# Welcome to SpaceXSI

More possibility. Less energy. A research foundation for useful computation, space-inspired thinking and open coordination.

Type: Project vision · Reviewed: 4 October 2026

## The thesis

SpaceXSI begins with a practical question: how can digital infrastructure deliver more useful outcomes without treating energy as an unlimited input? Its vision connects the resource discipline of space engineering, the optimization potential of SI, and the coordination tools of crypto.

The ambition is an ecosystem where a computation is evaluated by what it accomplishes, the resources it consumes, and the evidence behind the measurement. More activity alone is not the goal. Better use of capacity is.

## Why this matters

The IEA’s 2025 analysis estimates that data centres used approximately 415 TWh of electricity in 2024. Its base case projects approximately 945 TWh in 2030. These figures cover the entire data-centre sector, not SI alone. The forecast is a scenario, not a fixed outcome.

For SpaceXSI, this establishes a research priority: examine the physical consequences of digital growth alongside its technical possibilities.

Source: [Energy and AI: energy demand](https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai) — IEA · 2025. Data-centre demand estimates and scenario-based forecasts. Sector totals include workloads beyond intelligence.

## Three connected pillars

| Pillar | Question | Direction |
| --- | --- | --- |
| Energy | What does useful work cost? | Measure electricity, quality, latency and utilization together. |
| Space | What can constrained systems teach us? | Explore solar potential, autonomy and lifecycle engineering. |
| Crypto | How do participants coordinate? | Evaluate transparent records, verification and settlement. |

## Choose your reading path

Research chapters explain published findings. Design chapters describe proposed SpaceXSI directions. Worked examples explain calculations with selected assumptions. This separation makes the vision concrete without confusing a proposal with a measured result.

- Foundations: Energy landscape → Efficiency & measurement → SI & computation.
- Space thesis: Solar & orbital power → Computing in orbit → Future pathways.
- Project design: Crypto & coordination → System architecture → Research roadmap.
- Reference: Glossary → Sources & methodology.

---

# The energy landscape

Digital growth is also a physical infrastructure problem: power, cooling, capacity and time.

Type: Research · Reviewed: 4 October 2026

## Computation has a physical footprint

Every digital request becomes work performed by chips, memory, storage and networking equipment. The facility also needs power delivery and environmental control. An efficient accelerator is therefore only one part of an efficient service.

The IEA describes demand across servers, storage, networking and cooling, with a composition that varies between facilities. Its 2025 outlook highlights a timing mismatch: compute deployments can move faster than the energy infrastructure needed to support them.

Source: [Energy and AI: energy demand](https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai) — IEA · 2025. Data-centre demand estimates and scenario-based forecasts. Sector totals include workloads beyond intelligence.

## Growth and efficiency are different variables

Total demand can rise even while individual operations become cheaper. If energy per task falls by 20% but task volume doubles, combined energy consumption rises by 60%, before changes in overhead. This is arithmetic, not a project forecast.

Track both intensity and scale: energy per accepted result and energy consumed across all results. A system can improve the first metric while worsening the second.

## Keep four objectives separate

| Objective | What improves it | What it does not establish |
| --- | --- | --- |
| Energy efficiency | Less electricity for equivalent work | Lower emissions in every location. |
| Carbon intensity | Cleaner electricity at the time of use | Lower absolute energy demand. |
| Reliability | Resilience and sufficient capacity | Resource efficiency. |
| Economics | Lower delivered cost and better utilization | Lower lifecycle impact. |

## Flexibility is a resource

Some work must finish immediately; other work can move within a deadline. Radovanović and colleagues describe scheduling flexible compute around changing grid carbon intensity while respecting capacity constraints. This is an operational mechanism rather than a replacement for adequate power supply.

Our design inference is to separate interactive work from flexible batch work before assigning resources. Scheduling decisions need explicit deadlines, location restrictions and data-transfer costs.

Source: [Carbon-Aware Computing for Datacenters](https://arxiv.org/abs/2106.11750) — Radovanović et al. · 2021 paper. Scheduling flexible compute around time-varying grid carbon intensity.

## Implications for SpaceXSI

- Make energy an input to workload decisions, not only a dashboard after execution.
- Compare equivalent service levels rather than headline chip specifications.
- Report absolute demand alongside per-task savings.
- Treat available power, local impacts and supply constraints as design inputs.

---

# Efficiency & measurement

Useful work per watt is a principle. Reproducible measurement makes it meaningful.

