Space Data Centres

• Lonestar Data Holdings — Lunar Data Centre Commercial Pilot Progression (ongoing through 2026, with Mission 2 targeting 2027 lunar surface deployment) Lonestar, a Florida-based startup, has continued advancing its lunar data centre concept beyond the proof-of-concept stage demonstrated aboard the IM-1 Intuitive Machines lunar lander in February 2024. The company is now in active commercial discussions with sovereign government clients — specifically positioning the Moon as a regulatory-neutral, radiation-hardened, and geopolitically insulated archival jurisdiction. The strategic logic is that data stored on the lunar surface falls outside any existing national data sovereignty framework, a proposition that has attracted interest from financial institutions and government archivists. The competitive moat being constructed here is jurisdictional arbitrage combined with physical inaccessibility — a genuinely novel defensible position. Teams tracking this space should monitor whether Lonestar secures a named anchor customer before its Mission 2 payload commitment window closes in late 2026.

• ESA ASCEND Study — Commercial Orbital Data Centre Feasibility Outputs (study phase concluded 2025; policy and procurement implications materialising through 2026–2027) The European Space Agency's ASCEND (Advancing Space Cloud for European Net zero and Data sovereignty) study, led by a consortium including Thales Alenia Space, Airbus Defence & Space, and CloudFerro, produced feasibility outputs indicating that a solar-powered orbital data centre could achieve net-zero operational energy consumption by leveraging unobstructed solar flux (approximately 1,361 W/m² vs. terrestrial averages of 100–250 W/m²). The study's most strategically significant finding is that orbital data centres could address both the EU's AI compute energy crisis and its data sovereignty imperatives simultaneously. ESA is now in the process of translating these findings into procurement frameworks — watch for calls for tender from the European Commission's Horizon Europe and CASSINI programmes in Q3–Q4 2026. This positions European industrial primes for first-mover advantage in sovereign orbital compute, at the direct expense of US hyperscaler dominance in EU cloud markets.

• Starship Block 2 / IFT-9 Cadence and Cost-Per-Kilogram Trajectory (ongoing; payload-to-orbit economics improving through 2026) SpaceX's Starship programme — following the resolution of the IFT-7/IFT-8 anomaly sequence — has re-established launch cadence in 2026. While SpaceX does not publish official $/kg figures, independent analysis from BryceTech and Quilty Analytics suggests that at full reusability, Starship's marginal cost per kilogram to LEO could fall below $200 in the 2027–2028 timeframe, compared to Falcon 9's ~$2,700/kg. This is the single most important enabling variable for orbital data centre economics: a 1,000-server rack weighing approximately 10 metric tonnes currently costs ~$27M to orbit on Falcon 9; at Starship economics, that drops to ~$2M. No orbital data centre business case is viable without this cost compression. Competing heavy-lift programmes — including Blue Origin's New Glenn (now operational), Rocket Lab's Neutron (targeting 2026 first flight), and ESA's Ariane 6 — all remain materially more expensive, giving SpaceX structural leverage over any commercial orbital compute customer.

• Microsoft Azure Space & Orbital Computing Partnerships — Quiet Infrastructure Buildout (ongoing; Azure Space partner programme expanding through 2026) Microsoft has been systematically expanding its Azure Space partner ecosystem, integrating ground station networks (KSAT, Viasat), satellite connectivity (Starlink integration for enterprise edge), and — less publicly — exploring compute-at-orbit concepts through its partnership with Axiom Space for ISS-adjacent commercial infrastructure. While Microsoft has not announced a dedicated orbital data centre product, its pattern of infrastructure pre-positioning mirrors the pre-cloud terrestrial buildout of 2006–2010. The strategic risk for AWS and Google Cloud is that Azure Space's sovereign cloud positioning (already dominant in NATO/defence markets) could extend into orbital compute as a natural adjacency, locking in government clients before the infrastructure physically exists. Investment teams should note that Microsoft's FY2026 Azure capex guidance of ~$80B+ includes line items for "next-generation compute infrastructure" that analysts have not fully decomposed.

• Thermal Management and Power Architecture Breakthroughs for Microgravity Compute (research-stage; 12–24 month commercialisation horizon) A persistent engineering barrier for orbital data centres is heat rejection: in microgravity, convective cooling is absent, and radiative cooling requires large deployable surfaces that add mass and complexity. Researchers at NASA's Glenn Research Center and MIT Lincoln Laboratory have published work in 2025–2026 on two-phase loop heat pipe systems and deployable radiator architectures specifically designed for high-density compute in orbital environments. Separately, Nvidia's H100/H200 GPU power density (~700W per chip) creates acute thermal challenges that current orbital thermal architectures cannot accommodate at scale — meaning the first generation of orbital data centres will likely be CPU-bound or use lower-power ASICs, not GPU clusters. This is a material constraint on the addressable use cases (archival, routing, and edge inference — not training) that investment teams should factor into TAM assessments.

