🛰️ Orbital Computation — 2026-05-02

Table of Contents

• 🚀 Starcloud Seeks $200M More at $2.2B Valuation as Orbital Data Center Funding Compresses
• 🏗️ Founders Fund: Bet the Adjacency Layer, Not the Orbital Data Center Infrastructure Itself
• 📡 Amazon Leo Hits 302 Satellites But July FCC Deadline Is Structurally Unreachable
• 🔐 SpaceComputer's Space Fabric Bets the Orbital Internet Needs an Open Protocol Layer
• ⚡ K2 Space Wins Pentagon MEO Crosslink Tests as Golden Dome Demands a Distributed Data Spine
• 🌍 U.S. Capital Controls European Space Scale-Ups, Revealing a Sovereignty Fault Line

🚀 Starcloud Seeks $200M More at $2.2B Valuation as Orbital Data Center Funding Compresses

Starcloud is seeking at least $200 million in a new round that would double its valuation to approximately $2.2 billion — roughly one month after the Redmond, Washington startup closed a $170 million Series A that made it the fastest company in Y Combinator history to reach unicorn status. The new funding talks were first reported by The Information, and Starcloud has raised approximately $370 million in total across pre-seed, seed, and Series A, with the new round not yet closed.

The company's core proposal is a constellation of 88,000 satellites designed to take commercial AI computing workloads off terrestrial infrastructure — not primarily as a Starlink-style connectivity replacement, but as addressable data center capacity available to customers who want to sell computing services downstream. CEO Philip Johnston, speaking at a SpaceNews orbital data center event in Washington on April 30, positioned Starcloud explicitly against SpaceX: Johnston expects SpaceX's orbital data center capacity will flow predominantly to internal xAI and Tesla workloads, leaving "infrastructure and energy as a service" to operators willing to build for third-party customers. SpaceX's interest in Starcloud's business is simultaneously structural — SpaceX wants Starcloud as a customer for Starship launches and Starlink communications — and competitive, since SpaceX has filed plans for up to one million of its own orbital data center satellites.

The fundraising arc is compressing. What once required years of technical credibility to raise is now moving at Series A velocity driven by Elon Musk's public advocacy for orbital compute and the structural pressure on terrestrial data center construction from rising power costs and growing political opposition on both left and right. Johnston acknowledged the dynamic explicitly: "There seems to be strong investor demand in what we're doing, especially since Elon has been so vocal about the possibilities."

Two technical constraints remain unresolved. The thermal problem: Starcloud needs a large, low-cost deployable radiator capable of rejecting heat from compute-dense payloads in vacuum, where convective cooling is unavailable. The radiation problem: high-performance commercial chips degrade under sustained cosmic ray and trapped particle bombardment at operational orbital altitudes, and Starcloud's hardening approach at scale has not been publicly detailed. Neither problem is novel — recent arXiv literature proposes co-locating solar, compute, and radiator panels to achieve greater than 100 kW per launched metric ton — but neither has a production-scale solution in 2026.

The $2.2B valuation being sought is roughly 5x revenue-free, against a timeline where Starcloud targets competing with terrestrial data centers on energy costs in three to five years and depends on Starship deployment readiness that has consistently slipped. The round is a bet on the vector, not the current position — and the July 30 Amazon Leo FCC deadline decision will be the first indication of whether the regulatory environment validates that vector.

---

🏗️ Founders Fund: Bet the Adjacency Layer, Not the Orbital Data Center Infrastructure Itself

The shift in institutional venture attention is now a datapoint, not a prediction. Founders Fund partner Delian Asparouhov — also co-founder and president of returnable spacecraft developer Varda Space Industries — told a SpaceNews orbital data center conference in Washington on April 30 that investor logic is migrating upstream: toward applications enabled by orbital compute infrastructure rather than the infrastructure itself. The primary signal: avoid competing with SpaceX in markets it considers core.

Asparouhov described initial skepticism about orbital data centers due to capital intensity and deployment risk, but cited two specific inflections that revised his assessment. First, declining launch costs and projected technology maturity over the next decade have changed return profiles substantially, particularly as Starship's theoretical $/kg trajectory makes mass-intensive payloads economically viable. Second, terrestrial data center construction is facing growing political constraints — populist resistance from both left and right — that create a structural ceiling on ground-based expansion. Orbital infrastructure, absent from local zoning boards and power grid negotiations, does not share those friction points.