Type: Research + examples · Reviewed: 4 October 2026

## Define the measurement boundary

A measurement may describe a chip, server, rack or entire facility. Changing the boundary changes the number. Accelerator telemetry should not be presented as the complete energy cost of a service.

A proposed benchmark should identify hardware, software version, workload, input size, warm-up, measurement period and instrument. It should disclose whether idle power, networking, storage and cooling are included.

## Use a portfolio of metrics

| Metric | Definition | Purpose |
| --- | --- | --- |
| Energy per task | Measured energy ÷ accepted tasks | Compare equivalent useful work. |
| Useful throughput | Accepted results ÷ elapsed time | Separate valid output from activity. |
| Latency | Request-to-result duration | Check service requirements. |
| Utilization | Used capacity ÷ available capacity | Identify idle resources. |
| PUE | Facility energy ÷ IT equipment energy | Describe facility overhead. |
| Carbon estimate | Energy × emissions factor | Evaluate electricity-related emissions with stated assumptions. |

Source: [Power Usage Effectiveness](https://www.thegreengrid.org/node/372) — The Green Grid. The definition of facility energy divided by IT equipment energy.

Source: [Power measurement for MLPerf](https://mlcommons.org/working-groups/benchmarks/power/) — MLCommons. Methods connecting measured power, energy and benchmark performance.

## A facility overhead example

The Green Grid defines PUE as total facility energy divided by IT equipment energy. It does not measure the usefulness of the computation. A low-PUE facility can still run an inefficient workload.

If IT equipment uses 100 kWh, PUE 1.4 implies 140 kWh at the facility boundary. At PUE 1.2, the same IT load implies 120 kWh: a reduction of 20 kWh, or about 14.3% of baseline facility energy. This example assumes equivalent IT work and measurement periods.

```text
PUE = E_facility / E_IT
Saving (%) = (E_baseline − E_candidate) / E_baseline × 100
```

Source: [Power Usage Effectiveness](https://www.thegreengrid.org/node/372) — The Green Grid. The definition of facility energy divided by IT equipment energy.

## Preserve output quality

MLCommons develops MLPerf power measurement methods to connect energy use with benchmark performance. The relevant lesson is to measure completed work and its acceptance criteria alongside electricity.

A model that saves power but produces unusable output is not an equivalent improvement. Neither is a batched service that misses the required deadline. Quality, latency and reliability must remain inside the agreed envelope.

Source: [Power measurement for MLPerf](https://mlcommons.org/working-groups/benchmarks/power/) — MLCommons. Methods connecting measured power, energy and benchmark performance.

## Reading the homepage model

The homepage uses a baseline of 100 relative units. An assumed 35% improvement yields 65 units for the same workload. The slider changes an assumption; it does not gather infrastructure measurements.

A future benchmark should replace assumed savings with repeated measurements, documented baselines and a distribution of results rather than one best run.

```text
Candidate energy = Baseline × (1 − assumed saving / 100)
```

---

# SI & computation

Choose the right task, model, hardware and moment—then measure the complete result.

Type: Research + design · Reviewed: 4 October 2026

## The intelligence layer

SI is the intelligence theme in SpaceXSI. The practical research here concerns learning-based systems, inference and resource optimization. Original source titles retain the terminology used by their publishers.

The useful question is which system can satisfy a task reliably within an energy and latency budget. Some tasks suit small specialized models; others require larger systems. Conventional software remains appropriate where a model contributes little value.

## Training and inference

Training develops model parameters from data. Inference applies a trained model to new inputs. Their execution patterns and constraints differ, so they need separate evaluations.

Measure the full request lifecycle: input preparation, execution, memory movement, postprocessing and retries. Token count or accelerator activity alone is not a complete definition of useful output.

## An optimization stack

NVIDIA’s TensorRT documentation describes lower-precision inference formats and quantization schemes. Whether an implementation saves energy must be tested on its hardware and workload; precision alone supplies no universal savings percentage.

| Technique | Opportunity | Trade-off to test |
| --- | --- | --- |
| Task routing | Use a sufficient smaller system | Routing mistakes and quality regressions. |
| Quantization | Reduce numerical precision and storage | Accuracy, hardware support and overhead. |
| Batching | Share execution resources | Queue time and latency. |
| Caching | Reuse valid results | Staleness, privacy and hit rate. |
| Flexible scheduling | Move work within deadlines | Data transfer and delayed completion. |

Source: [Quantization schemes](https://docs.nvidia.com/deeplearning/tensorrt/latest/inference-library/quantized-types-schemes.html) — NVIDIA · TensorRT documentation. Lower-precision inference formats and their implementation constraints.

