-kilowatt racks.

The required area depends on factors including:

  • operating temperature;
  • radiator orientation;
  • solar exposure;
  • infrared radiation from Earth;
  • surface degradation;
  • thermal-transfer efficiency;
  • spacecraft architecture.

Higher computing density can reduce spacecraft volume but makes heat transport more difficult. Conversely, spreading components across a larger structure increases mass, interconnect distance and shielding requirements.

Supporters of orbital computing argue that thermal-control engineering is well established at conventional spacecraft power levels, while critics question how efficiently it can scale to data-center or gigawatt-class systems.

Economics

Launch cost is generally considered a major determinant of whether large orbital data centers can compete with terrestrial infrastructure.

Published concepts commonly assume substantial reductions in launch cost compared with conventional expendable or partially reusable launch systems.

Additional economic issues include:

Hardware obsolescence
Artificial-intelligence processors can become commercially outdated within several years, substantially faster than the design lifetime of many conventional satellites.
Radiation qualification
Radiation-tolerant computing hardware can cost more or provide lower performance than equivalent terrestrial hardware.
Maintenance
Terrestrial data centers allow technicians to replace failed components. Distributed orbital systems require redundancy, remote reconfiguration, robotic servicing or replacement spacecraft.
Manufacturing
Constellations containing thousands of high-power computing spacecraft would require production volumes substantially beyond those of most existing satellite systems.
Launch dependence
Many proposed business models depend on future reusable launch systems achieving cost and flight-rate targets that had not yet been demonstrated at full commercial scale as of 2026.

Economic estimates published by project developers vary widely and should generally be interpreted as projections rather than demonstrated operating costs.

Orbital debris

Large orbital-computing constellations could significantly increase the number and mass of spacecraft in low Earth orbit.

Concerns include:

  • collision probability;
  • debris generation;
  • conjunction-management capacity;
  • spacecraft failures;
  • end-of-life disposal;
  • cascading collision scenarios associated with the Kessler syndrome.

The scale of some proposals has therefore prompted calls for more detailed regulatory and environmental assessment before deployment.

Astronomy

Astronomers have raised concerns that very large constellations could affect optical and infrared observations through reflected sunlight, thermal emissions or radio interference.

Large spacecraft equipped with extensive solar arrays or radiator structures could present different observational effects from smaller communications satellites.

Atmospheric effects of reentry

Large-scale satellite replacement would produce repeated atmospheric reentries.

Researchers have studied possible effects from aluminium oxides and other materials deposited into the upper atmosphere, although the long-term environmental consequences remain uncertain.

Material degradation

Spacecraft surfaces experience ultraviolet radiation, atomic oxygen, charged particles and thermal cycling.

These effects can reduce the performance of solar arrays, radiator coatings and other exposed materials over time.

Governance and sovereignty

Orbital computing also presents legal and political questions.

Potential issues include:

  • which jurisdiction applies to data processed aboard spacecraft;
  • responsibility for data originating in one country but processed by an operator registered in another;
  • cybersecurity and access control;
  • export-control restrictions;
  • liability for collisions or infrastructure failures;
  • concentration of computing capacity among countries or companies with independent launch systems.

Some analysts have also cautioned that regulators should distinguish between demonstrated capabilities and long-term capacity projections when evaluating very large proposed systems.

Market context

Orbital computing emerged during the mid-2020s as a distinct category within the broader commercial-space and artificial-intelligence infrastructure markets.

Projects range considerably in scale.

A useful classification is:

  1. Operational and near-term systems — organizations that have flown computing hardware or deployed early orbital nodes, including Axiom Space, ADA Space/Zhejiang Lab, Starcloud and Kepler Communications.
  2. Technology suppliers and infrastructure enablers — organizations providing processors, networking, thermal systems, radiation-tolerant hardware or software, including NVIDIA, Ramon.Space, Sophia Space, Skyloom, OrbitsEdge, Hewlett Packard Enterprise and Huawei-associated open-source technologies.
  3. Research programs — initiatives examining technical, economic or environmental feasibility, including Google Project Suncatcher, ASCEND and university-led satellite-computing programs.
  4. Large proposed constellations — systems proposed by SpaceX, Blue Origin, Cowboy Space and other organizations that would require large numbers of dedicated spacecraft.

The eventual market structure remains uncertain. Orbital computing could develop primarily as a specialized extension of satellite edge processing, or it could expand into a larger infrastructure market if launch, thermal-control and spacecraft-manufacturing costs decline sufficiently.

Timeline

Date Event
2008 China begins research into space-based solar-power technologies later applicable to high-power orbital infrastructure
November 2018 Huawei Cloud initiates and open-sources KubeEdge
December 2021 Tiansuan Constellation satellite launched with a KubeEdge-based cloud-native computing stack
April–July 2022 NTT and SKY Perfect JSAT establish Space Compass
November 2022 European Commission contracts the Thales Alenia Space-led ASCEND consortium
September 2023 Huawei's Data Center 2030 report lists space-based data centers among several potential future infrastructure patterns
June 2024 ASCEND feasibility results published
September 2024 Lumen Orbit, later renamed Starcloud, publishes an orbital-computing white paper
February 2025 Lonestar operates off-Earth data-storage hardware aboard an Intuitive Machines lunar mission
14 May 2025 First 12 satellites of the Three-Body Computing Constellation launched
August–September 2025 Axiom Space deploys AxDCU-1 aboard the International Space Station
4 November 2025 Google announces Project Suncatcher
November 2025 Starcloud-1 launches carrying an NVIDIA H100 GPU
November 2025 Beijing announces a dedicated space data-center development program
11 January 2026 Axiom orbital-computing nodes and Kepler optical-network satellites launched
January 2026 Alibaba Qwen3 reported deployed to the Three-Body constellation
January 2026 Blue Origin announces TeraWave
30 January 2026 SpaceX files for a proposed orbital-computing constellation of up to one million satellites
February 2026 Starcloud files for a proposed constellation of up to 88,000 satellites
February 2026 Zhejiang Lab reports multiple artificial-intelligence models operating on the Three-Body constellation
March 2026 NVIDIA announces the Space-1 Vera Rubin Module
19 March 2026 Blue Origin files Project Sunrise
2026 Cowboy Space develops its Stampede orbital-computing proposal
July 2026 Shanghai announces the Xingshu Plan
August 2026 Additional private investment announced for Starcloud
31 August 2026 Shanghai opens the Songjiang space-computing hub
Early 2027 Planned: Google/Planet Labs Suncatcher prototypes and additional commercial orbital-computing demonstrations
2028 Planned: European ASCEND demonstration and additional megawatt-class orbital-computing demonstrations
Retrieved from "https://primo.ai/index.php?title=Computer_Networks&oldid=43465"