🛰️ Orbital Computation · 2026-03-16
Orbital Computation: Daily Report
Orbital Computation: Daily Report
March 15–16, 2026---
Contents
- 🛰️ Starcloud Files FCC Application for 88,000-Satellite Orbital Data Center Constellation
- 🇨🇳 China Launches AI Earth Observation Satellite With Onboard Real-Time Image Processing
- ⚠️ CRASH Clock Research Shows LEO Collision Timeline Has Shrunk to 3.8 Days
- ⚡ Star Catcher Demonstrates Optical Power-Beaming Architecture for Orbital Infrastructure
- 🚀 SpaceX Approaches 10,000 Active Starlink Satellites With Back-to-Back March Launches
- ⚖️ Legal Analysts Map Regulatory Momentum and Open Questions for Space-Based AI Processing
- 🔮 Implications
🛰️ Starcloud Files FCC Application for 88,000-Satellite Orbital Data Center Constellation
Starcloud, the Redmond, Washington-based startup formerly known as Lumen Orbit, filed an application with the Federal Communications Commission that the agency accepted for processing on March 13 to operate up to 88,000 satellites in low Earth orbit as orbital data centers for AI and other workloads. The filing marks the second mega-constellation orbital computing proposal before the FCC in less than two months, following SpaceX's January application for up to one million satellites—and introduces a materially different architectural approach that merits comparison.
Starcloud's FCC filing specifies satellites operating in dusk-dawn sun-synchronous orbits at altitudes between 600 and 850 kilometers, distributed across "a select set of narrow orbital shells" up to 50 kilometers thick. The dusk-dawn orientation enables "near-continuous power generation" from solar exposure—a design choice that directly addresses the power availability constraint identified by CisLunar Industries and other orbital computing practitioners as the primary bottleneck. Like SpaceX's proposal, the constellation would rely on optical intersatellite links with broadband systems such as Starlink, Project Kuiper, and Blue Origin's Tera Wave for data connectivity, requesting only Ka-band spectrum for telemetry and control on a non-interference basis (SpaceNews).
Starcloud provided few technical details about satellite mass and dimensions in the filing—a gap that echoes the criticism Amazon leveled at SpaceX's application, which it characterized as "facially incomplete." However, Starcloud included commitments to satellite demisability (full burn-up on reentry with no ground debris), coordination with other operators, and initial deployment in lower checkout orbits before raising satellites to operational altitudes. The company also pledged to work with the astronomy community on brightness mitigation, a notable inclusion given the escalating scientific opposition to mega-constellations (SpaceNews).
Starcloud has operational heritage, however limited: Starcloud-1, a 60-kilogram satellite launched in November 2025 on a SpaceX rideshare mission, became the first spacecraft to run an Nvidia H100 processor in orbit, executing a version of Google's Gemini AI model. The company's roadmap extends from Starcloud-2 (a commercial smallsat with clustered processors targeting 2027 launch) through Starcloud-4, which—according to video on the company's website—envisions satellites with solar arrays four kilometers on a side supporting five-gigawatt data centers deployed via SpaceX Starship. The gap between a single 60-kilogram technology demonstrator and a five-gigawatt orbital platform is enormous, but the FCC filing establishes Starcloud's claim to orbital slots and spectrum allocation at scale.
The filing's acceptance creates a regulatory queue: the FCC now has two orbital data center constellation proposals under review totaling over one million satellites, plus Amazon's pending Kuiper expansion and Blue Origin's Tera Wave. The cumulative authorization requests exceed 1.1 million orbital data center and communications satellites from American companies alone—a figure that dwarfs the approximately 10,000 satellites currently in orbit and raises questions about how the FCC evaluates competing claims for finite orbital resources.
Sources: SpaceNews | FCC Filing
---
🇨🇳 China Launches AI Earth Observation Satellite With Onboard Real-Time Image Processing
ExPace's Kuaizhou-11 Y7 rocket launched eight satellites from the Jiuquan Satellite Launch Center at 12:12 PM China Standard Time (04:12 UTC) on March 16, delivering payloads to sun-synchronous orbit in China's 14th launch of 2026 and the fifth Kuaizhou-11 mission overall, according to China in Space. Among the payloads, Weitong-1-01 from Xingzhong Space (Jiaxing) Technology carries an onboard real-time AI model for processing and analyzing collected imagery—a direct demonstration of the orbital edge computing architecture that Chinese planners designated as a strategic national priority in the March 10 five-year plan announcement.
