Orbital computing is moving from concept to early validation, with European semiconductors and space supply chains offering thematic exposure

Overview

The report discusses the feasibility, evolution path, economics, and stock implications of orbital computing as a new AI infrastructure theme. Its central view is that orbital computing is not likely to replace hyperscale terrestrial data centers in the short term, but has more realistic starting points in scenarios such as orbital edge AI, on-orbit inference, earth observation, defense, maritime, disaster monitoring, and autonomous spacecraft. Over the long term, if reusable rockets, mass production of satellites, optical inter-satellite links, thermal dissipation, power systems, and radiation-tolerant computing continue to improve, orbital computing could evolve into a distributed AI infrastructure layer around Earth.

Core views

Core views in the report include: first, the commercial starting point for orbital computing is "process data where it is generated" rather than building a space version of a hyperscale data center. Second, space offers potential advantages such as high solar exposure, vacuum optical communication, and reduced constraints from terrestrial power grids, land, and water resources, but introduces new constraints including launch, mass, thermal management, radiation, maintenance, and regulation. Third, economics are not only about US$/kg at launch; they focus on capex/W, kg/kW, compute density, launch cadence, and time-to-power. Fourth, most of the supply-chain value accrues in optical/RF, power and thermal management, radiation-hardened semiconductors, and supporting components for core AI chips. Fifth, STMicroelectronics and Infineon are viewed as key beneficiaries in European semiconductor coverage due to their space-grade power management, radiation-hardened semiconductor, communications, and control electronics capabilities.

Analysis framework

The report uses a thematic supply-chain research and economics framework: it first defines orbital computing architecture and a three-stage evolution path, then uses SpaceX-related launch research as a cost benchmark to assess potential economics around capex/W, launch costs, satellite hardware, compute payload, service life, and radiation losses, and finally maps implications across semiconductor, optical communications, RF, batteries, thermal management, and satellite system supply chains.

Methodology notes

• Industry evolutionThree-stage orbital computing roadmap

  Orbital edge computing, orbital cloud/distributed computing, space AI infrastructure

  The report divides orbital computing into three stages: in the short term, satellites process imagery and inference tasks on orbit; in the mid term, multiple compute satellites use a LEO network to schedule and shift workloads; in the long term, modular compute satellites with solar arrays, radiators, and optical links form a space AI infrastructure layer.

• Economic analysisCost-per-watt framework

  Shifting from US$/kg launch cost to capex/W and US$/W/year

  The report argues that declining launch costs are a necessary but insufficient condition; investment judgement should instead focus on unit computing-power cost, annualized cost over five-year life, compute density, launch cadence, and deployment speed.

• Supply-chain mappingOrbital computing technology stack

  Compute payload, solar power and thermal management, optical networking, formation flying, storage and orchestration

  The report breaks orbital computing into hardware and system layers: AI chips and memory provide compute, the satellite platform provides power and cooling, optical inter-satellite links provide networking, and software handles storage, scheduling, and fault management.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• STMicroelectronics

  Key European semiconductor beneficiary and enabling player in the LEO and orbital computing supply chain

  Strengths

  Has radiation-hardened semiconductor, RF front-end module, BiCMOS, and PLP capabilities; management has guided that cumulative sales over the next three years are "well above US$3bn"; terminal, satellite, and gateway revenues can be modeled separately.

  Weaknesses

  Revenue realization depends on LEO constellation buildout, terminal scaling, and launch cadence; orbital data centers remain a longer-term opportunity.

  Comparison

  In European coverage, the report places it alongside Infineon as a key enabling player; STMicro recently hosted LEO-themed events and disclosed more detailed BOM and technology positioning.

  Risks

  Fewer satellite launches than expected, delayed commercial adoption, pricing pressure, supply-chain bottlenecks, and slower-than-expected pace of space infrastructure buildout.

• Infineon Technologies AG

  Space-grade power management and radiation-hardened high-reliability semiconductor supplier

  Strengths

  Well positioned in space-grade power management, power MOSFETs, solid-state relays, and space Schottky diodes, fitting satellite power, attitude-control, and communication system requirements.

  Weaknesses

  Orbital computing contribution may be more long-term; short-term revenue sensitivity depends on execution of satellite platform and power-system orders.

  Comparison

  Named with STMicro as one of the preferred orbital-computing enablers in the report's European semiconductor coverage.

  Risks

  Launch cost reductions may lag expectations, long space qualification cycles, intensifying competition, and customer project delays.

• NVIDIA / AMD / Broadcom / Micron / TSMC / SK hynix / Samsung

  Related supply chain for core AI silicon, memory, communication chips, and advanced manufacturing

  Strengths

  AI accelerators, SoCs, memory, modems, and advanced node manufacturing remain the foundation for orbital computing payloads.

  Weaknesses

  The space environment imposes stricter radiation, power, thermal, and reliability requirements, so standard terrestrial data-center solutions cannot be directly ported.