## Space-native intelligence

NASA’s EO-1 mission used the Autonomous Science Experiment for onboard intelligent scheduling. This is a historical example of useful autonomy in orbit.

A research opportunity is to extract relevant information near the sensor before downlink. Our inference is that compact results may reduce communication requirements, but that benefit must be weighed against onboard power, compute mass and reliability.

Source: [Earth Observing-1: Ten Years of Innovation](https://science.nasa.gov/earth/earth-observatory/earth-observing-1-ten-years-of-innovation/) — NASA. A historical example of onboard intelligent scheduling and autonomous science.

## A proposed evaluation sequence

- Publish a fixed task suite and quality criteria.
- Change one optimization at a time against a documented baseline.
- Measure energy, latency, accuracy and retries together.
- Include warm-up and idle conditions in the report.
- Repeat under representative loads rather than only full utilization.

---

# Solar & orbital power

Sunlight is an opportunity. Reliable delivered electricity is an engineered system.

Type: Research + examples · Reviewed: 4 October 2026

## Separate two concepts

Using solar electricity onboard a spacecraft and beaming power from space to Earth address different problems. Onboard generation supplies local equipment. Power delivery to Earth adds conversion, transmission and receiving infrastructure.

NASA’s 2024 study considered space-based power delivery for a possible 2050 operating horizon. Under its assumptions, costs exceeded terrestrial sustainable alternatives, while capability gaps could change future economics. These conclusions should not be substituted directly for onboard-compute economics.

Source: [Space-based solar power study](https://www.nasa.gov/organizations/otps/space-based-solar-power-report/) — NASA · 2024. A 2050-oriented study of power delivery to Earth, distinct from onboard electricity use.

## Generation, storage and mission life

NASA’s spacecraft power guidance accounts for eclipse periods, solar angles and end-of-life performance. Panels cannot generate from direct sunlight in Earth’s shadow; usable output may degrade with mission age.

Generation must supply the compute load, spacecraft services, losses and storage recharge. A sun-synchronous orbit is not automatically a guarantee of continuous sunlight in every configuration and season.

Source: [Small spacecraft: power subsystems](https://www.nasa.gov/smallsat-institute/sst-soa/power-subsystems/) — NASA. Solar arrays, storage, eclipses and end-of-life power budgeting.

## A simplified storage calculation

For an assumed 10 kW total load during a 30-minute eclipse, 5 kWh must reach the load. At an assumed 80% usable storage fraction, nominal capacity starts at 6.25 kWh, before extra losses and reserve.

This is a planning example, not a spacecraft specification. Real design also includes degradation, charge/discharge limits, thermal behavior and failure cases.

```text
E_storage ≥ P_eclipse_load × eclipse duration / usable storage fraction
```

## Compare power architectures

| Architecture | Potential benefit | Required evidence |
| --- | --- | --- |
| Grid-connected compute | Mature operation and servicing | Local supply, overhead and emissions. |
| Ground renewables + storage | Cleaner supply and flexibility | Hourly matching and storage losses. |
| Orbital onboard solar | Power where spacecraft work occurs | Sunlight profile, mass and mission life. |
| Power beaming to Earth | Remote orbital generation | Conversion chain, reception and lifecycle economics. |

## A grounded research direction

Start with terrestrial experiments that can be repeated. Treat orbital generation as a use-case-specific frontier. Compare delivered useful work across the full system rather than assuming one location is inherently best.

---

# Computing in orbit

Power, heat, communications, deployment and lifetime have to work together.

Type: Research + examples · Reviewed: 4 October 2026

## Space is not free cooling

NASA’s thermal guidance explains that vacuum removes convection: heat moves through internal conduction and exchanges with the external environment through radiation. Electronics still need a thermal path to a surface capable of rejecting heat.

Radiator area, temperature, emissivity, orientation and incoming solar or planetary radiation affect the balance. A mission must manage both useful heat rejection and unwanted heating.

Source: [Small spacecraft: thermal control](https://www.nasa.gov/smallsat-institute/sst-soa/thermal-control/) — NASA. Heat transfer in vacuum and spacecraft thermal management.

## A first-order radiator calculation

Our idealized calculation assumes emissivity 0.9, a 300 K radiator and negligible incoming heat. Using σ ≈ 5.67 × 10⁻⁸ W·m⁻²·K⁻⁴ gives roughly 413 W per square metre. Rejecting 1 MW needs about 2,420 m² of effective emitting area.