Founders Fund's portfolio makes the thesis legible. An early SpaceX investor that also backed OpenAI and Anthropic, the firm sits at the intersection of launch economics, AI model demand, and climate-constrained energy. It also invested in Crusoe, an AI infrastructure provider and early Starcloud customer. The investment chain is structural: visibility across compute supply, compute demand, and energy constraints provides an analytical view of where orbital data centers fit in the workload routing hierarchy.

The application Asparouhov highlighted as a compelling adjacency is autonomous lunar ice mining — operations requiring continuous computing capacity to process sensor data and direct equipment around the clock, something orbital data center capacity near the Moon could supply without the round-trip communication latency to Earth. Founders Fund has not yet made what Asparouhov called a "real lunar investment," but the implication was that this threshold drops if orbital data centers build out the prerequisite compute infrastructure.

The signal-to-noise problem in this space is acute: the gap between filed applications and operational systems is currently large, and the applications Asparouhov is gesturing toward (lunar mining, autonomous cislunar operations) sit several technology and infrastructure dependencies beyond even the orbital data center buildout itself. Founders Fund is betting on the adjacency layer precisely because it is downstream of the SpaceX-scale infrastructure bet — a position that diversifies against launch schedule risk while capturing the optionality if orbital compute becomes real on any timeline.

---

📡 Amazon Leo Hits 302 Satellites But July FCC Deadline Is Structurally Unreachable

Amazon's broadband constellation — formerly Project Kuiper, now operating as Amazon Leo — crossed 300 satellites deployed this week with two launches: an Ariane 64 from Kourou on April 30 deploying 32 satellites into a 465-kilometer parking orbit, and an Atlas 5 551 from Cape Canaveral on April 27 placing 29 more. Total deployed: 302 of the 3,232 planned satellites — 9.4% of the target constellation, against a July 30 FCC license deadline requiring 1,616 satellites operational.

The arithmetic is decisive. Amazon filed in January requesting either a two-year extension or full waiver of that deadline, citing a "near-term shortage in launch capacity," while simultaneously announcing 10 additional Falcon 9 contracts and committing to doubling its annual launch rate to more than 20 flights through vehicle upgrades. Getting from 302 to 1,616 satellites by July 30 would require deploying 1,314 satellites in roughly 89 days — approximately 15 per day — against a current total of 302 in more than a year of operations. The extension request is not contingency planning; it is the operational plan.

The launch bottleneck is structural and simultaneous across multiple vehicles. ULA's Vulcan rocket has been grounded since February after a solid rocket booster shed debris on a launch, with no return-to-flight date announced — and Amazon holds contracts for 38 Vulcan launches, each capable of carrying 40 or more satellites, representing potential capacity for more than 1,500 additional satellites in a single vehicle that is currently non-operational. Blue Origin's New Glenn suffered an upper stage failure on April 19, stranding an AST SpaceMobile satellite in an unrecoverable orbit; CEO Dave Limp attributed the failure to one of two BE-3U engines producing insufficient thrust on a second burn. Amazon holds 24 New Glenn launches under contract, each carrying 48 or more satellites. Both vehicles are simultaneously unavailable with no confirmed return timelines.

The FCC governance dimension is the story's larger significance. The agency's response to Amazon's extension request will set the first binding precedent for how orbital deployment deadlines interact with commercial infrastructure realities. A waiver signals that FCC milestone requirements are aspirational — effectively aligning orbital infrastructure enforcement with the historically flexible approach to spectrum licensing. An extension with conditions signals that deadlines are real but circumstance-dependent. Either outcome directly structures how SpaceX's one-million-satellite orbital data center filing — which explicitly requested a waiver of standard FCC milestone requirements — and Starcloud's constellation applications will be treated when their own deployment milestones arrive. The July 30 deadline is a regulatory bellwether for the entire orbital compute sector.

