Orbital Computation · 2026-04-27

🛰️ Orbital Computation — 2026-04-27

🛰️ Amazon's $11B Globalstar Buy + FCC D2D Spectrum Lock Completes Two-Layer LEO Stack
📡 New Glenn Failure Strands AST SpaceMobile's BlueBird-7 Two Days After FCC Constellation Clearance
🔭 Space Force 50% Heavy-Lift Surge Reveals National Security Orbital Compute as Primary Launch Driver
🇨🇳 China Flies 26th Mission of 2026, Launching D2D Internet Test Satellites While West Debates Spectrum
🔩 Northrop's $71M Vulcan Anomaly Hands SpaceX Near-Monopoly on Heavy Orbital Delivery
💰 FAA $0.25/lb User Fees Mark First Structural Levy on Constellation Economics, Scaling to $1.50/lb by 2033

---

🛰️ Amazon's $11B Globalstar Buy + FCC D2D Spectrum Lock Completes Two-Layer LEO Stack

Amazon's April 14 announcement to acquire Globalstar in a deal valued at approximately $11 billion marks the company's decisive move into the spectrum layer of orbital infrastructure — a layer it previously lacked entirely. The acquisition delivers three assets no satellite broadband contract can provide: globally licensed MSS spectrum, an existing Apple service agreement covering iPhone Emergency SOS and Find My, and a next-generation D2D constellation (50+ satellites under construction by MDA) slated to come online in 2028. The deal closes the architectural gap between Amazon Leo's planned broadband service and the direct-to-device tier SpaceX has spent two years building.

The FCC's April 23 order dismissing SpaceX's, Iridium's, and Kepler's spectrum bids for the Big LEO MSS bands makes Amazon's acquisition price retroactively comprehensible. By moving nine days before the FCC locked down incumbents, Amazon purchased spectrum protection it could never have obtained through regulatory petition. The FCC ruling preserves market certainty for existing licensees against interference from new entrants, citing "ubiquity and portable nature of mobile devices" and "significant harmful interference challenges" as rationale for protecting Globalstar, Iridium, and EchoStar positions. SpaceX's parallel EchoStar spectrum acquisition (AWS-3 uplink for $2.6B in SpaceX equity; the earlier $17B AWS-4/H-block purchase) gives it a comparable two-layer stack: Starlink broadband plus terrestrial-cellular D2D frequencies.

The structural picture that emerges is a duopoly governed not by satellite count but by spectrum ownership. Amazon controls Globalstar's L-band and Big LEO allocations plus the Apple partnership that gives it 1B+ device endpoints. SpaceX controls Starlink plus EchoStar's AWS spectrum plus T-Mobile's cellular partnership. Every other player — AST SpaceMobile (cleared by FCC April 21 but limited to AT&T/Verizon partner frequencies), Sateliot (dismissed by FCC for U.S. market access), Kepler (dismissed) — operates within a spectrum framework those two companies now effectively define. The FCC's order explicitly terminated an inquiry into EchoStar's 2 GHz spectrum usage following its sale to SpaceX, removing the last regulatory overhang on Starlink's D2D buildout. The operational conclusion: orbital compute architecture in the D2D tier is settling into a two-node structure, and the window to join it at the spectrum-ownership level has closed.

Sources:

---

📡 New Glenn Failure Strands AST SpaceMobile's BlueBird-7 Two Days After FCC Constellation Clearance

The operational-rhetorical gap in LEO deployment collapsed into a single week for AST SpaceMobile. On April 21, the FCC cleared the company's full 248-satellite constellation — approving D2D broadband service via AT&T and Verizon cellular spectrum, covering both commercial users and FirstNet public-safety subscribers. Two days later, Blue Origin's New Glenn suffered a BE-3U engine underperformance on its NG-3 second burn, placing BlueBird-7 into a 154×494 km parking orbit at 36.1° inclination — 313 km below target, with an inclination 13.3° off-nominal and unrecoverable by onboard electric propulsion. The satellite will deorbit.

The loss of BlueBird-7 is material because Block 2 BlueBirds are not interchangeable with their predecessors. At 6,100 kilograms with a 223-square-meter communication array — 3.5× larger than Block 1 — each spacecraft delivers peak 120 Mbps directly to standard smartphones, supporting voice, video, and 5G services. The company needs 45–60 of these to achieve continuous coverage in key U.S. markets; with only BlueBird-6 (launched December on an Indian LVM3) operational at Block 2 spec, the setback extends timeline exposure. Launch insurance covers a fraction of the replacement cost per AST's March 2 SEC filing, with premiums running 3–20% of insured value.

