ENGINEERING ARCHITECTURE

ORBITAL COMPUTE / HOSTED PAYLOAD SPACE DePIN

Move selected workloads
beyond the terrestrial limit.

Proposed modular low-Earth-orbit infrastructure designed around radiative heat rejection, solar generation, optical cross-links, physical isolation and staged hosted payloads. Its advantage must be demonstrated workload by workload against strong terrestrial alternatives.

REFERENCE ORBIT505 KM SSO / LEO
MODULE CLASS4 KG HOSTED PAYLOAD
DESIGN LIFE5 YEARS
LINK LAYERS-BAND + 1550 NM OISL
THE ENGINEERING REALITY

Vacuum is not a coolant.
Radiation is the answer.

Without atmospheric convection, space hardware must conduct heat away from processors and reject it as infrared radiation. The design therefore begins with thermal architecture—not marketing.

STEFAN–BOLTZMANN HEAT REJECTIONP = εσA(Tᵣ⁴ − Tₛ⁴)

Radiated power scales with emissivity, radiator area and the fourth power of radiator temperature.

Orbital compute infrastructure concept
CONCEPT VISUAL / ORBITAL COMPUTENOT TO SCALE

Five stages from
silicon to deep space.

01 / CAPTURE

Processor die

Carbon-based thermal interface moves heat into a direct cold plate.

02 / TRANSPORT

Loop heat pipe

Two-phase working fluid carries heat through a passive capillary loop.

03 / BUFFER

Phase-change mass

PCM absorbs transient compute peaks before radiator equilibrium.

04 / SPREAD

Radiator wing

High-emissivity surfaces maximize effective rejection area.

05 / REJECT

Infrared radiation

Waste heat leaves the spacecraft as electromagnetic radiation.

SCIENTIFIC NOTE — The 2.7 K cosmic background does not create “free cooling.” Radiator sizing, orientation, absorptivity, view factors, solar load and Earth albedo govern real thermal performance.

A compact validator
with spacecraft discipline.

The current reference architecture models a low-power, radiation-aware compute and storage payload carried by a commercial host bus.

TOTAL MODULE MASS4.0 kg

PC-104 class stack, enclosure, shielding, clock and dual S-band antenna allowance.

AVERAGE / PEAK POWER6 W / 15 W

Dynamic hot/cold storage zones reduce idle thermal load.

RADIATION DESIGN100 krad

Modeled five-year TID tolerance using graded-Z aluminum and tantalum concepts.

MEMORY PATHSTT-MRAM

Magnetic non-volatile state storage for improved SEU tolerance.

SPACE–GROUND LINK2.25 GHz

S-band telemetry concept at up to 1024 kbps with Doppler tracking.

SPACE–SPACE LINK10 Gbps

Proposed 1550 nm optical inter-satellite mesh.

Observe. Predict. Move.
Isolate. Service.

Physical intervention is rare and expensive in orbit. The updated architecture therefore treats autonomy as a closed-loop resilience system that must first be matured and measured on terrestrial hardware.

01 / OBSERVE

Health telemetry

Continuously monitor temperature, voltage, memory integrity and communication state.

02 / PREDICT

Failure scoring

Detect degradation trends early enough to create a safe migration window.

03 / MIGRATE

Workload transfer

Move encrypted shards, validation duties and recoverable state to healthy nodes.

04 / ISOLATE

Fault containment

Power-gate and logically quarantine damaged cells, links or compute cores.

05 / SERVICE

Lifecycle extension

Later phases explore standard interfaces for modular orbital inspection or replacement.

MATURITY BOUNDARY — The five-step loop is a target architecture. Prediction horizon, lossless migration, isolation effectiveness and serviceability are not operating claims until demonstrated through fault injection, hardware-in-the-loop trials and flight evidence.

Design for loss.
Prove the recovery path.

Launch and initial commissioning do not establish durable service. POP’s 6 September requirements framework now treats attitude, power, thermal, communications, navigation, control authority, servicing and disposal as explicit failure domains—each with detection, containment, recovery and evidence gates.

