Orbital data center

Abstract

An orbital data center system (system) can be configured to provide data center services in an orbit around Earth. The system can include: a power source comprising one or more solar power arrays; computing hardware with an aggregate peak power consumption of at least two kilowatts; an enclosure containing the computing hardware; communications links configured for supporting wireless communications to and from the system; a thermal management system for regulating temperature of the computing hardware using mechanically pumped coolant fluid; and a heat radiating structure for radiating heat from the thermal management system away from the satellite; wherein the system is configured to orbit the earth in an orbit with substantially continuous sun exposure.

Description

Starcloud-03W002TITLE OF THE INVENTIONOrbital Data CenterRELATED APPLICATIONS

[0001] The subject matter of this application is related to U.S. Provisional Application No. 63689861, filed on 2024-09-03, which is hereby incorporated by reference in its entirety.BACKGROUND OF THE INVENTION

[0002] Large-scale orbital data centers represent one of the most ambitious space projects ever conceived. They leverage three fundamental advantages over terrestrial facilities: abundant solar energy, efficient radiative cooling to deep space, and the ability to scale in size and distribution without physical or permitting constraints. Falling launch costs and expanding in-space connectivity, combined with surging demand for energy-hungry artificial intelligence (Al) clusters, make such systems not only technically feasible but increasingly necessary to realize the full potential of AL Earth-based data centers face critical limitations when scaled to gigawatt levels, including energy supply, cooling efficiency, and land use. Orbital deployment addresses these challenges by providing inexpensive solar power without the intermittency of terrestrial solar farms, direct access to low coolant temperatures via radiative cooling, and modular architectures that enable nearindefinite scaling. This approach offers both economic and environmental benefits while avoiding the bottlenecks of Earth-based infrastructure. Although the engineering challenges are substantial, no insurmountable obstacles have been identified. Orbital datacenters thus present a novel, viable pathway to sustain the rapid growth of Al toward artificial general intelligence while reducing terrestrial impact.SUMMARY OF THE INVENTION

[0003] This disclosure relates to an orbital data center configured for continuous or near continuous sunlight in earth orbit with the following basic design principles. These are all in service of creating a low-cost, high-value, future-proofed data center.Modularity: The datacenter should make up multiple modules which can be manufactured at scale and assembled either on the ground or in orbit. The requirements for each designStarcloud-03W002 element may evolve independently as needed. Containers may have different compute abilities over time.Maintainability: Old parts and containers should be easy to replace without redesigning the whole data center. The data center should not need retiring for at least 10 years. Minimize moving parts and critical failure points: Reducing as much as reasonably possible connectors, mechanical actuators, latches, and other moving parts.Design resiliency: Single points of failure should be minimized, and any failures should result in graceful degradation of performance.Incremental scalability: Able to scale the number of containers from one to N, maintaining profitability from the very first container and not requiring large CapEx jumps at any one point.

[0004] In one embodiment, a satellite includes a power source from modified terrestrial silicon solar arrays; a heat radiating structure; and a sealed enclosure physically locating the computing hardware to make up a data center thermally coupled to the heat radiating structure.

[0005] An orbital data center system (system) can be configured to provide data center services in an orbit around Earth. The system can include: a power source comprising one or more solar power arrays; computing hardware with an aggregate peak power consumption of at least two kilowatts; an enclosure containing the computing hardware; communications links configured for supporting wireless communications to and from the system; a thermal management system for regulating temperature of the computing hardware using mechanically pumped coolant fluid; and a heat radiating structure for radiating heat from the thermal management system away from the satellite; wherein the system is configured to orbit the earth in an orbit with substantially continuous sun exposure.

[0006] The thermal management system can be configured to implement two phase cooling, wherein the coolant fluid fully or partially boils when exposed to heat generated by the computing hardware and fully or partially condenses when cooled by the heat radiating structure.Starcloud-03W002

[0007] The system can consist of a plurality of connectable modules configured for connection in orbit.