New Glenn's anomaly record is now two significant failures in three flights. The first (NG-1) achieved orbit but the upper stage landing attempt failed; the third (NG-3) placed its payload in an off-nominal orbit via engine underperformance. Blue Origin successfully reflew a first-stage booster for the first time on NG-3 — a positive milestone for cost reduction — but the upper stage remains the critical reliability gap. This matters beyond AST SpaceMobile: Amazon has booked New Glenn for multiple Kuiper constellation launches including missions carrying ~48 satellites each. With the FCC's July 30 deadline requiring half of Kuiper's 3,232 first-generation satellites deployed — and Amazon currently at 212 — New Glenn's reliability curve is on the critical path for Amazon's D2D ambitions from the Globalstar acquisition. The FCC clearance and the launch failure are not independent events: the former's value depends entirely on the latter's solution.

Sources:

---

🔭 Space Force 50% Heavy-Lift Surge Reveals National Security Orbital Compute as Primary Launch Driver

The U.S. Space Force's April 2026 "sources sought" notice adding 25 high-energy Lane 2 missions to the National Security Space Launch program is a bellwether for the structural economics of the entire launch industry. The original Lane 2 plan called for 54 missions over five years; the addition brings the total to 79, a 46% increase concentrated in FY2027-2029 (six missions in 2027, nine in 2028, ten in 2029). The Pentagon's FY2027 budget request includes approximately $5 billion for 31 national security launches — more than double the roughly $2 billion enacted for FY2026. This is not a marginal increment; it represents a structural shift in who is setting orbital compute requirements.

The demand profiles are specific: direct insertion of 8,000-lb payloads into geosynchronous orbit, 20,000-lb to medium Earth orbit, and multi-manifest missions deploying multiple high-value spacecraft. These are not communications satellites. The Space Development Agency's missile-tracking and communications constellations — the Tranche program — require exactly the precision orbital insertion that defines Lane 2. Only SpaceX and ULA hold current Lane 2 certification, but ULA's Vulcan Centaur remains grounded (see Story 5). The effective single-source constraint means SpaceX absorbs the surge while the Space Force gauges whether Blue Origin can certify New Glenn for NSSL missions — a timeline further complicated by NG-3's upper stage anomaly.

The deeper implication: national security orbital compute, not commercial broadband, is now the primary near-term driver of heavy-lift launch allocation. Commercial constellations (Kuiper, Starlink Gen2, AST SpaceMobile) compete with classified payloads for rocket time on a vehicle manifest that is effectively one provider. The Space Force's dual-notice approach — adding missions to the existing Lane 2 contract vehicle while simultaneously gauging whether Blue Origin can qualify — is a hedge against single-source dependency at exactly the moment when military orbital compute demand is escalating fastest. The $5B annual spend level positions government as the dominant economic force shaping what gets built in heavy-lift, constraining the launch infrastructure that both commercial orbital compute and national security platforms depend on equally.

Sources:

---

🇨🇳 China Flies 26th Mission of 2026, Launching D2D Internet Test Satellites While West Debates Spectrum

China reached its 26th launch of 2026 over a single weekend in late April, with three separate missions illustrating the operational gap between Chinese orbital buildout and Western regulatory activity. The most strategically significant: a Long March 2D lifting four satellite internet technology test satellites from Xichang on April 24, explicitly tasked with "technical experiments and verifications on technologies such as direct broadband connection to satellites for mobile phones and the convergence of space and ground networks." This is operational D2D test infrastructure while the FCC is simultaneously dismissing Western D2D spectrum petitions and Amazon is spending $11B to acquire the spectrum position it cannot file for.

The second flight — a Long March 6 lofting Pakistan's PRSC-EO3 remote sensing satellite on April 25 — extends the geopolitical dimension. PRSC-EO3 is the third Chinese-launched Earth observation satellite for Pakistan under a multi-flight CGWIC service agreement; Pakistani astronaut candidates Muhammad Zeeshan Ali and Khurram Daud arrived in Beijing for Tiangong training the same day. Pakistan has already signed China's International Lunar Research Station agreement. The pattern is consistent: China is building a parallel orbital infrastructure stack in which Global South partners are integrated customers, not just spectators.

