Orbital data centers can enable liquid cooling, but traditional cooling equipment value per megawatt may be halved

Overview

The report views orbital data centers as modified satellites carrying AI server racks, rather than entirely new computing infrastructure. Their core still consists of a computing bus, solar arrays, and a thermal management system. Because there is no convection in a vacuum environment, heat generated by servers needs to be collected by direct-to-chip cold plates, transferred through a primary liquid loop and simplified CDUs into a water-free coolant secondary loop, and ultimately rejected to deep space by deployable radiator wings through infrared radiation.

Core views

The front half of orbital liquid cooling is highly similar to terrestrial direct-to-chip liquid cooling. The real change occurs at the final heat-rejection stage: radiator wings replace facility water systems, chillers, cooling towers, and dry coolers. Therefore, orbital data centers will not eliminate cooling demand, but will reconfigure the cooling equipment mix and reduce traditional equipment value per megawatt from $1.3 million to $1.8 million on the ground to $0.6 million to $0.8 million. Because orbital capacity is expected to account for only about 11% of the global total by 2031, the report believes this change does not yet constitute a material near- to medium-term threat to terrestrial cooling demand. Long-term opportunities come more from radiator wings, integrated aerospace thermal management technologies, and related M&A or partnerships, rather than traditional aftermarket service.

Analysis framework

The research first reviews the heat-transfer chain of terrestrial liquid-cooled data centers, then builds orbital cooling assumptions based on established satellite and spacecraft thermal management principles. It then compares, item by item, the chillers, computer room air conditioners, rear-door heat exchangers, CDUs, cold plates, manifolds, software, accessories, and radiator wings required by the two architectures, and estimates the overall market opportunity and potential exposure of each supplier based on equipment value per megawatt. Finally, it combines orbital capacity scenarios, SPCX's tendency toward vertical integration, and on-orbit maintenance conditions to assess the investment implications for existing equipment companies.

Methodology notes

• Engineering physicsStefan-Boltzmann law

  Thermal radiation flux is mainly determined by surface temperature and emissivity

  The report uses an aerospace radiator emissivity of about 0.92 and includes absorption of solar and Earth radiation to estimate that a double-sided radiator at about 20 degrees Celsius can achieve net heat rejection of about 633 watts per square meter.

• Architecture comparisonMapping of terrestrial and orbital liquid-cooling components

  Compare the complete chain from chip heat absorption to final heat rejection for the two types of data centers

  Cold plates, manifolds, and primary liquid loops are largely retained, while the orbital system uses a water-free coolant secondary loop and radiator wings to replace facility water systems, chillers, and air-side heat-rejection equipment.

• Market opportunity estimationEquipment value per megawatt analysis

  Estimate supplier revenue opportunities based on equipment value configured per megawatt of compute load

  The report assigns component values item by item and includes accessories and redundant configurations, resulting in estimated orbital cooling value of $0.6 million to $0.8 million per megawatt and terrestrial liquid cooling value of $1.3 million to $1.8 million per megawatt.

• Scenario analysis2031 orbital capacity penetration scenario

  Assess substitution risk based on the ratio of orbital capacity to global installed data center capacity

  The report expects orbital data centers to reach about 27 GW by 2031, or about 11% of global capacity, and on this basis judges that the near- to medium-term impact on the terrestrial liquid-cooling market is limited.

Asset mapping & comparison

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

• VRT (Vertiv)

  Exposure as a core equipment supplier for terrestrial and orbital liquid cooling

  Strengths

  The report estimates its terrestrial liquid-cooling value at about $1.4 million to $1.8 million/MW and orbital value at about $0.5 million to $0.8 million/MW, both at relatively high levels among the listed incumbent suppliers.

  Weaknesses

  The orbital architecture eliminates multiple types of facility-level equipment and simplifies CDUs, leading to a decline in value per megawatt.

  Comparison

  Compared with most peers, its combination of cold plates, manifolds, CDUs, and software enables higher retained value under the orbital architecture.

  Risks

  SPCX vertical integration, radiator wing value not included, limited on-orbit service revenue, and changes in the final equipment configuration.

• SU.FP (Schneider Electric)

  Exposure as a data center liquid-cooling and management software supplier

  Strengths

  The report estimates its terrestrial liquid-cooling value at about $1.4 million to $1.8 million/MW and orbital value at about $0.5 million to $0.8 million/MW, with opportunities related to CDUs, cold plates, manifolds, and software.

  Weaknesses

  The contribution of traditional facility-level heat-rejection equipment is reduced in the orbital architecture, and the orbital market remains small in the near to medium term.

  Comparison

  The value range given by the report is broadly comparable to VRT and higher than most single-component suppliers.

  Risks

  Market share may be diverted by vertically integrated orbital operators or specialized aerospace thermal management companies.

• NVT (nVent)

  Exposure as a supplier of liquid-loop components such as CDUs and manifolds

  Strengths

  The orbital architecture still requires CDUs and manifolds, and the report estimates its orbital value at about $0.2 million to $0.4 million/MW.

  Weaknesses

  Product coverage is relatively concentrated and lacks the cold plate and software value that the report attributes to leading integrated suppliers.

  Comparison

  Orbital value is lower than VRT and Schneider, but higher than companies lacking key liquid-cooling components.

  Risks

  CDU simplification leads to lower unit prices, and operator in-house development may further compress the serviceable market.

• TT, CARR, JCI, ETN

  Traditional HVAC, electrical, and partial liquid-cooling component suppliers

  Strengths

  Continue to benefit from the much larger scale of terrestrial data center construction; some companies can still supply simplified CDUs required for orbit.

  Weaknesses

  Chillers, computer room air conditioners, rear-door heat exchangers, cooling towers, and dry coolers are largely no longer needed in the orbital architecture.