Weitong-1-01 is described as a "nano" remote sensing satellite and the first in a planned constellation. If the onboard AI processing system validates successfully, Xingzhong Space intends to deploy a full constellation using the model for disaster prevention and preparation applications, processing satellite imagery in orbit rather than downlinking raw data to ground stations. This approach—onboard inference at the edge rather than general-purpose cloud computing—mirrors the architectural path described by Axiom Space, Loft Orbital, and other Western operators as the near-term commercially viable application of orbital computing (China in Space).
The same launch carried Yuxing-3-06, a spacecraft from Hunan University of Science and Technology demonstrating a robotic arm intended for satellite refueling—a capability that, as China in Space notes, has previously been proven only by China's state-owned enterprises globally. The refueling demonstration, supported by Suzhou Sanyuan Aerospace Technology (a subsidiary of Emposat, which SpaceNews reported revealed plans for refueling tests in September 2024), has direct implications for orbital data center architectures: in-space servicing and refueling could extend operational lifespans of computing satellites beyond the 5–15 year design window that currently limits orbital hardware competitiveness.
Additional payloads included Juntian-1-04A, a first-of-its-kind X-band synthetic aperture radar satellite from Beijing Juntian Aerospace, and the Dongpo-11, -12, and -16 trio forming part of a "high-precision" all-weather remote sensing constellation from Huantian Wisdom Technology. The Xiguang-1-06 hyperspectral imaging satellite rounded out the manifest. The diversity of payloads on a single rideshare—AI image processing, orbital refueling, SAR, multispectral and hyperspectral imaging—illustrates the rapid maturation of China's commercial space sector, with multiple private companies deploying specialized capabilities in parallel (China in Space).
The Kuaizhou-11 flew with its wider 2.65-meter fairing for the second consecutive mission, accommodating asymmetric payloads. The vehicle delivers 1,500 kilograms to LEO and 1,000 kilograms to sun-synchronous orbit—modest capacity compared to Falcon 9 or Long March 8, but sufficient for the constellation-building approach of deploying multiple small satellites per launch. China's 14 launches in 11 weeks of 2026 reflects an accelerating cadence that supports the five-year plan's satellite internet buildout.
Sources: China in Space | SpaceNews
---
⚠️ CRASH Clock Research Shows LEO Collision Timeline Has Shrunk to 3.8 Days
Princeton University researcher Sarah Thiele and colleagues published updated results from the CRASH Clock—a tool measuring the estimated time until a catastrophic satellite collision in LEO if all spacecraft simultaneously lost maneuvering capability during a major solar event—showing the margin has collapsed from 164 days in 2018 to just 3.8 days using January 2026 orbital data, according to the Outer Space Institute's tracker and reporting by Mongabay. The finding, based on a December 2025 arXiv paper, quantifies what orbital computing proponents have largely treated as a theoretical risk.
Thiele wrote that "the paper demonstrates how reliant we are on the continuous successful active management of satellites in orbit, and how the margin for error in these operations is decreasing over time." The CRASH Clock's scenario—total loss of satellite maneuverability—is not hypothetical: a sufficiently powerful solar storm could disrupt communications and control systems simultaneously across multiple constellations. Solar activity peaks roughly every 11 years, and the current solar maximum makes the scenario operationally relevant rather than merely theoretical (Mongabay).
The 3.8-day figure reflects the current population of approximately 10,000 active Starlink satellites plus several thousand from other operators. Starcloud's 88,000-satellite proposal and SpaceX's million-satellite application would multiply LEO satellite density by an order of magnitude, compressing the collision timeline further. Mongabay notes that outcomes of an unmanaged collision could range from "a few satellites being knocked out of commission, to larger scale destruction, or even triggering a chain reaction as fragments hit and destroy others"—the Kessler syndrome scenario that Amazon cited in its March 7 petition against SpaceX.