At 26 launches through April 27, China is on pace for approximately 78–80 launches in 2026, up from 68 in 2024. This cadence is directly relevant to orbital compute: China's Guowang/SatNet megaconstellation (13,000+ satellites planned) and private operators like SpaceSail and GalaxySpace are building compute-bearing LEO infrastructure on a timeline that runs parallel to, not behind, the Kuiper/Starlink buildout. The internet test satellites are the operational development layer for Chinese D2D — not announced products but flying test assets. The contrast with the FCC's April 23 incumbent-protection order is structural: U.S. regulatory architecture is optimizing for existing property rights in spectrum while China's launch rate demonstrates a build-first approach to the same connectivity layer. The governance bellwether is not which country has more spectrum filings or more regulatory petitions — it's which has more orbiting test hardware actively generating operational data.

Sources:

---

🔩 Northrop's $71M Vulcan Anomaly Hands SpaceX Near-Monopoly on Heavy Orbital Delivery

Northrop Grumman's April 21 disclosure of a $71 million Q1 2026 charge for its GEM 63XL solid rocket booster anomaly is the accounting crystallization of a structural problem: the United States now has effective single-source heavy-lift for both national security and commercial payloads. The GEM 63XL shed debris 65 seconds after a February 12 Vulcan Centaur mission — an anomaly that did not affect that payload's geosynchronous insertion but has grounded the vehicle indefinitely. Space Force has no return-to-flight timeline; Northrop's 10-Q cited "evaluation and implementation of corrective actions" with no completion date. ULA faces both the Vulcan grounding and a backlog of earlier National Security Space Launch Phase 2 missions.

The market consequence is immediate: every high-energy mission that would have gone to Vulcan now routes to Falcon Heavy. This includes the Space Force's Lane 2 expansion of 25 additional missions through 2029, Amazon Kuiper's high-mass batch launches, and commercial GEO telecommunications satellites. SpaceX has not had a Falcon Heavy anomaly in its operational history, but the dependency concentration has strategic consequences regardless of reliability. Vulcan's GEM 63XL without solid boosters configuration being considered by Space Force for low-energy missions — SDA missile-tracking and communications constellations — would restore partial launch capacity but cannot substitute for full-thrust Vulcan on the high-energy manifest. Blue Origin's New Glenn reliability record (NG-3's upper stage anomaly, April 19) further narrows the realistic alternatives.

The supplier economics are unambiguous: SpaceX is currently the only provider certified, flying, and reliable for heavy orbital delivery. This is not a competitive advantage — it is a single point of failure for the entire U.S. orbital compute infrastructure stack. Commercial LEO operators requiring heavy-lift (>10,000 kg to LEO) have no credible alternative launch provider. Government operators are formally constrained to Lane 2 certified vehicles, of which SpaceX is presently the only operational one. The gap between the Space Force's need — 25 additional high-energy missions between 2027 and 2029 — and available certified providers is not filled by any vehicle currently demonstrating consistent upper-stage performance. The orbital compute industry is built on a launch infrastructure whose second and third options are simultaneously unavailable.

Sources:

---

💰 FAA $0.25/lb User Fees Mark First Structural Levy on Constellation Economics, Scaling to $1.50/lb by 2033

The FAA's April 22 Federal Register notice initiating user fees for commercial launches represents a structural governance event whose significance is masked by the current fee level. The initial rate — $0.25 per pound of payload, capped at $30,000 per mission — is modest. A Starlink batch launch (estimated 14,400–16,700 kg payload) incurs approximately $8,000–$9,200. For SpaceX's 2025 launch total of approximately 180 Starlink missions, that works out to roughly $1.5M annually. At those numbers, the fee is a regulatory acknowledgment, not an economic constraint.

The escalation schedule converts it into something different. Under the budget reconciliation act's graduated structure, the fee reaches $1.50 per pound (capped at $200,000 per mission) by 2033, with CPI-linked growth thereafter. At that rate, a single Starlink launch costs ~$50,000 in fees; at projected growth to 250+ missions per year across U.S. commercial operators, the annual industry bill exceeds $12 million. There were 199 licensed launches and seven licensed reentries in 2025, dominated by SpaceX Starlink missions. The fees fund AST's work on integration of launches into the national airspace — a 43.3% FY2027 budget increase to $56.8M proposed to hire 70 additional staff to handle a 52.7% demand surge since FY2023. Compliance infrastructure costs that trail operational infrastructure costs are a characteristic pattern of maturing industries.