• Orbital Solar Energy as a Compute Subsidy [HIGH] The intersection of Space-Based Solar Power (SBSP) programmes and orbital data centre energy economics represents a structural disruption to terrestrial data centre energy cost models. The UK's Space Energy Initiative (targeting a 2035–2040 demonstration), JAXA's ongoing SBSP research, and the ESA ASCEND study all converge on the same insight: orbital compute powered by unobstructed solar flux could achieve near-zero marginal energy cost per compute cycle, fundamentally undermining the energy cost moat of terrestrial hyperscalers in markets where electricity prices are high or carbon constraints are tightening. Incumbents at risk: Terrestrial colocation operators (Equinix, Digital Realty) in high-electricity-cost markets (EU, Japan, Singapore). Potential winners: Vertically integrated orbital operators (Lonestar, future ESA-backed consortia, and any SpaceX orbital compute venture). KPIs to monitor: (1) EU electricity price trajectory for industrial consumers (€/MWh industrial benchmark); (2) ESA/EC procurement tender issuance dates for orbital compute; (3) Starship $/kg actuals vs. BryceTech projections.

• Sovereign Data Jurisdiction Arbitrage — Regulatory Moat Formation [HIGH] No existing international treaty framework (including the Outer Space Treaty of 1967 or the Moon Agreement of 1979, which the US has not ratified) clearly assigns national jurisdiction to data stored on orbital platforms or the lunar surface. This legal ambiguity is simultaneously a risk and a moat: the first operator to establish a commercially operational orbital or lunar data centre and secure a legal opinion from a major jurisdiction (US, EU, or UK) on data residency will have a first-mover regulatory moat that could persist for years before legislative clarity emerges. Incumbents at risk: EU cloud providers operating under GDPR data residency requirements (who currently benefit from regulatory complexity as a barrier to US hyperscaler entry) — orbital data centres could bypass GDPR residency rules entirely. Potential winners: Lonestar Data Holdings; any operator with a US Article VI national authorisation for commercial space activities. KPIs to monitor: (1) US FCC or DOC commercial space authorisation filings for orbital compute payloads; (2) EU Commission statements on GDPR applicability to orbital data; (3) Lonestar customer contract announcements.

• Latency Physics as a Structural Ceiling on Orbital Compute Use Cases [MEDIUM] A frequently underappreciated constraint: a server in LEO (400–600 km altitude) has a minimum round-trip latency of ~5–10ms to ground, rising to 270ms+ for GEO. This permanently excludes orbital data centres from real-time transactional compute (financial trading, gaming, interactive AI inference) and limits viable use cases to archival storage, batch processing, AI training (if power economics work), and deep-space relay. This is not a disruption signal in the traditional sense — it is a ceiling signal that defines the addressable market. Incumbents at risk: Any orbital data centre business plan that projects hyperscaler-equivalent TAM without accounting for latency constraints is structurally overvalued. Potential winners: Operators who correctly scope to archival, sovereignty, and batch workloads — Lonestar's lunar positioning is actually well-aligned with this constraint. KPIs to monitor: (1) Published latency benchmarks from any orbital compute demonstration mission; (2) Customer use case disclosures in Lonestar or competitor commercial agreements.

• Radiation Hardening and Fault Tolerance — The Hidden Moat [MEDIUM] Commercial off-the-shelf (COTS) compute hardware degrades rapidly in the radiation environment of LEO and beyond due to single-event upsets (SEUs) and total ionising dose (TID) effects. Radiation-hardened compute (rad-hard ASICs, ECC memory architectures, triple-modular redundancy) is currently manufactured by a small number of specialists: BAE Systems (RAD750, RAD5545 processors), Microchip Technology (RTG4 FPGA), and Renesas (formerly Intersil). Any orbital data centre operator must either pay the significant cost premium for rad-hard components or accept higher fault rates with software-layer redundancy. This creates a component supply moat for incumbent rad-hard manufacturers and a systems integration moat for defence primes with space heritage. Incumbents at risk: Commodity cloud hardware vendors (Dell, Supermicro) who have no space-qualified product lines. Potential winners: BAE Systems Space, Microchip Technology, and any systems integrator (Thales Alenia Space, Northrop Grumman) with existing rad-hard supply chain relationships. KPIs to monitor: (1) COTS vs. rad-hard cost differential trend (currently ~10–100x premium for rad-hard); (2) New entrant announcements in radiation-hardened GPU/AI accelerator space.