Effective emitting area is not necessarily panel footprint: geometry may allow both sides to contribute. Real design includes view factors, thermal gradients, environmental heat and margin.

```text
Q ≈ ε × σ × A × (T_radiator⁴ − T_environment⁴)
```

Source: [Small spacecraft: thermal control](https://www.nasa.gov/smallsat-institute/sst-soa/thermal-control/) — NASA. Heat transfer in vacuum and spacecraft thermal management.

## Match work to communications

A job importing large datasets from Earth has a different burden from a job processing data already in orbit. Sustained transfer, contact interruptions and result size matter alongside peak link speed.

Turyshev’s April 2026 preprint models power, thermal rejection, communications, utilization and mission lifetime together. It identifies space-native preprocessing and relay-integrated edge work as plausible early regimes. Its conclusions depend on the model assumptions.

Source: [Orbital Data Centers: Spacecraft Constraints and Economic Viability](https://arxiv.org/abs/2604.27197) — Slava G. Turyshev · April 2026 preprint. Model-dependent analysis of power, mass, communications, utilization and lifetime. A preprint, not demonstrated operation.

## Evaluate the lifecycle

The ASCEND consortium’s 2024 study summary says substantial emissions improvements would require a launcher with ten times lower lifecycle emissions in its study context. Avoiding water-based cooling does not settle manufacturing, deployment and replacement impacts.

ESA’s 2026 environment report reinforces the importance of orbital sustainability. A credible infrastructure proposal includes collision avoidance and end-of-life disposal along with its payload budget.

Source: [ASCEND feasibility study results](https://www.thalesaleniaspace.com/en/press-releases/thales-alenia-space-reveals-results-ascend-feasibility-study-space-data-centers-0) — Thales Alenia Space · 2024. A consortium summary of orbital data-centre feasibility and lifecycle conditions.

Source: [Space Environment Report 2026](https://www.esa.int/Space_Safety/Space_Debris/ESA_Space_Environment_Report_2026) — ESA · September 2026. Orbital sustainability, reentry and disposal trends.

## Study the workload before the scale

| Workload | Why investigate it | Main constraint |
| --- | --- | --- |
| Observation preprocessing | Inputs originate in space | Onboard power and reliability. |
| Relay-integrated processing | Work occurs within a communications system | Link availability and integration. |
| Earth-facing batch compute | Timing may be flexible | Transfer and delivered cost. |
| Distributed training | A long-term capacity concept | Interconnect, replacement and utilization. |

> Research synthesis: this comparison identifies questions to test, not deployed SpaceXSI services.

---

# Crypto & coordination

Coordinate the work transparently without replicating every computation onchain.

Type: Design direction · Reviewed: 4 October 2026

## A focused role for crypto

Crypto can provide shared records between participants who do not rely on one operator. Candidate SpaceXSI uses include contribution records, benchmark commitments, resource access and settlement of accepted work.

The efficiency thesis favors separating execution from agreement about execution. Repeating a heavy workload on every validator would add avoidable work.

## Consensus matters

Ethereum’s documentation explains how proof-of-stake replaces energy-intensive mining with stake-based consensus. This illustrates the importance of network architecture, but a saving percentage always depends on the baseline and system being compared.

Evaluate operational energy, security assumptions, throughput, settlement time and data availability together. One fee or transaction metric cannot resolve the entire comparison.

Source: [Ethereum energy consumption](https://ethereum.org/energy-consumption/) — ethereum.org. The energy implications of proof-of-stake; reported estimates have specific dates and boundaries.

## Offchain execution, onchain commitments

A hash can show that a record has not changed. It cannot establish the truth of a physical energy measurement. Meter calibration, trusted collection and independent checks remain necessary.

- Execute a task on suitable compute infrastructure.
- Collect a signed result and measurement record.
- Retain detailed evidence outside the settlement layer under an explicit policy.
- Commit a content reference or hash that makes later changes detectable.
- Accept and settle according to the declared verification process.

## Recognize outcomes rather than activity

A proposed contribution mechanism could recognize accepted work inside a quality and resource envelope. It must address duplicate tasks, forged telemetry, selective reporting and collusion.

Potential controls include challengeable records, sampling and independent reruns. These controls also consume resources; their overhead belongs in the system-level efficiency calculation.