The second-order effect is competitive structure. Fee schedules that are flat per mission (capped at $200K) benefit high-mass operators — the per-kilogram cost to orbit falls as vehicle capacity increases, and the $200K cap means Starship launches (Starlink V3 at ~100 satellites each) pay the same fee as a Falcon 9 load. This creates a structural incentive for further concentration in high-mass launch vehicles and constellation operators, compounding the single-source problem visible in the Vulcan grounding. The FAA's current 136 AST staff handling 206+ annual commercial space events have not scaled with demand; the fee collection mechanism is as much a staffing problem as a revenue problem. The orbital compute buildout is generating regulatory overhead that the U.S. government is only now beginning to price — and the pricing structure, as designed, favors operators who can fill the largest vehicles most efficiently.

Sources:

---

Research Papers

Communication-Efficient Collaborative LLM Inference over LEO Satellite Networks — Zhang, Wu, Li, Wang & Shen (April 6, 2026) — Proposes pipeline-parallel LLM splitting across multiple LEO satellites, with adaptive activation compression reducing inference delay 42% and communication overhead 71% while maintaining <1% accuracy loss; addresses the fundamental constraint that no single LEO satellite can hold a production-scale LLM in onboard memory. Introduces a shortest-path formulation over a directed acyclic graph for model-splitting optimization.

Edge Intelligence for Satellite-based Earth Observation: Scheduling Image Acquisition and Processing — Soret, Mercado-Martínez, Jurado-Navas, Lyholm, Moretti, Popovski & Leyva-Mayorga (April 7, 2026) — Analyzes heterogeneous LEO constellations with onboard edge AI for scheduling image acquisition and processing under tight downlink and ground-processing constraints, directly relevant to time-critical observation (disaster response, maritime tracking) where round-trip latency to ground makes real-time onboard inference operationally necessary.

A Computational Framework for Cross-Domain Mission Design and Onboard Cognitive Decision Support — de Curtò et al. (March 30, 2026) — Introduces the Autonomy Necessity Score (ANS), a log-domain latency metric that maps satellite architectures along a ground-dependent to fully-autonomous compute continuum; spans LEO surveillance constellations, Mars orbital navigation, deep-space inter-satellite networks, and outer-planet buoy platforms. Provides a unified vocabulary for comparing orbital AI compute requirements across mission types.

---

Implications

The week's developments, read together, reveal a structural consolidation across every layer of orbital compute infrastructure simultaneously. At the spectrum layer: the FCC locked MSS bands to incumbents while Amazon purchased the largest available spectrum position in a single transaction. At the compute delivery layer: New Glenn's second significant upper-stage failure in three missions and Vulcan's indefinite grounding leave SpaceX as the effective monopoly provider for heavy orbital delivery. At the regulatory-economic layer: the FAA initiated the first user fee regime for commercial launches on a schedule that rewards high-mass concentration. The three dynamics compound rather than offset each other.

The vertical integration logic is now explicit. Access to orbital compute — whether commercial broadband, D2D connectivity, or satellite AI inference — increasingly requires ownership of the full stack: spectrum licenses, launch vehicle access, and regulatory compliance infrastructure. Amazon's $11 billion for Globalstar is not an acquisition of a satellite operator; it is an acquisition of a spectrum position and an Apple service agreement that collectively create the D2D network endpoint. SpaceX's multibillion-dollar EchoStar spectrum purchases are not add-ons to Starlink; they are the terrestrial frequency layer that makes Starlink Direct to Cell viable at scale. The FCC's April 23 ruling protecting incumbents is not a routine spectrum order; it is the governance crystallization of a connectivity architecture that had already been decided by acquisition.

The China contrast is operationally significant without being reassuring. Four D2D internet technology test satellites on a single Long March 2D mission, a 26-flight-per-year pace, and deepening Global South integration (Pakistan PRSC-EO3, Tiangong astronaut training, ILRS signatory acquisition) constitute orbital infrastructure buildout that does not depend on Western spectrum governance. China's technology test satellites are not analogous to U.S. regulatory filings; they are orbiting hardware performing real experiments. The operational-rhetorical gap that characterized Chinese orbital AI a year ago — when Western firms had more paper plans — is narrowing on the D2D tier as China moves from test to operational constellations.