SAFE STATE

Protect the spacecraft first

Power-positive safe mode, command, timing and telemetry must outrank revenue workloads during attitude, battery or thermal degradation.

CONTINUITY

Migrate without false finality

Authenticated checkpoints, separated replicas and partition-safe operation must preserve state while clearly marking when global finality is unavailable.

MANOEUVRE AUTHORITY

Separate advice from control

Conjunction inputs need source provenance and uncertainty records; safety-critical commands require legally accountable, threshold authorization.

SERVICE OR REPLACE

No rescue dependency

Standard servicing interfaces may be evaluated, but continuity cannot depend on a successful rendezvous. Propellant reserves, abort rules, replacement and disposal remain part of the lifecycle model.

CURRENT EVIDENCEPRE-FLIGHT REQUIREMENTS FRAMEWORK
GOA R&D GATESFault injection · network degradation · power/thermal safe mode · conjunction exercise · servicing simulation
PUBLIC CLAIM LIMITDefined requirements—not resilient, serviceable, flight-qualified or operational capability

EVIDENCE TRIGGER / A real proximity mission reinforced the difference between commissioning and recoverable service: Katalyst LINK commissioning record and reported Swift approach outcome. These third-party events inform POP requirements; they do not validate POP capability.

Rent the bus.
Own the service layer.

Rather than funding a dedicated mega-constellation, PoP’s model places lightweight validator and storage payloads on compatible host spacecraft or deploys software onto qualified orbital compute.

  1. Hardware-hosted: integrate self-contained 1U/3U-class modules with commercial satellite buses.
  2. Software-hosted: deploy containerized validation software to compatible Linux/RTOS flight computers.
  3. Hybrid ground sector: combine commercial ground-station networks with community tracking nodes.
  4. Rolling replenishment: replace a portion of nodes annually to manage degradation and orbital decay.
5-YEAR PARTIAL-SCOPE MODEL

$11.24M internal model

Illustrative 48-node hosted-payload programme scope covering stated manufacturing, integration, ground/regulatory operations and replenishment assumptions. It is not a quoted price or complete delivered-compute TCO.

INITIAL MODEL$5.28M
5-YEAR OPS MODEL$1.40M
REPLENISHMENT MODEL$4.56M

Different physics.
One comparison boundary.

PoP is compared with credible terrestrial designs—not an artificially weak baseline. No column establishes a winner until equivalent workloads and full lifecycle boundaries are measured.

PARAMETERGRID-CONNECTED CAMPUSDEDICATED-GENERATION CAMPUSRENEWABLE-LINKED CAMPUSPOP ORBITAL ARCHITECTURE
Power boundaryUtility supply, UPS and backup generationOnsite generation, fuel supply and optional grid supportGeneration, storage, intermittency and grid contractSolar generation, batteries, distribution loss and eclipse reserve
Thermal boundaryAir or liquid cooling; PUE and WUEFacility cooling plus generation heat and parasitic loadsFacility cooling, location and seasonal constraintsConduction, transport and radiator area/mass per kilowatt
Network boundaryFibre routes and endpoint latencyFibre routes and endpoint latencyFibre or remote backhaul and endpoint latencyInter-satellite links, downlink, gateways and fibre backhaul
MaintenanceHuman access, repair and hardware refreshHuman access plus generation maintenanceHuman or remote access plus storage replacementAutonomous recovery, replenishment and failed-node replacement
Security & sovereigntyJurisdiction and architecture dependentJurisdiction and architecture dependentJurisdiction and architecture dependentArchitecture dependent; physical separation does not remove legal control
Full-cost boundarySite, grid, facility, cooling, network, refresh and financeSite, generation, fuel, cooling, network, refresh and financeSite, generation, storage, cooling, network, refresh and financePayload, bus, shielding, thermal, launch, insurance, ground, replacement, disposal and finance

Comparison framework: POP Terrestrial Alternatives vs Orbital Benchmark, 5 September 2026. Values remain unpopulated until a common workload and evidence class are available.

Measure every burden.
Claim only the evidence.