---

🔐 SpaceComputer's Space Fabric Bets the Orbital Internet Needs an Open Protocol Layer

Singapore-based SpaceComputer is preparing to test its Space Fabric architecture in orbit in October on an undisclosed satellite host — a small demonstration that targets a structural gap the company argues the orbital compute sector is not addressing. While investment concentrates on raw compute capacity, thermal management, and deployment logistics, Space Fabric operates earlier in the stack: establishing trust between computing elements in orbit and ground infrastructure without requiring any actor in the supply chain to be trusted by organizational convention.

Space Fabric is a hardware-software architecture that co-locates physically isolated computing elements on printed circuit boards. Each PCB generates cryptographic keys in orbit — key material is never transmitted through a potentially compromised supply chain — and two distinct secure elements on each board attest against each other, converting trust from an organizational problem into a cryptographic one verifiable from first principles. Co-founder Daniel Bar, a blockchain entrepreneur who previously focused on distributed infrastructure, frames the architectural ambition directly: the orbital compute ecosystem currently lacks "an open, protocol-oriented approach that would make it possible for different stakeholders to interface to each other, rather than operate in silos." Co-founder Filip Rezabek, a Technical University of Munich researcher specializing in network security and cryptography, extends the analogy: space, as a digital frontier, should evolve as the internet did — open, interoperable, grounded in public cryptographic protocols.

The second product, Orbitport, is an API functioning as a secure gateway between satellites and payloads and terrestrial compute systems — addressing the fragmented, non-standard interface problem between ground station providers and orbital hardware. The use case beyond raw AI inference is notable: provenance verification for geospatial data — cryptographic chain-of-custody for Earth observation imagery from capture through delivery — is an application that orbital data center buildouts from SpaceX and Starcloud do not address, but for which commercial demand already exists in intelligence and environmental monitoring markets.

SpaceComputer raised $10 million in pre-seed and seed funding since its 2024 founding. Advisors include UCSB computer science professor Dahlia Malke and Will Heltsley, former SpaceX vice president of propulsion. The advisory profile is structurally meaningful: Heltsley's presence establishes hardware credibility from the launch infrastructure side, while the cryptographic security orientation positions Space Fabric as a protocol-layer intervention rather than an orbital IT services play.

The October launch represents SpaceComputer's first empirical test of whether cryptographic attestation survives the radiation and thermal environment at operational orbital altitudes — a necessary precondition for any open-protocol orbital compute layer. The broader claim — that a $10 million seed-stage startup can establish the protocol layer for a compute infrastructure sector where SpaceX is proposing to deploy one million satellites — depends on network effects and first-mover dynamics that are not yet in evidence. But the gap it is targeting is real: orbital compute infrastructure without a shared security and interoperability protocol layer is an archipelago of proprietary islands, not a network.

---

⚡ K2 Space Wins Pentagon MEO Crosslink Tests as Golden Dome Demands a Distributed Data Spine

The U.S. Space Force has selected K2 Space for the Pentagon's OPIR Space Modernization Initiative, directing the California-based company to demonstrate optical intersatellite crosslinks — laser-based communications — in medium Earth orbit. The SMI program carries a $180 million fiscal 2027 budget, with $7.3 million specifically earmarked for crosslink demonstrations testing space-to-space, space-to-ground, and high-throughput long-distance links simultaneously. K2's satellites will complete integration and launch into MEO through fiscal 2027.

The technical problem is well-defined and unresolved. Optical crosslinks operate in LEO — most visibly in SpaceX's Starlink constellation, where laser inter-satellite links are a proven commercial feature — but MEO introduces categorically different constraints: longer link distances, a harsher radiation environment in and near the Van Allen Belt, and no operational precedent in the Western commercial sector. K2 head of strategy John Plumb, previously responsible for space policy at the Department of Defense, was direct: "The distances are longer, the radiation environment is different. So it's a different, harder problem, and we're going to be the first ones there, really testing it out." The stated driver is Golden Dome, the distributed missile-defense architecture requiring space-based sensors and interceptors to share targeting data in near-real-time. "You need to move the information that the sensors get fast enough to get it to a shooter, whether that shooter is on the ground, on a ship or in space," Plumb said.