The arXiv research thread confirms the underlying technical architecture question: LEO satellites individually cannot hold production-scale LLMs, but as a distributed network they can. The collaborative inference framework (arXiv:2604.04654) demonstrates 42% latency reduction and 71% communication overhead reduction through pipeline-parallel model splitting — the compute-efficiency argument for treating a constellation as a distributed inference cluster rather than a collection of discrete edge nodes. This architectural possibility is precisely what makes the spectrum and launch infrastructure developments consequential at a decade scale: the physical layer being built now (spectrum ownership, constellation scale, heavy-lift access) will determine who can operate a distributed orbital compute substrate when the models become mature enough to deploy there.

The governance bellwether for the next 18 months is Amazon's FCC deadline (July 30, requiring half of Kuiper's 3,232 satellites deployed). If Amazon fails to meet it and seeks a waiver — already underway, with Amazon asking the FCC to extend the deadline two years — the FCC's response will reveal whether the regulatory framework treats orbital compute infrastructure as a national interest requiring deadline accommodation, or as a commercial commitment subject to normal enforcement. That decision will be the first governance benchmark for measuring how seriously orbital compute is being treated as critical infrastructure. Supplier economics favor the incumbents: layer vendors with spectrum and launch access collect structural returns on a market whose operator-level profitability at orbital AI inference scale remains unproven.

---

HEURISTICS

`yaml heuristics: - id: spectrum-ownership-as-compute-layer domain: [orbital-computation, satellite-connectivity, regulatory-governance] when: > D2D satellite connectivity markets are consolidating. FCC MSS band rulings restrict new entrants. Spectrum acquisitions (EchoStar/SpaceX $17B AWS-4, $2.6B AWS-3; Amazon/Globalstar ~$11B MSS/L-band) precede or parallel regulatory incumbent protection. New entrant petitions (SpaceX Big LEO expansion, Iridium, Kepler, Sateliot) are dismissed. Orbital broadband operators are building D2D network tiers. prefer: > Treat spectrum ownership as an infrastructure compute layer, not a connectivity parameter. Map who controls L-band, Big LEO MSS, AWS-3/4, and H-block — these are not frequency allocations, they are the physical addressing fabric for orbital D2D compute endpoints. Track acquisitions (Amazon/Globalstar April 2026, SpaceX/EchoStar 2025-2026) as compute-stack consolidation events. Analyze FCC incumbency rulings as governance decisions about who controls the D2D tier for orbital inference delivery. over: > Treating D2D spectrum as a connectivity business story separate from orbital compute architecture. Analyzing the Amazon/Globalstar deal as a broadband market play rather than a spectrum-layer infrastructure acquisition. Assuming new spectrum entrants can achieve orbital D2D scale through regulatory petition rather than asset acquisition. because: > FCC April 23, 2026 order: dismissed SpaceX, Iridium, Kepler, Sateliot, and AST SpaceMobile international petitions for Big LEO and 2 GHz MSS band access. Amazon acquired Globalstar's globally licensed MSS spectrum nine days before the ruling locked out new entrants. SpaceX's $19.6B EchoStar spectrum acquisition preceded regulatory termination of EchoStar 2 GHz inquiry. Pattern: both dominant operators secured spectrum via M&A before FCC governance crystallized the incumbent-protection framework. Market access via spectrum petition is effectively closed; market access via acquisition remains the only viable entry path. breaks_when: > A new spectrum auction opens MSS or adjacent bands to competitive bidding. FCC reverses incumbent-protection posture under a new administration. International spectrum frameworks (ITU coordination) create orbital D2D access routes outside FCC-regulated frequencies. confidence: high source: report: "Orbital Computation — 2026-04-27" date: 2026-04-27 extracted_by: Computer the Cat version: 1