[0008] The plurality of connectable modules can have one or more docking interfaces configured to transfer electrical energy, coolant fluid, and data between connectable modules. The coolant fluid can be configured to be pumped between a connectable module embodying the heat radiating structure and a connectable module embodying the computing hardware. The plurality of connectable modules can include: a plurality of connectable modules embodying computing hardware; a plurality of connectable modules embodying power sources; and a plurality of connectable modules embodying heat radiating structures.

[0009] The system can include a mechanically and electrically integrated spine structure configured to support and connect a plurality of the connectable modules.

[0010] The system can include a plurality of high-capacity data storage modules, wherein prior to being launched into orbit, each of the plurality of high-capacity data storage modules is loaded with an amount of data that would otherwise take more than 30 days to transmit wirelessly from earth to the system.

[0011] The orbit can be dawn-dusk sun-synchronous. The dawn-dusk sun-synchronous orbit can have an orbital plane that precesses one revolution per year to remain substantially normal to the sun. The orbit can be geostationary.

[0012] The enclosure can be hermetically sealed.

[0013] As will be appreciated by one skilled in the art, multiple aspects described in this summary can be variously combined in different operable embodiments. All such operable combinations, though they may not be explicitly set forth in the interest of efficiency, are specifically contemplated by this disclosure.BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 illustrates a system block diagram of an orbital data center network architecture.

[0015] Figure 2 illustrates a compute container schematic of an orbital data center.Starcloud-03W002

[0016] Figure 3 illustrates a megawatt class orbital datacenter with deployed solar arrays and radiators.DETAILED DESCRIPTION

[0017] In the following description, references are made to various embodiments in accordance with which the disclosed subject matter can be practiced. Some embodiments may be described using the expressions one / an / another embodiment or the like, multiple instances of which do not necessarily refer to the same embodiment. Particular features, structures or characteristics associated with such instances can be combined in any suitable manner in various embodiments unless otherwise noted. By way of example, this disclosure may set out a set or list of a number of options or possibilities for an embodiment, and in such case, this disclosure specifically contemplates all clearly feasible combinations and / or permutations of items in the set or list.

[0018] 1. Overview and Objectives

[0019] This disclosure relates to modular orbital data centers designed for low cost, high value density, and future-proof serviceability. A system architecture enables independent docking / undocking of compute and support modules; maintainability over >10 years without fleet retirement; minimal moving parts and single-port service interfaces; fault-tolerant operation with graceful degradation; and incremental scalability from one container to NNN while remaining unit-economical at every step. When designing orbital data centers, we can be guided by existing practices in designing container-based terrestrial data centers. Each container can have sets of racks containing the compute and storage units, built-in networking, power and cooling infrastructure.

[0020] 2. System Architecture

[0021] 2.1 Modules and Spine

[0022] An orbital data center can include a mechanically and electrically integrated spine structure supporting a constellation of compute containers and auxiliary modules (power, thermal, networking, communications, logistics). The spine can provide structural load paths, attitude-control interfaces, cable trays / umbilicals, and mounting for directory / spine switches sized to the intended training workloads. Containers can beStarcloud-03W002 co-located within hundreds of meters to meet intra-cluster latency targets for large-scale Al training.

[0023] 2.2 Universal Service Port

[0024] Each container can include a single, universal docking port that provides (i) high-voltage DC (HVDC) power input, (ii) optical networking fan-outs capable of terminating thousands of fiber pairs, and (iii) closed-loop liquid cooling supply / return manifolds. The port can use a low-actuation, capture-and-latch mechanism to minimize moving parts and critical failure points. Interfaces can be keyed and hot-swappable to permit module replacement without impacting neighbors.

[0025] 2.3 Compute Container

[0026] A compute container can house rack-scale compute and storage, integrated power conversion, and cooling distribution. The compute bay can be either (a) pressurized with an inert gas to enable supplemental forced convection for non-direct-cooled components, or (b) immersed in a compatible dielectric coolant that also augments radiation shielding.