  Comparison

  The report estimates the orbital cooling value of these companies mostly at about $0.2 million/MW, significantly below VRT and Schneider.

  Risks

  If the long-term share of orbital capacity rises significantly, the value of their traditional facility-level equipment may face structural erosion.

• NVDA

  Potential beneficiary of orbital AI compute loads and GPU demand

  Strengths

  Orbital data centers use GPU clusters to run AI workloads, and computing capacity expansion can increase demand for accelerated computing.

  Weaknesses

  Large-scale deployment is constrained by power supply, heat-rejection area, structural weight, and launch capability.

  Comparison

  Its opportunity mainly comes from computing hardware and does not directly correspond to the cooling equipment value estimated in this report.

  Risks

  Orbital computing economics, launch progress, and thermal management scalability fall short of expectations.

• SPCX (SpaceX)

  Key driver of orbital data center deployment and potential vertically integrated operator

  Strengths

  It is at the core of the rapidly developing space economy and is actively promoting commercialization of orbital data centers; vertical integration helps control system design and deployment.

  Weaknesses

  Multi-megawatt AI loads require large-scale solar arrays and radiator wings, and on-orbit repair is difficult.

  Comparison

  Compared with external equipment suppliers, SPCX is more likely to internalize part of its thermal management capabilities.

  Risks

  Launch delays, insufficient technical transparency, deployment costs, radiator wing reliability, and orbital capacity falling below expectations.

Key data

• Orbital cooling equipment valueApproximately $0.6 million to $0.8 million/MWExcludes radiator wings and assumes the unit value of space equipment is comparable to ground equipment; the final configuration may lead to differences.
• Terrestrial liquid-cooling equipment valueApproximately $1.3 million to $1.8 million/MWIncludes chillers, installation, piping, pumping, heat-rejection equipment, and other items.
• Orbital CDU valueApproximately $150,000 to $200,000/MWAbout 50% of terrestrial liquid-cooling CDUs, because it mainly consists of pumps and heat exchangers.
• Orbital cold plate valueApproximately $200,000 to $250,000/MWBroadly consistent with terrestrial direct-to-chip liquid cooling.
• Orbital manifold valueApproximately $50,000 to $100,000/MWUsed to distribute and recover coolant between CDUs and individual cold plates.
• Radiator net heat-rejection capacityApproximately 633 watts/square meterBased on about 20 degrees Celsius, double-sided radiation, and 0.92 emissivity, including the impact of solar and Earth radiation.
• Radiator area required for a 100 kW platformApproximately 158 square metersCorresponds to solar array area of about 300 to 500 square meters; the report example also uses about 400 square meters.
• Share of surface area for power supply and heat rejection facilitiesMore than 95%Shows that orbital data centers are more like power supply and heat-rejection platforms than platforms dominated by computing hardware surface area.
• 2031 orbital data center capacityApproximately 27 GWAbout 11% of global installed data center capacity.
• On-orbit aftermarket service opportunityExtremely limited, potentially close to zeroDedicated repair missions are costly and usually require years of planning, making it difficult to form a service-revenue tail similar to ground equipment.

Impact & implications

In the near to medium term, terrestrial data centers will still account for the vast majority of capacity, and AI infrastructure expansion and liquid-cooling penetration will continue to support existing cooling equipment suppliers. If orbital deployment accelerates, the revenue structure will shift from chillers, cooling towers, and air-side equipment to cold plates, manifolds, simplified CDUs, radiator wings, and integrated thermal management systems, but both traditional equipment value per megawatt and aftermarket service revenue may decline. Manufacturers with comprehensive liquid-cooling product portfolios are relatively more resilient; specialized aerospace thermal management start-ups may become partnership or acquisition targets.

Risks

• The orbital cooling architecture remains a research assumption built on existing satellite technology, and future public technical solutions may differ from the report's assumptions.
• The value per megawatt assumes space equipment prices are comparable to ground equipment; final size, materials, reliability, and aerospace certification requirements may significantly change costs.
• The $0.6 million to $0.8 million/MW orbital valuation excludes radiator wings, which may be one of the higher-value and most engineering-intensive new categories in the system.
• SPCX's preference for vertical integration may reduce the external market share available to existing cooling equipment vendors.
• Orbital equipment is difficult to repair or replace, leaving suppliers with almost no traditional aftermarket service revenue and exposing them to higher upfront reliability requirements.
• The approximately 27 GW and 11% 2031 capacity forecast depends on aggressive deployment assumptions, launch schedules, and orbital computing economics.
• Radiator area, solar array area, mass, folding deployment, and structural stability may limit multi-megawatt system scaling.
• In the long term, if orbital capacity significantly exceeds the report's expectations, demand for traditional chillers, cooling towers, and air-side equipment may face a greater impact.

What to watch

• Specific liquid-cooling, radiator wing, and solar array architectures for orbital data centers disclosed by SPCX and other operators.
• Whether orbital capacity can approach the planned scenario of about 27 GW by 2031 and about 11% of global capacity.
• Radiator wing cost per megawatt, mass, area, folding deployment reliability, and the commercial supplier landscape.
• Whether existing cooling equipment companies can enter the supply chain for aerospace-grade cold plates, manifolds, CDUs, and water-free coolant loops.
• Whether SPCX chooses internal development or external procurement of thermal management systems, and the impact of its vertical integration on supplier share.
• Partnerships, investments, or M&A activity by traditional cooling companies around orbital thermal management start-ups.
• Whether orbital equipment design life, redundancy standards, and failure rates can offset the constraints from the lack of repair conditions.
• Whether deployment growth of terrestrial direct-to-chip liquid cooling continues to be significantly higher than that of orbital data centers.