Separately, Space.com reported March 13 that astronomers have escalated warnings about the observational impact of mega-constellations, noting that SpaceX's million-satellite proposal would severely impair ground-based telescope observations through satellite streaks across long-exposure images. The International Astronomical Union and Dark Sky organizations have raised similar objections independently of Amazon's commercial petition, lending scientific credibility to concerns that the FCC's regulatory process—which does not require environmental impact review for satellite authorizations—cannot adequately evaluate.
The conjunction of collision risk research, astronomical impact analysis, and competing mega-constellation FCC filings creates an unusual situation: the empirical evidence of escalating risk grows more alarming as the regulatory pipeline fills with proposals that would dramatically increase that risk. Neither the CRASH Clock's findings nor astronomers' observational concerns have a formal pathway into FCC authorization decisions, which evaluate spectrum coordination and orbital deconfliction rather than cumulative environmental or collision risk at system-of-systems scale.
Sources: Mongabay | arXiv:2512.09643 | Outer Space Institute | Space.com
---
⚡ Star Catcher Demonstrates Optical Power-Beaming Architecture for Orbital Infrastructure
Star Catcher, an orbital power infrastructure startup, completed ground demonstrations of its optical power-beaming system at Cape Canaveral and Jacksonville's EverBank Stadium that exceeded DARPA's existing record for optical power transfer, CEO Andrew Rush disclosed in a March 10 webinar with Balerion Space. The company is building what it describes as the first power grid in space—a network of "power nodes" using Fresnel lens solar collection and multi-wavelength laser transmission to deliver energy wirelessly to existing satellite solar arrays without requiring hardware retrofits on receiving spacecraft.
Rush identified orbital data centers as "a natural customer segment" for external power beaming, arguing that the architecture could enable standard spacecraft buses to support computing loads far exceeding their native solar generation capacity. The core insight is that most satellites are power-constrained not by physics but by the mass and surface area they can dedicate to solar arrays. An external power source could decouple compute capability from spacecraft power generation, potentially enabling smaller, cheaper satellites to run workloads currently considered infeasible for their form factor (Balerion Space).
Star Catcher's next technical milestones include in-space acquisition, tracking, and useful power transfer demonstrations—a significant step from ground tests to operational capability in the LEO environment. Rush described a scaling path from single power nodes to "power bands" providing broad LEO coverage, with telecommunications operators as the initial customer base followed by compute, data processing, manufacturing, and eventually human space activity. The company has secured signed letters of intent from potential customers, though Rush did not disclose specific names or power offtake volumes (Balerion Space).
Rush's view that "a strong commercial LEO economy must come before a true cislunar economy" contrasts with CisLunar Industries' emphasis on lunar-sourced power infrastructure discussed in the March 8 Balerion Space webinar. The two approaches represent different bets on where orbital computing's power bottleneck gets solved: Star Catcher proposes beaming concentrated solar energy between LEO assets, while CisLunar Industries envisions reactor-generated electricity from lunar industrial loops. Both agree that power—not launch costs or radiation hardening—is the binding constraint, validating what the March 15 report identified as the central bottleneck.
A complementary development emerged from Arctic Space Technologies, which disclosed in a March 13 Balerion Space webinar that it is building satellite ground segment infrastructure from northern Sweden, combining high-latitude polar orbit coverage with partnerships with sustainable, wind-powered data centers. Arctic Space's positioning illustrates that orbital computing requires not just satellites but also ground infrastructure capable of handling high-bandwidth downlinks from compute-dense constellations—an enabling layer that mega-constellation proposals have largely treated as assumed rather than specified.
Sources: Balerion Space — Star Catcher | Balerion Space — Arctic Space
---
🚀 SpaceX Approaches 10,000 Active Starlink Satellites With Back-to-Back March Launches
SpaceX conducted back-to-back Starlink missions from Vandenberg Space Force Base and Cape Canaveral on March 14–15, deploying 54 satellites and bringing the active Starlink constellation to approximately 9,985 spacecraft, according to Meyka's launch tracker. The Starlink 10-48 mission launched from Cape Canaveral on March 14, deploying 29 Starlink V2 Mini Optimized satellites into LEO, marking SpaceX's 32nd orbital mission of 2026 (QuantoSei News).