  Comparison

  The report sees core AI chips as necessary, but value also extends to power, thermal management, and communication hardware around the compute payload.

  Risks

  Radiation-hardening adaptation cost, launch mass limits, life-loss degradation, thermal design, and in-orbit maintenance constraints.

• Optical/RF, power and thermal management supply chain

  Critical peripheral enabling layer for the orbital computing platform

  Strengths

  Phased-array/RF, optical interconnect, batteries, solar power, radiators, and power systems determine whether a satellite can stably run compute workloads.

  Weaknesses

  System complexity is high; progress in a single technology area is not sufficient to secure overall economics.

  Comparison

  The report views this layer as equally important as core compute chips, since an orbital platform must carry both power and thermal systems.

  Risks

  Optical link bandwidth, formation-flying distances, radiator efficiency, supplier concentration, and space debris risk.

Key data

• Predicted space economy size in 2035US$1.8trThe report cites a World Economic Forum forecast.
• Orbital computing capex/W forecastabout US$60/W in 2030, US$32/W in 2031, US$15/W in 2035, US$9/W in 2040Based on Morgan Stanley's citation of a SpaceX-related framework.
• Annualized cost estimate for 2031about US$6.5/W/yearCalculated over a five-year useful life, close to current Blackwell sector benchmark US$6.8/W/year.
• Starship launch cost assumptionabout US$500/kg in 2030, below US$200/kg in 2035, below US$150/kg in 2040The report says launch cost declines need to be combined with launch cadence and computing load per deployment.
• Early AI satellite power and massabout 150kW total power, 120kW compute power draw, about 2.1 tons massBaseline early-compute satellite setting in the report model.
• Later-stage AI satellite power and massabout 913kW total power, 730kW compute power draw, about 6.1 tons massBenefits from larger launch vehicles, lighter solar arrays, and more efficient radiators.
• Compute density improvementabout 56kW/ton to 80kW/ton, and eventually to 120kW/tonHighlights the importance of kg/kW, radiator area, and satellite hardware cost.
• Radiation loss assumptionabout 4% per year, about 20% cumulatively over five yearsSupports redundant capacity, fault detection, and periodic replacement requirements.
• STMicroelectronics LEO sales forecastFY26 US$815mn to FY28 US$1.8bnThe report suggests this forecast may be conservative if satellite launch activity exceeds the model.
• STMicroelectronics FY26 terminal sales estimateUS$546mn, 67% of LEO salesManagement identifies terminal products as the largest expected revenue contribution.

Impact & implications

The investment implication is that the orbital computing theme may expand the long-term addressable market for semiconductors and space infrastructure, but gains will not be concentrated only in core AI GPUs. More direct beneficiaries include radiation-hardened power devices, guidance/control computing, memory, FPGA, communications and attitude-control electronics, phased-array/RF, optical links, batteries, power management, and thermal systems. For equities, the report highlights STMicroelectronics and Infineon as having strong enabling positions within European semiconductors, and it also mentions related supply-chain names such as NVIDIA, AMD, Broadcom, Micron, TSMC, SK hynix, Samsung, Analog Devices, Microchip, Shanghai Fudan, Qorvo, Amphenol, Coherent, Lumentum, GS Yuasa, Lite-On, and LG Energy Solution.

Risks

• Orbital computing may lack cost competitiveness early on, with economics dependent on whether reusable rockets and mass production of satellites are actually achieved.
• Declining launch cost alone is not enough to ensure commercialization; sufficient launch frequency and compute load per deployment are also required.
• Space radiation can cause memory upsets, compute errors, and performance degradation, requiring shielding, error correction, redundancy, and periodic replacement.
• Cooling is not free in space; heat can only be rejected through radiators, and radiator area and efficiency affect both mass and cost.
• Orbiting hardware is difficult to repair, requiring autonomous fault detection, workload rerouting, spare capacity, and replacement cycles.
• Space debris, low-orbit congestion, and collision risk are rising, potentially increasing insurance, regulatory, and operational complexity.
• International space regulation, data sovereignty, and cross-border data compliance may constrain certain applications.
• Specialized materials, space-grade components, and a limited number of global launch service providers may create supply-chain bottlenecks.

What to watch

• Actual launch cost, launch frequency, and payload capability of Starship and other heavy reusable rockets.
• Whether orbital computing capex/W declines quickly along the path outlined in the report starting in the early 2030s.
• Progress in follow-on validations by Starcloud, NVIDIA, SpaceX Starmind, Google Project Suncatcher, and Chinese on-orbit AI constellations.
• Whether STMicroelectronics LEO-related revenue scales according to the FY26-to-FY28 ramp model, especially terminal, satellite, and gateway revenue categories.
• Infineon order intake, qualification, and customer expansion in space-grade power management and radiation-hardened devices.
• Engineering progress in optical inter-satellite links, radiators, solar arrays, batteries, and autonomous orchestration software.
• Whether more explicit frameworks emerge for regulation, space traffic management, and LEO debris governance.