POP does not assume that moving compute into orbit makes it greener. The 9 September measurement framework requires terrestrial and proposed orbital systems to be compared for the same useful workload and service level, with every result labelled by evidence class.

TERRESTRIAL SYSTEM

Measure the full site

Include electricity source and losses, compute, networking, cooling, water, heat, buildings, land, noise, onsite generation, manufacture, refresh and disposal.

ORBITAL SYSTEM

Count Earth and orbit

Include payload and bus manufacture, launch, failures, solar and battery systems, radiators, links, gateways, replenishment, ground operations and end of life.

PROVENANCE

Show how each result arose

Record the instrument or source, reporting period, workload denominator, boundary, exclusions, uncertainty, calculation version and assurance status.

CLAIMS CONTROL

No automatic green advantage

“Zero-carbon,” “zero-water,” “clean” or “greener than terrestrial” remain prohibited unless matched lifecycle evidence and independent review support the exact statement.

STATUS — This is a measurement and disclosure framework, not environmental performance evidence. POP has not yet measured or independently assured energy, water, emissions, heat, materials, launch, replacement or disposal impacts for an operational orbital system.

Evidence trigger: Oracle’s 8 September 2026 Project Jupiter announcement describes planned public onsite reporting and independent assessments; it does not validate POP or provide POP measurements. POP framework: Environmental Measurement and Disclosure Framework, 9 September 2026.

From 64 terabytes
to petabyte-class orbit.

The storage roadmap separates near-term design tiers from long-range research ceilings. “Usable” capacity is shown after the architecture’s modeled parity, logging, mirroring and replica reserve.

3U
DESIGN TIER

SAV‑1 Micro‑Vault

64 TB usable

128 TB raw target · 4 kg class · 15 W peak · 2 W standby. A compact physically isolated archive and key-custody payload.

REDUNDANCY RESERVE
50%
FORM
CubeSat / hosted canister
SS
SCALE TIER

SAV‑2 Orbital Data Centre

1.0 PB usable

2.0 PB raw target · approximately 150 kg · 120 W peak. Active-active mirroring and Proof-of-Replica capacity reserve.

REDUNDANCY RESERVE
50%
FORM
Dedicated SmallSat
R&D
RESEARCH CEILING

Maximum-capacity vault

10–50 PB raw target

Long-range single-node research envelope; the upper case implies about 25 PB usable after parity/EDAC. It is not a current flight specification.

REQUIRES
Thermal + radiation validation
STATUS
Research architecture
01 / INGEST WINDOWAuthenticated burst

Data enters only during a scheduled, cryptographically authorized communications window.

→
02 / BUS ISOLATIONTransponder off

Duty-cycled RF and power-domain isolation remove the continuous network path.

→
03 / SHARD & VERIFYK-of-N recovery

Erasure coding, replica proofs and cross-node shards protect against device loss.

→
04 / CONTROLLED EGRESSPolicy-gated release

Multi-party authorization reopens a bounded retrieval channel.

MATURITY NOTE — These are internal engineering targets derived from the Space Airgap Vault architecture documents. Capacity, power, mass, shielding and lifetime require component selection, thermal-vacuum testing, radiation analysis and launch-partner acceptance.

Count the complete system.
Then test the advantage.

The economic thesis is not “space is cheap.” Hosted payloads may reduce ownership of a dedicated spacecraft and launch stack, but any claimed advantage must survive comparison with grid-connected, dedicated-generation and renewable-linked campuses after all lifecycle costs and useful workload output are counted.

GRID-CONNECTED

Count the queue

Include power availability, site work, cooling, fibre, permitting, maintenance and financing.

DEDICATED GENERATION

Count the fuel system

Include onsite generation, fuel exposure, reliability, emissions, cooling and capital.

RENEWABLE-LINKED

Count variability

Include generation profile, storage, curtailment, modular compute and backhaul.

POP ORBITAL

Count every launched subsystem

Include payload, bus, shielding, radiators, links, launch, insurance, ground, replacement, disposal and finance.