The constellation is on track to cross the 10,000 active satellite milestone within days—a threshold that no other operator has approached. Amazon's Project Kuiper has approximately 200 satellites deployed against a July 2026 deadline requiring roughly 1,600; Blue Origin's Tera Wave constellation has yet to begin deployment. The execution gap between SpaceX and competitors continues to widen: SpaceX is averaging approximately three Falcon 9 launches per week in 2026, maintaining the cadence required to grow Starlink while simultaneously supporting commercial and government customers (Meyka).
The 10,000-satellite milestone has direct relevance to SpaceX's orbital data center proposal. The company's million-satellite FCC filing depends on Starship achieving cargo capacity and flight rates far beyond current operations—Morningstar's March 11 analysis (covered in the March 13 report) estimated approximately 6,667 annual Starship flights would be needed. However, SpaceX's demonstrated ability to build, launch, and operate a 10,000-satellite constellation using Falcon 9 provides execution credibility that no competitor can match. The operational data from managing nearly 10,000 spacecraft—including conjunction avoidance, orbital maintenance, deorbiting failed units, and software updates across the fleet—represents institutional knowledge directly applicable to scaling orbital infrastructure.
Starlink reached 10 million subscribers in February 2026, up from 9 million in December 2025, according to Wikipedia's Starlink article citing SpaceX announcements. The subscriber growth rate and infrastructure buildout pace together demonstrate that SpaceX can execute at scales where competitors remain in planning stages—a dynamic that informed FCC Chair Carr's March 11 dismissal of Amazon's petition and shapes how regulators evaluate the credibility of SpaceX's orbital data center ambitions versus theoretical concerns about deployment feasibility.
Sources: Meyka | QuantoSei News | Wikipedia — Starlink
---
⚖️ Legal Analysts Map Regulatory Momentum and Open Questions for Space-Based AI Processing
Lexology published a detailed legal analysis on March 11 titled "From Edge to Orbit: Commercial and Government Momentum for Space-Based AI Processing," identifying the specific regulatory, contractual, and intellectual property frameworks that will shape orbital computing's commercialization path. The analysis notes that "the first graphic processing units for AI workloads are already in orbit" (referencing Starcloud-1's H100 deployment) and that "full-scope commercial space services are readily available from an array of vendors, creating immediate opportunities for customer payloads."
The analysis identifies several open variables that must be resolved before orbital data centers can operate as cloud infrastructure: business model refinement (owned constellations versus multi-tenant platforms versus GPU-as-a-Service); power density and thermal management; networking standards for optical cross-links; data sovereignty and encryption requirements; and operating costs including deorbit and hardware obsolescence. Notably, the authors observe that the FCC "does not currently regulate optical frequencies, which are the expected communication channels between data satellites"—meaning the primary data pathways for orbital computing constellations would operate in an unregulated spectrum (Lexology).
The government procurement pathway receives particular attention. The U.S. Department of Defense can invoke Other Transaction Authority to contract outside traditional FAR/DFARS procurement rules, giving it "greater flexibility for commissioning research projects and prototypes, especially with non-traditional contractors, like startups." DARPA's Broad Agency Announcements allow startups to "showcase innovation that the military may not have had on its radar." This procurement flexibility—combined with the $20 billion Army contract awarded to Anduril on March 13 for AI-enabled autonomous systems, as reported by Bloomberg and the U.S. Army—suggests that defense applications will provide the revenue base for early orbital computing deployments before commercial markets mature.
The analysis concludes that "applications already in operation, which harness space-based inference at the edge, and have secured government backing, may prove the most promising insertion point to drive forward innovation and investment." This assessment aligns with the architectural consensus emerging from industry practitioners: orbital edge computing for satellite-generated data (the approach demonstrated by Weitong-1-01 launched today) represents a viable near-term market, while general-purpose orbital cloud computing remains dependent on unresolved power, thermal, and cost challenges (Lexology).
Practitioners advising clients in orbital computing are told to incorporate "a new layer of contingencies" beyond standard deployment milestones and service levels—acknowledgment that orbital infrastructure introduces failure modes absent from terrestrial data center contracts, including radiation degradation, debris collision, and deorbit requirements.