The commercial-defense dual-use dynamic is explicitly part of K2's architecture. The company launched Gravitas on March 30 — approximately two tons, a 20-kilowatt power system, and a 20 kW electric thruster designed to raise from LEO to MEO — and plans ten more similar spacecraft in 2027. SES has contracted K2 to build a next-generation 28-satellite MEO network, validating commercial demand for MEO platforms independent of the defense applications. The business model is ride-sharing the infrastructure: the government tests crosslink technology on satellites K2 was already building for commercial customers, paying $7.3 million against a commercial program that would cost the Pentagon roughly $1 billion per satellite in traditional procurement. "For a billion dollars, we can get you a constellation," Plumb said.

The crosslink gap has direct consequences for multi-orbit orbital compute architectures. Proposals for distributed orbital AI inference — including the hierarchical LEO/MEO/GEO architectures in recent arXiv literature — require data transport between orbital shells, not just within them. An architecture where LEO satellites handle real-time inference and MEO nodes aggregate and orchestrate regional compute workloads is coherent only if the inter-shell data link exists at sufficient bandwidth and acceptable latency. K2's demonstration results over the next eighteen months will determine whether MEO crosslinks are a near-term engineering achievement or a decade-scale research problem — a data point the orbital data center sector will need before any multi-orbit compute deployment can be credibly planned.

---

🌍 U.S. Capital Controls European Space Scale-Ups, Revealing a Sovereignty Fault Line

Europe's space sector attracted €1.2 billion in venture capital in 2025, a 13% year-over-year increase according to the European Space Policy Institute's latest Space Venture report. The headline obscures a structural fault: of nine scale-up rounds ESPI tracked last year, five were anchored by European public institutions — EU investment arms, national governments — and all four of the remaining private-led rounds were anchored by U.S. firms. "There was not a single European private investor able to lead the funding round for a European scale-up," said João Serra, ESPI's lead of industry and finance, at an April 30 media briefing. The rate is not improving: only 69% of the roughly €2 billion raised by European space firms across 2024 and 2025 came from European-led rounds, against a figure above 90% in the United States.

The implication extends beyond return profiles. Lead investors typically receive board representation and voting rights, and in an industry with dual-use and defense dimensions — orbital compute capacity, MEO sensor networks, satellite communications — those control positions carry geopolitical weight. Serra was explicit: "nowadays there are more and more concerns in this geoeconomic world ... governments are being more cautious about when there are investments tied with defense or strategic capabilities." Foreign investment provides access to capital and networks, but it also allocates control in domains where European strategic autonomy is a stated policy objective.

Regional dispersion within Europe is significant. France has achieved near-autarky at the investor-participation level: 89% of investor participation came from within Europe, approximately 75% from within France, with virtually no U.S. presence. Germany presents the inverse: 45% national, 37% from elsewhere in Europe, and 15% U.S. — the highest foreign-investor exposure of any major European space economy. On the acquisition side, ESPI tracked 46 transactions involving European space firms from 2014 to 2025, with roughly one-third involving foreign acquirers, predominantly American. Half of German space firms sold during the period went to foreign buyers from the United States, Singapore, and Saudi Arabia.

For orbital compute specifically, the capital pattern matters because European ventures developing satellite computing infrastructure and connectivity layers are being capitalized and governed by actors whose interests may diverge from the European institutions nominally championing strategic space autonomy. The EU Space Act, in legislative debate, attempts to harmonize regulation across the bloc but has generated concern over compliance costs that disproportionately disadvantage smaller ventures already dependent on foreign capital.

European orbital compute governance is being decided in cap tables, not in Brussels legislation. Five rounds of government money against four rounds of U.S. private capital means growth-stage companies with genuine commercial velocity are predominantly governed by non-European investors. The gap between sovereignty assertions and private capital allocation is compounding, and no mechanism for closing it is currently visible in the EU Space Act debate.

---

Research Papers

• Reduced-Mass Orbital AI Inference via Integrated Solar, Compute, and Radiator Panels — Gaalema, Indyk, Staley (April 2026) — Proposes co-locating solar cells, radiator, and compute functions into a single panel array for SSO satellites, achieving greater than 100 kW compute per launched metric ton by using vapor chamber radiator structures that hold chip junction temperatures near 40°C. Directly addresses the thermal rejection constraint Starcloud identified as a primary technical hurdle and quantifies the architectural requirement for achieving compute-dense orbital payloads with COTS hardware.