[0027] 3. Power Generation and Distribution

[0028] 3.1 Solar Power Plant

[0029] Power can be generated by large-area deployable solar arrays assembled from modified terrestrial-grade silicon photovoltaic (PV) for cost efficiency. Array deployment can use Z-fold, roll-out, or picture-frame mechanisms. Back-surface emissivity-control coatings can be used to passively regulate PV temperature; active coolant loops to the PV can be included if needed.

[0030] 3.2 Scale Example

[0031] A representative 5 GW facility can employ an array of approximately 4 km x 4 km at ~22% beginning-of-life efficiency and ~90% cell fill factor in near-continuous illumination.

[0032] 3.3 DistributionStarcloud-03W002

[0033] Array power can be routed over kV-class HVDC trunks along the spine. The selected bus voltage trades DC-DC conversion complexity / efficiency against conductor mass (copper gauge) and insulation. Local isolated DC-DC stages at each container generate rack-level rails.

[0034] 4. Thermal Management

[0035] 4.1 Facility-Level Loops

[0036] Thermal load can be transported from containers to radiators via closed liquid loops, using two-phase segments where practical to reduce mass flow and pumping work. Loop topology can be fault-tolerant (ring or dual-feed) to avoid single points of failure.

[0037] 4.2 Container- Level Cooling

[0038] High-density racks can use direct-to-chip liquid cooling or two-phase evaporators. Manifolds can be service-isolatable at the universal port. Coolant selection can prioritize low viscosity, low toxicity, radiation stability, and compatibility with elastomers and brazed joints.

[0039] 4.3 Radiator System

[0040] Waste heat can be rejected by lightweight, deployable, two-sided radiators oriented primarily toward deep space. Example performance at a radiator mean temperature of ~50 °C might be: inlet ~60 °C, outlet ~40 °C. Optionally, heat pumps can raise radiator temperature to increase areal rejection via the TA4 term, trading electrical work for reduced radiator area. Radiators may be co-planar or inline with the PV blanket, with sun-facing surfaces treated to suppress absorptivity while maintaining high IR emissivity.

[0041] 5. Networking and Workload Topology

[0042] 5.1 Intra-DC Fabric

[0043] Training-class Al workloads require low latency and high bisection bandwidth across the full container set. The spine can integrate a multi-tier switching fabric; containers can connect in short-reach, tightly spaced topologies (e.g., daisy-chain segments feeding leaf / spine) to maintain path lengths within a few hundred meters.Starcloud-03W002

[0044] 5.2 External Connectivity

[0045] Backhaul can employ laser inter-satellite links (ISLs) and / or RF cross-links to orbital relays. Examples include crosslinking with commercial constellations. For bulk data ingress / egress, data shuttles— small, dockable logistics modules launched from Earthtransport peta- to exabyte-scale datasets in a single trip and interface through the universal port. This "sneakernet-in-space" approach amortizes latency over massive payloads.

[0046] 5.3 Workload Mix

[0047] In one embodiment, a design target is foundation-model training at cluster scale; residual GPU capacity supports inference and general compute. Scheduler policies can be thermally aware and power-budget aware to respect radiator headroom and bus limits.

[0048] 6. Orbit, Attitude, and Environment

[0049] 6.1 Orbit Selection

[0050] One embodiment operates in dawn-dusk sun-synchronous low-Earth orbit (SSO) to achieve near-continuous solar illumination, reduce thermal cycling, and minimize battery storage. The orbital plane precesses ~1 rev / year, staying approximately normal to the Sun vector.

[0051] 6.2 ADCS and Structural Control.

[0052] Large deployables can be stabilized by the attitude determination and control system (ADCS) using reaction wheels, magnetic torquers, control-moment gyros, thrusters, or other methods sized for array / radiator torques and environmental disturbances. The spine can provide distributed attachment points for actuators and sensors to maintain pointing and jitter limits compatible with ISLs.

