All articles tagged with #orbital data centers
SpaceX, in partnership with Nvidia, plans to launch its first AI-powered orbital data centers—the Vera Rubin NVL72 system—by Q4 2027, with broader deployments in 2028, moving the timeline up from 2028. The push depends on Starship readiness and ongoing test flights, though analysts see upside as SpaceX builds out orbital compute.
Rising AI compute demand is driving interest in orbital data centers powered by solar energy, which could slash electricity bills but require massive solar arrays, radiators, and launch costs; feasibility studies show near-term use may be limited to specialized workloads due to power, cooling, debris, and communication constraints.
Five‑month‑old Orbital has asked the FCC for permission to deploy up to 100,000 low‑Earth‑orbit data‑center satellites, each around 100 kW, to deliver about 10 gigawatts of compute power for AI, with sats at 500–850 km using optical intersatellite links, planning a scaled‑down demo and a 2028 Orbital‑1 launch as it eyes a full constellation in the coming decade, competing with Starcloud, Cowboy Space and SpaceX's Starlink while addressing manufacturing and space‑traffic challenges.
Cowboy Space has raised $275 million at a $2 billion valuation to develop vertically integrated space infrastructure that couples rocket upper stages with power and in‑orbit compute to host AI workloads. The plan envisions thousands of satellites forming orbital data-center networks, with a first mission this year to demonstrate wireless power beaming and a long‑term goal of a one‑megawatt data center in orbit by 2028. The venture, led by Index Ventures with new backers including IVP, Blossom Capital and SAIC, aims to manufacture an integrated rocket/upper stage/compute system and to grow an orbital AI compute ecosystem far beyond current terrestrial limits.
SpaceX has unveiled a plan to deploy up to one million solar-powered satellites to provide AI computing from orbit, including a gargantuan “mini” satellite that dwarfs Starship; a proposed $20 billion Terafab facility would generate up to 200 billion AI or memory chips per year, while astronomers warn such a constellation could severely hinder deep-space observations, and analysts estimate deployment costs could exceed $1 trillion.
SpaceX’s plan to orbit up to 1 million data-center satellites is under FCC review. Gwynne Shotwell says the target is aspirational and may not be reached, but the company is starting with a high ceiling. Satellites could orbit Earth, the Moon, or even the Sun to manage congestion, while critics cite light pollution, atmospheric impact, and space-safety concerns. Rival firms are proposing their own large constellations, and SpaceX has signaled a cautious, phased approach with testing before scaling.
Elon Musk teased a rough rendering of orbiting data centers that would be longer than the ISS, powered by large solar arrays to support 100 kW of AI computing per satellite with future megawatt-scale versions, and a plan to launch up to 1 million satellites; experts warn about collision risk, light pollution, and atmosphere/astronomy interference, while SpaceX argues heat rejection in space and pursues a Terafab chip factory in Austin to supply AI hardware.
Astronomers warn SpaceX’s FCC filing to deploy up to one million additional satellites for orbital data centers could keep satellites in sunlight at midnight and increase atmospheric pollutants if they deorbit, jeopardizing ground-based astronomy. Regulators are fast-tracking environmental reviews amid opposition from astronomers and Amazon as the company expands its Starlink megaconstellation.
Blue Origin has filed with the FCC seeking authority to launch up to 51,600 low-Earth-orbit satellites for orbital data centers under Project Sunrise, supplementing its 5,408-satellite TeraWave plan and using laser links to serve terrestrial data centers and government customers; the move intensifies the space data-center race, drawing objections from SpaceX while other players like Starcloud eye the market and Amazon’s Leo/AWS could intersect in the future.
Blue Origin has filed with the FCC to pursue Project Sunrise, a megaconstellation of up to 51,600 satellites in sun-synchronous orbits (500–1,800 km) to host space-based data centers that would supplement terrestrial AI computing, joining ongoing industry pushes by SpaceX and Nvidia-backed teams in the race for orbital compute.
Nvidia unveiled the Space-1 Vera Rubin Module, a space-ready AI computing system designed to run on satellites and enable onboard data centers, aiming to outperform the H100 and support in-space AI analytics for satellite constellations; it integrates IGX Thor and Jetson Orin tech but faces radiation and cooling challenges and is not yet commercial, with Planet Labs and other partners testing Nvidia hardware for space-derived Earth-imaging analytics.
NVIDIA announced Space Computing, bringing data-center‑class AI to orbital data centers and space-edge environments via the Space‑1 Vera Rubin Module, IGX Thor, and Jetson Orin. Partners such as Aetherflux, Axiom Space, Kepler, Planet, Sophia Space and Starcloud will power next‑gen space missions with on‑board AI, real‑time analytics and autonomous operations. The lineup also includes ground‑processing acceleration with RTX PRO 6000 Blackwell Server Edition, while Space‑1 Rubin promises up to ~25x more AI compute for space inferencing versus the H100; IGX Thor and Jetson Orin are available now, with Space‑1 Rubin Module coming later.
Starcloud has filed with the FCC to operate up to 88,000 low‑Earth‑orbit satellites to serve as orbital data centers for AI and other computing needs, placing them in dusk‑dawn sun‑synchronous shells at 600–850 km. The plan envisions intersatellite optical links and Ka‑band telemetry/control, with safety measures including coordination with other operators and full demisability on reentry. Starcloud‑1 has already flown a small satellite with an Nvidia H100 processor; Starcloud‑2 is planned for 2027, and there are concepts for Starcloud‑3 and Starcloud‑4, including a massive satellite array with multi‑kilometer dimensions. The Starcloud filing would create a constellation far larger than existing systems like Starlink (about 10,000 satellites) but smaller than SpaceX’s unfunded one‑million‑satellite proposal.
SpaceX and xAI reportedly plan a million-satellite constellation to run AI data centers in orbit. While proponents cite solar efficiency and lower terrestrial costs, experts warn it could trigger debris cascades (Kessler syndrome), overheating and radiation risks for space hardware, brightening the night sky and harming astronomy, and disruption to Earth observation and climate monitoring; even if feasible, AI training would still require Earth-based resources, and governance gaps could complicate implementation.
SpaceX has asked the FCC to authorize a constellation of up to one million satellites in low Earth orbit that would act as an orbital data center, using near-constant solar power and intersatellite optical links (with Ka-band as a backup) to deliver AI compute from space. The satellites would operate at 500–2,000 km with 30-degree and sun-synchronous inclinations to maximize sun exposure; SpaceX says the system could reduce terrestrial data-center costs and environmental impact while enabling rapid AI processing, though the filing provides few technical details and seeks waivers from usual deployment milestones with no schedule or price disclosed. The plan aligns with SpaceX’s AI ambitions and broader corporate moves, including IPO considerations and potential ties to xAI.