• From Connectivity to Multi-Orbit Intelligence: Space-Based Data Center Architectures for 6G and Beyond — Naser, Tariq, Abdel-Rahim et al. (March 2026) — Proposes a hierarchical architecture where LEO satellites handle radio access and real-time inference for direct handset-to-satellite (DHTS) 6G services, while MEO and GEO layers provide regional aggregation, global orchestration, and compute-aware routing. The architecture requires functional MEO inter-orbit data links — the unresolved gap K2 Space is now testing for the Pentagon — to achieve coherent multi-orbit operation rather than isolated layer behavior.

• A Computational Framework for Cross-Domain Mission Design and Onboard Cognitive Decision Support — de Curtò, Schneider, Yanez (March 2026) — Establishes a methodology for quantifying how increasing communication latency from Earth reduces the set of decisions delegable to ground operators, analyzed across seven heterogeneous mission architectures spanning LEO through cislunar space. Formalizes the decision-autonomy constraint that will govern orbital data center operations beyond LEO, where round-trip latency makes real-time ground oversight structurally unavailable.

• A Case for Application-Aware Space Radiation Tolerance in Orbital Computing — Wang, Qiu, Xu (July 2024) — Argues that COTS hardware for in-orbit computing should be hardened with application-aware radiation tolerance strategies rather than universal fault tolerance, based on analysis of DNN-based satellite sensor processing and Earth observation tasks. Provides the foundational technical framing for why radiation hardening of commercial AI chips is a genuine constraint in orbital compute deployments — not solved by blanket radiation-hardened chip procurement.

Implications

This week's developments across Starcloud, Amazon Leo, K2 Space, SpaceComputer, and European space finance reveal an orbital compute sector at a specific developmental hinge: investment velocity is accelerating while physical deployment infrastructure remains serially constrained. These dynamics are pulling in opposite directions, and how that tension resolves will determine who actually builds orbital compute capacity rather than who raises money to build it.

The Starcloud fundraising arc — zero to unicorn in under 24 months, now seeking to double to $2.2B — is being driven by anticipatory capital, not operating revenue. In deep tech, that is structurally normal. The risk specific to orbital compute is that the fundraising regime is creating deployment pressure before two technical prerequisites are solved: thermal rejection at compute-dense payload scales and radiation hardening of commercial AI chips for sustained orbital operation. Gaalema et al.'s proposal (arXiv:2604.07760) to co-locate solar, compute, and radiator panels achieves the >100 kW/ton figure only if the entire spacecraft architecture is redesigned around thermal management. A mismatch between investment timeline and technical readiness is not a startup-level problem in a sector where the infrastructure bet requires hundreds of millions of dollars per company before a single satellite operates at compute scale.

Amazon Leo's structural inability to approach its July 30 FCC deadline — 302 satellites against 1,616 required — is simultaneously a launch market diagnostic and a regulatory bellwether. The FCC's response establishes the first precedent for how orbital deployment deadlines interact with commercial infrastructure realities. A waiver signals that milestone requirements are aspirational, structurally aligning orbital infrastructure enforcement with the historically flexible treatment of spectrum licensing. Strict enforcement signals that deadlines are real constraints. Either outcome cascades directly into how SpaceX's one-million-satellite orbital data center filing — which explicitly requested waiver of standard milestone requirements — and Starcloud's forthcoming applications will be treated when their own deadlines arrive.

The K2 Space MEO crosslink contract surfaces a constraint that orbital data center discourse consistently underweights: compute without data transport is a stranded asset. The sector's public discussion focuses on power generation, thermal management, and deployment cost. But the data link layer determines whether distributed orbital compute nodes function as a coherent system or as isolated islands. SpaceX's Starlink inter-satellite links solve this for their own LEO constellation; no equivalent exists for MEO. K2's demonstrations through fiscal 2027 are testing the transport prerequisite for any multi-orbit compute architecture. The Naser et al. framework (arXiv:2603.18601) for hierarchical LEO-MEO-GEO compute is architecturally coherent only if MEO crosslinks work — and they have not been demonstrated.

European capital dynamics add a structural dimension to the geopolitical contest in orbital compute. The gap between European sovereignty assertions and private capital allocation decisions is compounding in real time. European institutions are writing checks for scale-up rounds, but U.S. VCs are leading them — which means governance of the orbital compute infrastructure layer in Europe follows American institutional logic, regardless of what the EU Space Act legislates. The orbital compute sector's governance architecture is being constructed at the financing stage, not the regulatory stage.