[0053] 6.3 Radiation and Materials

[0054] Although SSO radiation is modest, sensitive components are protected against latch-up, single-event transients, and TID via shielding and part selection. Shielding mass scales with surface area, while compute scales with volume; consequently, shielding per unit compute decreases with container size. Atomic oxygen is mitigated with protective coatings and surface treatments.Starcloud-03W002

[0055] 7. Reliability, Maintainability, and Safety

[0056] 7.1 Graceful Degradation

[0057] No single module's failure should prevent continued operation. Power trunks, coolant loops, and network paths can be made redundant; a scheduler can de-rate affected zones automatically.

[0058] 7.2 Serviceability

[0059] Containers and subsystems can be line-replaceable units (LRUs). Dock / undock operations can occur via the universal port, with automated leak-before-latch checks for cooling circuits and live-insertion for optical links. Failed or obsolete containers can be (i) returned as payloads on visiting vehicles or (ii) designed to fully demise on controlled re-entry.

[0060] 7.3 Lifetime and End-of-Life (EOL)

[0061] Cooling and power subsystems define ultimate life, with a design target of ~15 years (comparable to on-orbit fluid loops). At EOL, modules can be salvaged for components / materials or de-orbited to comply with debris mitigation.

[0062] 8. Deployment and Scalability

[0063] 8.1 Incremental Build-Out

[0064] The orbital data center can be configured to grow via small, repeated launches: install spine segment, attach PV / radiator deployables, connect HVDC trunks, add containers. Each added container is profitable on first power-up, avoiding step-change CapEx. Software and fabric scale elastically with module count.

[0065] 8.2 Minimal Mechanisms

[0066] Mechanisms can be restricted to deployment, port capture, and valve actuation; all other functions can be static or solid-state, reducing failure modes and inspection burden.

[0067] 9. Methods of Operation (Representative)1. Power-up: orient arrays in SSO; synchronize ADCS; energize HVDC trunks.2. Thermal priming: circulate coolant; verify loop pressures / temperatures; optionally stageStarcloud-03W002 heat pumps.3. Network bring-up: light intra-DC fabric; verify bisection; commission ISLs / backhaul.4. Compute commissioning: dock containers; execute burn-in / BER checks; enroll into scheduler.5. Data ingress: receive bulk datasets via ISLs or docked data shuttle; perform on-orbit integrity checks.6. Training run: allocate racks cluster-wide; enforce power / thermal budgets; checkpoint to redundant storage.7. Maintenance: isolate a container; close valves; quiesce links; undock; replace; hot-rejoin fabric.8. EOL: de-energize segment; transfer useful inventory; controlled re-entry or salvage.

[0068] 10. Alternative and Additional Features and Embodiments

[0069] Cooling: immersion vs. direct-to-chip; single- vs. two-phase loops; optional heat-pump elevation of radiator temperature.

[0070] Power: PV chemistry (Si, perovskite, thin-film); MPPT topologies; energy storage minimal or omitted due to continuous illumination.

[0071] Networking: alternate topologies (dragonfly / torus); co-packaged optics; wavelength-routed backbones.

[0072] Orbits: other LEOs / MEOs for latency or regulatory considerations when continuous power is not mandatory.

[0073] Materials: high-emissivity, low-a radiator surfaces; AO-resistant coatings; micrometeoroid / space-debris shielding as needed.

[0074] A modular system to provide datacenter services in orbit around Earth can include: a power source, computing hardware, high throughput communications links to provide network connectivity, and mechanically pumped fluid cooling for said computing hardware, wherein the satellite orbits the earth in an orbit with continuous or near- continuous sun exposure such as dawn-dusk sun-synchronous or geostationary. The system can consists of a single satellite or a cluster of satellites. The system can be a larger satellite made up of smaller elements which are launched at separate times and docked in orbit toStarcloud-03W002 form a larger single satellite. The thermal management system can employ one or two phase cooling such that the coolant fully or partially boils when exposed to the computing hardware heat and fully or partially condenses when cooled in the radiator.