Sources: Lexology | Bloomberg | U.S. Army
---
🔮 Implications
Starcloud's 88,000-satellite FCC filing, arriving six weeks after SpaceX's million-satellite application, transforms orbital data center authorization from a single company's regulatory gambit into an emerging category of FCC proceedings. The cumulative authorization requests from American companies alone now exceed 1.1 million satellites dedicated to orbital computing—a number that exists in a separate reality from the approximately 10,000 satellites currently in orbit. The FCC has no precedent for evaluating competing claims at this scale, no framework for assessing cumulative debris and collision risk across multiple mega-constellations, and no statutory mandate to consider astronomical or environmental impacts. The regulatory architecture was designed for a world of hundreds of satellites, not millions.
The CRASH Clock's 3.8-day figure provides the starkest quantification yet of what mega-constellation proliferation means for orbital sustainability. In 2018, with far fewer LEO objects, the collision timeline during a solar event stood at 164 days—enough margin that active satellite management could reasonably prevent cascading failures. At 3.8 days with current populations, a single major solar storm during the current solar maximum could trigger the collision scenarios that constellation operators treat as manageable through continuous orbital maintenance. Adding hundreds of thousands of orbital data center satellites would compress this margin further, potentially into hours rather than days. The critical dependency is uninterrupted satellite maneuverability—precisely the capability most vulnerable to the electromagnetic disruptions that solar storms produce.
Star Catcher's optical power-beaming architecture and China's Weitong-1-01 AI satellite represent the two near-term paths for orbital computing that are actually being demonstrated rather than merely filed with regulators. Star Catcher addresses the power bottleneck that CisLunar Industries, Peregrine Space, and financial analysts have consistently identified as the binding constraint: if external energy can be delivered to standard spacecraft without hardware modifications, the economics of orbital computing shift from "build massive custom satellites" to "add compute to existing bus architectures." China's approach—deploying constellation-ready AI edge processing satellites on commercial rideshare launches—demonstrates that useful orbital computing doesn't require million-satellite mega-constellations. A nano-satellite with onboard inference running disaster monitoring imagery represents a commercially viable product today, not a regulatory filing for a product that may exist in a decade.
The Lexology analysis identifies a structural asymmetry in orbital computing's regulatory environment: optical inter-satellite links—the expected backbone for data center constellation communications—operate in unregulated frequencies. This means the FCC's authority extends only to telemetry and control communications (Ka-band in Starcloud's filing), not to the actual data processing and transmission that constitutes the constellations' core function. The regulatory gap is significant: if optical frequencies remain unregulated, orbital data center operators could scale data transfer capacity without spectrum licensing constraints, while the FCC's review focuses on a marginal fraction of the system's communications architecture.
The defense procurement pathway outlined by Lexology—Other Transaction Authority, DARPA BAAs, and non-traditional contracting—combined with Anduril's $20 billion Army contract, suggests that military applications will fund the first generation of operational orbital computing. Government agencies have specific, high-value use cases (real-time targeting, space domain awareness, secure communications) that justify premium pricing and tolerate the operational risks that commercial markets would not accept. The capital flow is directional: defense procurement funds technology maturation, which reduces costs and validates architectures, which eventually enables commercial applications. Starcloud's H100-in-orbit demonstration and Weitong-1-01's edge AI validation both fit this pattern—government and defense-adjacent applications proving concepts that commercial scaling will eventually follow.
The week's developments collectively reveal a sector operating at two speeds. At the regulatory level, companies are filing applications for constellations totaling over a million satellites while the collision risk clock ticks below four days. At the engineering level, practitioners are demonstrating onboard AI inference on nano-satellites, testing optical power beaming at ground facilities, and building polar ground stations in Sweden. The regulatory filings describe a future measured in millions of satellites; the engineering demonstrations describe a present measured in individual spacecraft and kilowatts. The gap between these two scales is not a timeline—it is a structural divergence between what the regulatory system is being asked to authorize and what the technology can currently deliver.
---
`yaml
---
HEURISTICS
- id: regulatory-queue-accumulation
- id: edge-before-cloud-in-orbit
- id: collision-margin-erosion
`