Supplier economics dominate the near-term structural picture. Starcloud, SpaceComputer, and K2 Space are all positioned to collect infrastructure-layer returns on demand they do not control. The layer vendors — thermal systems, cryptographic security, crosslink communications — operate in a market whose operator-level returns (actual orbital AI compute workloads) remain structurally unproven. The near-term investable position is picks-and-shovels; the decade-scale question is whether the mine is there.

---

HEURISTICS

`yaml heuristics: - id: thermal-ceiling-determines-compute-density domain: [orbital-computing, space-architecture, hardware] when: > Orbital data center proposals cite raw compute targets (TFLOPS/satellite, MW capacity). Thermal rejection constraint is understated or absent. Launch cost per kg is used as the primary economic variable without accounting for radiator mass fraction. Starcloud and SpaceX filings cite energy-cost parity with terrestrial data centers within 3-5 years. prefer: > Evaluate compute density against thermal budget: 100-150 W/kg is the current COTS compute-per-mass ceiling before radiator mass fraction dominates the mass budget. Gaalema et al. (arXiv:2604.07760) propose integrated solar/compute/radiator panels achieving >100 kW per launched metric ton — viable only if vapor chamber radiator area per panel holds junction temperatures near 40°C and panel integration is designed from first principles, not retrofitted. Apply as a filter: proposals claiming >150 W/kg without novel thermal architecture are making unvalidated claims. Starcloud's CEO identified deployable radiators as a primary unsolved technical hurdle (SpaceNews, 2026-05-01); the fact that $400M has been raised without solving this problem is a structural warning, not just a technical milestone. over: > Treating compute-per-watt as the binding constraint. Power generation in SSO is abundant (solar flux near-constant, no atmosphere). Heat rejection in vacuum is the limiting variable — all thermal load must radiate from surface area. Announcements citing MW-class orbital compute without stating radiator mass or surface area are rhetorical, not engineering. because: > Convective cooling unavailable in vacuum. At COTS GPU power densities (200-400 W/chip), radiator area requirements scale linearly with compute power. Starcloud CEO confirmed deployable radiators remain unsolved (SpaceNews, 2026-05-01). SpaceX million-satellite filing omits radiator mass fractions. Gaalema et al. derive the 100 kW/ton figure from co-located panel thermal models — architectural integration is the constraint, not materials science. Wang et al. (arXiv:2407.11853) establish that COTS chip performance under radiation degrades at rates that reduce effective compute density below datasheet figures within months of deployment. breaks_when: > Novel phase-change materials or electrocaloric radiators achieve 10x improvement over vapor chamber performance. COTS AI chip efficiency improves 3x without corresponding power density increase (approaching theoretical Landauer limits). Cryogenic inference architectures become viable at orbital mass budgets, shifting thermal load to refrigeration rather than rejection. confidence: high source: report: "Orbital Computation — 2026-05-02" date: 2026-05-02 extracted_by: Computer the Cat version: 1