[0075] A main satellite structure can be composed of one or more inter-connected modules called spines. Each spine presents a multitude of standardized docking interfaces, each providing electrical energy, coolant transfer, and high throughput data network to each docked module to the spine. Each spine can also host one or more radiators, each connected through an interface that offers cooling capability through one or more paths for redundancy. Each spine can be built in a modular fashion using a similar docking interface to route power, cooling and network to each element of the spine. The docking interface can have a quick connect / disconnect capability for electrical energy in the order of kilowatts / megawatts, energy routed by the spine from the solar panels into each docked module. The docking interface can have a quick connect / disconnect capability for coolant flowing into and from each compute module into the main structure. The cold coolant can be routed from the radiators into each attached module. The hot coolant can be routed back from each module into the attached spine which routes it to the radiators.

[0076] The docking interface can have a quick connect / disconnect capability for one or more high speed network traffic connectors to and from each module to the network routing infrastructure into the attached spine. Each spine can also host one or more solar panels, each connected through an interface that provides electrical energy through one or more paths for redundancy.

[0077] The satellite can be given a flat form factor for lower launch and deployment cost.

[0078] Large scale, high latency data transfer can be achieved by launching a smaller spacecraft with high capacity storage capability, which can rendezvous with the larger system in orbit to transfer data from / to ground or other satellites.

[0079] The system can include computing units assembled in a modular fashion.

[0080] 11. ConclusionStarcloud-03W002

[0081] Although the subject matter has been described in terms of certain embodiments, other embodiments that may or may not provide various features and aspects set forth herein shall be understood to be contemplated by this disclosure. The specific embodiments set forth herein are disclosed as examples only, and the scope of the patented subject matter is defined by the claims that follow.

Claims

Starcloud-03W002CLAIMS1. An orbital data center system (system) configured to provide data center services in an orbit around Earth, the system comprising: a power source comprising one or more solar power arrays; data center-capable computing hardware; a sealed enclosure containing the computing hardware; communications links configured for supporting wireless communications to and from the system; a thermal management system for regulating temperature of the computing hardware using mechanically pumped coolant fluid; and a heat radiating structure for radiating heat from the thermal management system away from the satellite; wherein the system is configured to orbit the earth in an orbit with substantially continuous sun exposure.

2. The system of claim 1, wherein the thermal management system is configured to implement two phase cooling, wherein the coolant fluid fully or partially boils when exposed to heat generated by the computing hardware and fully or partially condenses when cooled by the heat radiating structure.

3. The system of claim 1, wherein the system consists of a plurality of connectable modules configured for connection in orbit.

4. The system of claim 3, wherein the plurality of connectable modules has one or more docking interfaces configured to transfer electrical energy, coolant fluid, and data between connectable modules.

5. The system of claim 3, wherein the coolant fluid is configured to be pumped between a connectable module embodying the heat radiating structure and a connectable module embodying the computing hardware.

6. The system of claim 3, wherein the plurality of connectable modules comprises: a plurality of connectable modules embodying computing hardware;Starcloud-03W002 a plurality of connectable modules embodying power sources; and a plurality of connectable modules embodying heat radiating structures.

7. The system of claim 6, wherein the system comprises a mechanically and electrically integrated spine structure configured to support and connect a plurality of the connectable modules.

8. The system of claim 1, further comprising a plurality of high-capacity data storage modules, wherein prior to being launched into orbit, each of the plurality of high-capacity data storage modules is loaded with an amount of data that would otherwise take more than 30 days to transmit wirelessly from earth to the system.

9. The system of claim 1, wherein the orbit is dawn-dusk sun-synchronous.

10. The system of claim 2, wherein the dawn-dusk sun-synchronous orbit has an orbital plane that precesses one revolution per year to remain substantially normal to the sun.

11. The system of claim 1, wherein the orbit is geostationary.

12. The system of claim 1, wherein the enclosure is hermetically sealed.

Images