Strengthening Moats:

• SpaceX is extending a structural moat that is difficult to overstate: Starship's cost-per-kilogram trajectory, combined with Starlink's existing orbital infrastructure and SpaceX's vertical integration across launch, satellite manufacturing, and ground systems, means that any orbital data centre operator will likely depend on SpaceX as a launch provider for the foreseeable future. This creates a toll-road dynamic — SpaceX captures value regardless of which orbital compute operator wins commercially. The moat is strengthening because no credible competitor (New Glenn, Neutron, Ariane 6) can match Starship's payload economics within the 2026–2030 window relevant to first-generation orbital data centre deployment.

• Thales Alenia Space and Airbus Defence & Space are strengthening their moats in the European sovereign orbital compute segment by virtue of their ASCEND study participation, their existing relationships with ESA and the European Commission, and their decades of space systems integration heritage. For any EU-funded orbital data centre programme, these primes are structurally advantaged in procurement — a regulatory and relationship moat that US hyperscalers cannot easily replicate.

Eroding Moats:

• Terrestrial colocation operators (Equinix, Digital Realty, NTT Global Data Centers) face a long-duration but structurally real threat to their energy cost moat. Their competitive advantage in high-density compute markets is partially predicated on access to cheap, reliable power — a moat that erodes as orbital solar economics improve and as terrestrial electricity prices rise under carbon transition pressures. This is a 10–15 year erosion dynamic, not a 2–3 year cliff, but the trajectory is directionally negative for their long-term pricing power in energy-constrained markets.

• AWS GovCloud and Microsoft Azure Government face a specific moat erosion risk in the sovereign/classified data market if orbital data centres establish a credible jurisdictional arbitrage proposition. Both have invested heavily in physical security and compliance infrastructure for government clients — but neither can offer the physical inaccessibility and jurisdictional novelty of a lunar or deep-orbital data store. This risk is currently LOW probability but HIGH impact if it materialises.

Emerging Moats:

• Jurisdictional first-mover advantage is a genuinely new moat category that did not exist 12 months ago in a commercially actionable form. The operator who achieves the first commercially operational orbital or lunar data centre, secures a national authorisation, and establishes a legal framework for data residency in space will hold a regulatory moat that could persist for 5–10 years before legislative frameworks catch up. Lonestar Data Holdings is currently the most advanced claimant to this position, though the moat is not yet locked — it requires a successful Mission 2 deployment and at least one anchor commercial contract.

• Thermal management IP is emerging as a defensible technical moat for the small number of research groups and companies developing microgravity-optimised heat rejection systems. NASA Glenn's two-phase loop heat pipe work, if commercialised through a SBIR/STTR spinout, could create a component-level moat analogous to the cooling IP moats held by companies like Vertiv in terrestrial hyperscale cooling.

1. Track Lonestar Data Holdings' Mission 2 Payload Commitment and Customer Disclosure — Lonestar's Mission 2 lunar deployment (targeting 2027) represents the first commercially meaningful test of the lunar data sovereignty proposition. Investment teams should monitor: (a) whether Lonestar secures a named anchor customer (particularly a financial institution or sovereign wealth fund) before end-2026; (b) the launch vehicle selection (currently expected to be Intuitive Machines' IM-3 or a competing lunar lander); and (c) any regulatory filings with the US Department of Commerce's Office of Space Commerce regarding commercial lunar data operations. A named anchor customer would be a high-conviction signal that the jurisdictional arbitrage proposition has cleared legal due diligence at an institutional level. Signal that would change this recommendation: A definitive legal ruling by a major jurisdiction (US, EU, UK) that GDPR or equivalent data residency laws apply to orbital/lunar data stores — this would neutralise the core value proposition.

2. Assess the ESA ASCEND Procurement Pipeline for European Industrial Prime Exposure — The translation of ESA ASCEND feasibility outputs into active procurement tenders (expected Q3–Q4 2026 under Horizon Europe / CASSINI) represents a capital allocation event for European aerospace primes. Investment teams with exposure to the European aerospace and defence sector should evaluate the technology differentiation trajectory of Thales Alenia Space, Airbus Defence & Space, and CloudFerro in the context of orbital compute system integration. The key signal to monitor is the issuance of an ESA or European Commission call for tender specifically referencing orbital data infrastructure — this would trigger a multi-year procurement cycle with significant industrial content. *Signal that would change this recommendation: ESA