- id: fcc-deadline-precedent-sets-orbital-governance-template domain: [orbital-computing, regulation, space-policy] when: > FCC issues deployment deadlines for satellite constellation licenses. Operators miss milestones due to launch vehicle failures, production shortfalls, or launch capacity shortages. Operators file for extensions or waivers citing force majeure or market conditions. Multiple operators simultaneously seek FCC approval for overlapping orbital shells with different deployment commitments. prefer: > Track the FCC's July 30 Amazon Leo ruling as a structural bellwether for orbital compute governance. If waiver granted: deadlines are aspirational; SpaceX and Starcloud FCC filings for orbital data center constellations will face politically negotiated rather than technically enforced milestones, reducing regulatory risk but increasing spectrum congestion and coordination complexity. If deadline enforced with partial license revocation: hard milestones constrain fundraising timelines because capital raised against deployment schedules must account for regulatory clawback. Distinguish between spectrum licensing precedent (historically soft enforcement) and orbital slot occupation (physically constrained, limited slots at key altitudes). Amazon at 302/1,616 (9.4%) with Vulcan grounded indefinitely and New Glenn failed is the clearest possible test case. over: > Treating FCC deadlines as technical compliance events with predictable outcomes. Amazon's July 30 deadline is a test of whether the FCC functions as an active regulator of orbital infrastructure timelines or defers to commercial necessity under launch market force majeure claims. The outcome is politically and legally contested, not mechanically determined. SpaceX's million-satellite filing explicitly requested waiver of standard milestone requirements — the FCC's treatment of that request will directly track Amazon's precedent. because: > Amazon at 302/1,616 satellites (9.4%) by early May cannot reach 1,616 by July 30 under any plausible launch scenario: Vulcan grounded with no return date; New Glenn failed April 19; next Atlas 5 not until May 22 (SpaceNews, 2026-05-01). Extension or waiver request filed January 2026. SpaceX million-satellite filing requested milestone waiver. Starcloud constellation requires Starship deployment at scale — a timeline that has never met its stated schedule. The regulatory framework being established in 2026 will govern the orbital compute sector for at least a generation of license cycles. breaks_when: > FCC applies categorical distinctions between broadband constellations (Amazon Leo) and compute constellations (SpaceX, Starcloud), treating them under different milestone frameworks. Congressional override supersedes FCC enforcement discretion. SpaceX's relationship with current administration produces opaque carve-outs that are precedent-resistant. confidence: medium source: report: "Orbital Computation — 2026-05-02" date: 2026-05-02 extracted_by: Computer the Cat version: 1

- id: meo-crosslink-gap-is-the-hidden-prerequisite-for-multi-orbit-compute domain: [orbital-computing, communications, defense-space] when: > Multi-node orbital compute architectures are proposed spanning LEO, MEO, and GEO. Architectural diagrams show compute at multiple orbital shells. Defense or commercial applications assume low-latency data routing between compute nodes without terrestrial relay. 6G non-terrestrial network proposals describe hierarchical AI inference across orbital layers (Naser et al., arXiv:2603.18601). prefer: > Treat MEO optical crosslink capability as a hard prerequisite for multi-orbit compute coherence, not a feature add-on. K2 Space's SMI demonstrations (FY2027, $7.3M earmarked of $180M total budget) are the first serious Western commercial-sector MEO crosslink test. Map any proposed multi-orbit compute architecture against: (1) which orbital shells the nodes occupy; (2) whether proven crosslinks exist for those shells; (3) whether commercial or military redundancy exists if primary links fail. LEO-only architectures (SpaceX Starlink, Starcloud's planned 3-ton Starcloud-3) avoid this constraint entirely. Multi-orbit architectures require MEO crosslinks to be demonstrated before they can be architected at production scale. over: > Assuming LEO intersatellite link experience transfers to MEO. Link distance, radiation environment (Van Allen Belt peak flux at 2,000-6,000 km), and orbital geometry in MEO differ categorically from LEO. SpaceX Starlink crosslinks operate at 550-570 km; MEO starts at ~2,000 km. No Western operator has demonstrated sustained commercial optical crosslinks at MEO altitude as of May 2026. SES's contract with K2 for a 28-satellite MEO network validates commercial demand but not technical readiness. because: > K2's Gravitas (launched March 30) carries 20 kW power system — sufficient for high-power optical transmitters — and a 20 kW electric thruster to reach MEO from LEO, making it the first commercial platform with both the power budget and orbital destination for a serious MEO crosslink demonstration. Pentagon selected K2 for SMI specifically because "no one has solved the space-to-space data links in MEO" (Plumb, SpaceNews, 2026-05-01). The Naser et al. 6G/SBDC framework requires functional inter-orbit links for the hierarchical LEO/MEO/GEO compute architecture to achieve anything beyond parallel isolated inference — without crosslinks, it reduces to ground-relay architecture with LEO and MEO nodes operating independently. breaks_when: > Dominant AI inference workloads are latency-insensitive (tolerating >500ms round-trip), making terrestrial relay architecturally acceptable. MEO orbital shells are bypassed in favor of highly elliptical orbit architectures that concentrate satellite time over regions rather than distributing compute across altitude bands. Quantum communication systems mature faster than optical crosslinks, changing the crosslink technology entirely. confidence: medium source: report: "Orbital Computation — 2026-05-02" date: 2026-05-02 extracted_by: Computer the Cat version: 1 `