ILLUSTRATIVE SCOPE DIFFERENCE96.40%

(1 − hosted scope ÷ selected dedicated-space scope) × 100

SELECTED DEDICATED-SPACE BASELINE100% selected scope
POP HOSTED-PAYLOAD PARTIAL SCOPEInternal manufacturing + rideshare model

This control compares two internal space-programme scopes. It is not a terrestrial-versus-orbital benchmark, a delivered-compute comparison or a verified saving.

48 HOSTED NODES / MODEL$5.28M

Internal initial manufacturing, launch and integration assumptions.

5-YEAR PARTIAL SCOPE / MODEL$11.24M

Stated ground/regulatory operations and rolling replenishment assumptions.

DEDICATED SMALLSAT DELTA / MODEL95.00%

Illustrative scope difference; equivalent service and full lifecycle boundaries are not established.

MEGA-CONSTELLATION DELTA / MODEL99.92%

Illustrative scope difference only; architecture and service scopes differ materially.

MODEL BOUNDARY — These figures are internal partial-scope assumptions, not contracted quotes, audited operating history, complete delivered-compute TCO or evidence that orbital compute is lower-cost. A decision-grade model must also account for payload hardware, memory and storage, bus and structure, shielding, power storage and losses, radiator area and mass, communications, insurance and failure reserve, cybersecurity, ground infrastructure, refresh, disposal and financing.
INITIAL WORKLOAD RULE — General-purpose terrestrial AI training is not claimed as an initial PoP target. Candidate workloads must demonstrate a location-specific advantage such as proximity to space-generated data, reduced raw-data downlink, sovereign continuity, distributed orbital verification or an orbital-native service requirement.

Own the critical IP.
Source the mature infrastructure.

The proposed Goa R&D centre is intended as the orbital engineering core. The updated roadmap avoids rebuilding an entire national space supply chain: shared qualification capacity and flight-proven subsystems would be sourced, while POP’s differentiated compute, custody and consensus research would remain internal.

Concept visualization of the proposed Goa immersion compute laboratory
ROOM 301 / IMMERSION COMPUTECONCEPT VISUAL · NOT BUILT
  1. Source externally: launch access, host buses, solar and attitude systems, ground-station capacity and shared environmental qualification.
  2. Develop in Goa: temporal consensus, clock attunement and space-ground state coordination.
  3. Protect the compute envelope: thermal-isolation logic, power gating, radiation-aware state and air-gap custody.
  4. Anchor trust: hardware attestation, secure key registration and provable machine-action modules.
  5. Integrate the fabric: optical links, silicon-photonics research, telemetry routing and hardware-in-the-loop simulation.
PARTNER BOUNDARY — The public site describes capability categories only. No supplier, facility, launch, government or institutional relationship is implied without an executed agreement and controlled evidence.
PHASE 3 ARCHITECTURE BLUEPRINT

POP-BIONODE / GLOBAL BLACKOUT WATCH

When the grid disappears,
the pulse remains.

Phase 3 explores a wrist-worn, ultra-low-power emergency wallet and alert interface designed to preserve authenticated messaging when conventional power, cellular and fibre infrastructure are unavailable.

SKIN HEAT + MOTION→HEARTBEAT ENTROPY→SPARSE RADIO BURST→WSN ORBITAL UPLINK
STATUS — The source pack contains mathematical specifications, a wearable manufacturing blueprint, a Python simulator and a dashboard prototype. It does not establish flight qualification, production readiness, direct-to-satellite regulatory approval or real-world blackout deployment.
Concept render of the POP-BioNode blackout-resilience watch
GRIDOFFLINEWSN LINKSTANDBYPOWERBIO-KINETIC
PROTECTED DILIGENCE

Move from public architecture to controlled evidence.

Qualified institutions can request the air-gap vault, Goa R&D, MVP/POC, Phase‑3 and financial source packs after verification, final NDA execution and board approval.

SOVEREIGN ORBITAL ARCHITECTURE

Connect orbital compute to jurisdiction, key ownership, provenance, measurable resource economics and institutional Layer‑1 settlement.