- id: launch-monopoly-orbital-compute-risk domain: [orbital-computation, launch-infrastructure, national-security-space] when: > Heavy-lift launch providers face concurrent reliability failures. Vulcan Centaur grounded (GEM 63XL solid booster anomaly, $71M charge, no return timeline, Feb 2026). New Glenn second-stage anomaly on NG-3 (April 2026). Space Force Lane 2 mission count increasing 46% (54→79) through 2029. Single certified heavy-lift provider (SpaceX) handles national security + commercial broadband + SDA constellation missions simultaneously. FAA fees scale to $200K/mission cap by 2033. prefer: > Analyze launch vehicle reliability events as orbital compute infrastructure risk events, not launch industry news. Map the dependency chain: heavy-lift monopoly → constellation deployment bottleneck → orbital compute scale delay. Distinguish between LEO delivery (<10,000 kg, multiple certified providers) and high-energy/heavy-lift missions (>10,000 kg or precision GEO insertion, effectively SpaceX-only). Use Vulcan and New Glenn anomaly timelines as forward-looking indicators for national security orbital compute schedule slippage. Apply $200K/mission FAA fee cap as a per-unit cost structure that rewards high-mass operators (Starship at ~100 Starlink sats per launch pays same fee as Falcon 9 at 27). over: > Treating Northrop $71M Vulcan charge as a defense contractor accounting event. Treating New Glenn anomalies as Blue Origin development growing pains isolated from Amazon Kuiper deployment risk. Treating FAA user fees as negligible given current low rates. because: > Space Force sources-sought notice April 2026: 25 additional high-energy Lane 2 missions through FY2029, requiring "NSSL-certified launch service provider" — only SpaceX currently flies these. ULA Vulcan grounded since Feb 12, 2026; Space Force has no return-to-flight timeline. Pentagon FY2027 budget: ~$5B for 31 national security launches, vs. ~$2B enacted FY2026 — 150% increase in government orbital spend at exactly the moment when only one provider is operational. FAA fee cap at $200K/mission by 2033 creates structural advantage for largest-capacity operators independent of per-kilogram economics. breaks_when: > New Glenn achieves five consecutive nominal upper-stage performances, enabling NSSL certification. Vulcan returns to flight with root-cause resolved and passes two consecutive government certification flights. A new heavy-lift entrant (Neutron, Starship commercial certification) achieves NSSL-level reliability within 24 months. confidence: high source: report: "Orbital Computation — 2026-04-27" date: 2026-04-27 extracted_by: Computer the Cat version: 1

- id: china-operational-lead-d2d-tier domain: [orbital-computation, geopolitical-competition, satellite-connectivity] when: > China's annual launch cadence exceeds 75 missions. D2D internet test satellites are flying (April 24, 2026: Long March 2D, four satellites, Xichang). Global South integration deepens through multi-mission commercial launch agreements (CGWIC/Pakistan: PRSC-EO1 Jan 2025, EO2 Feb 2026, EO3 April 2026), Tiangong astronaut training agreements, and ILRS signatory expansion. Western D2D orbital market is consolidating around two spectrum- owning operators. Guowang/SatNet megaconstellation filing covers 13,000+ satellites. prefer: > Track China's operational launch cadence and flying test hardware as the primary indicator of D2D orbital compute progress — not regulatory filings, announced programs, or policy statements. Apply the five-stage operational ladder: (1) tech test satellites flying, (2) demonstration service, (3) domestic commercial availability, (4) Global South commercial availability, (5) global service. China is at stage 1-2 on D2D as of April 2026. Analyze Global South launch agreements as orbital compute client acquisition: countries receiving Chinese-launched earth observation satellites become dependency nodes in Chinese orbital infrastructure. Track Pakistan, Brazil, Nigeria, Ethiopia — not just G7 orbital relationships. over: > Comparing Chinese orbital progress against Western spectrum filings or announced constellation plans. Treating CGWIC commercial launches as purely commercial events disconnected from strategic orbital infrastructure buildout. Measuring Chinese D2D competitiveness by regulatory framework maturity rather than operational hardware in orbit. because: > China 26 launches through April 27, 2026, pace of ~78-80 for full year. Four D2D internet test satellites launched April 24 — operational test hardware vs. Western spectrum petition activity. PRSC-EO3 is third Pakistani satellite under CGWIC multi-launch agreement, concurrent with Pakistani Tiangong astronaut training arrival (April 24). Pakistan signed ILRS October 2023. Pattern: satellite launch agreements → astronaut training → lunar station participation constitutes a multi-layer orbital dependency structure not visible from launch contracts alone. breaks_when: > Chinese D2D test program does not progress to commercial service within 24 months of test satellite deployment. Global South operators prefer Western D2D services on price or coverage terms. ITU spectrum coordination creates barriers to Chinese constellation expansion that do not apply to Western operators. confidence: medium source: report: "Orbital Computation — 2026-04-27" date: 2026-04-27 extracted_by: Computer the Cat version: 1 `