## Resolve the utility before the token

Network, supply, allocations and contract mechanics should follow these requirements. These docs establish design questions rather than inventing token parameters.

| Design question | Requirement |
| --- | --- |
| What action needs a token? | A specific coordination function. |
| What is priced? | An accepted-work unit and service envelope. |
| How is evidence trusted? | Attestation, audit and dispute rules. |
| Who changes parameters? | A defined governance scope. |
| Can the same goal be simpler? | Comparison with non-token alternatives. |

---

# System architecture

A modular path from task request to execution, evidence and coordination.

Type: Proposed architecture · Reviewed: 4 October 2026

## Five system layers

| Layer | Responsibility | Output |
| --- | --- | --- |
| Request | Define work and constraints | Task specification and acceptance test. |
| Scheduling | Select eligible resources | Assignment and policy trace. |
| Execution | Run the workload | Result and environment record. |
| Measurement | Capture resources and quality | Evidence with explicit boundaries. |
| Coordination | Record acceptance and settlement | Commitment and dispute status. |

## The workload lifecycle

- A requester defines inputs, deadline, location restrictions and acceptance criteria.
- A scheduler filters resources by capability and capacity.
- An operator executes the work and collects telemetry.
- A verifier checks the result and the evidence policy.
- An accepted record can be committed to a coordination layer and linked to settlement.

## A proposed evidence record

These are candidate fields, not a finished API. They should explain how a measurement was produced without exposing sensitive input data.

Comparable records need common definitions. If one operator counts only accelerator power while another counts the whole server, a shared leaderboard would mislead even if both readings were accurate.

```text
workload_id · software_digest · hardware_profile
measurement_boundary · start_time · duration
energy_used · accepted_work · quality_result
operator_signature · evidence_reference
```

## Represent incomplete evidence

A result can exist while its efficiency claim remains unverified. Missing telemetry should never silently become a zero-energy record. Separate output validity from measurement validity.

Differences between repeated runs may reflect variability rather than misconduct. Verification should distinguish unstable results, missing metadata, invalid outputs and policy violations.

## Keep execution environments modular

An initial prototype can use ground infrastructure. Orbital research can remain an additional environment rather than a prerequisite for every module.

This lets SpaceXSI test a narrow proposition first: whether transparent, resource-aware coordination improves a defined workload after all system overhead. Further scope should follow evidence.

---

# Future pathways

The future is a sequence of testable conditions, not one promised launch date.

Type: Research synthesis · Reviewed: 4 October 2026

## Improve existing infrastructure first

The nearest research path tests efficient execution on accessible ground resources. Model selection, utilization, scheduling and overhead can be evaluated separately.

Advance when savings repeat without unacceptable quality, latency or reliability penalties, and survive independent reruns under realistic load.

## Connect resource networks

A second path coordinates operators with different resources. Moving work adds transfer costs, privacy restrictions and timing uncertainty.

The decision gate is net benefit after communications, verification and settlement. The locally fastest resource may not be the best system-wide choice.

## Explore space-native work

Orbital-compute research suggests examining work whose inputs originate in space before assuming general-purpose terrestrial replacement. Our inference is to select workload fit before infrastructure scale.

An experiment should specify power, thermal limits, input/output traffic, fault behavior and lifetime. Progress requires meeting these constraints simultaneously.

Source: [Orbital Data Centers: Spacecraft Constraints and Economic Viability](https://arxiv.org/abs/2604.27197) — Slava G. Turyshev · April 2026 preprint. Model-dependent analysis of power, mass, communications, utilization and lifetime. A preprint, not demonstrated operation.

## Conditions for larger orbital systems

- Lower combined cost of spacecraft construction and deployment.
- Durable power and radiator hardware over a useful mission lifetime.
- Sustained communications matched to the workload.
- High utilization with a realistic replacement or servicing plan.
- Lifecycle assessment including manufacture, launch and disposal.

Source: [Orbital Data Centers: Spacecraft Constraints and Economic Viability](https://arxiv.org/abs/2604.27197) — Slava G. Turyshev · April 2026 preprint. Model-dependent analysis of power, mass, communications, utilization and lifetime. A preprint, not demonstrated operation.

Source: [ASCEND feasibility study results](https://www.thalesaleniaspace.com/en/press-releases/thales-alenia-space-reveals-results-ascend-feasibility-study-space-data-centers-0) — Thales Alenia Space · 2024. A consortium summary of orbital data-centre feasibility and lifecycle conditions.

## A proposed research roadmap

These phases are a proposed sequence, not completed milestones or fixed dates. Each should produce a reviewable deliverable before the next expands the scope.

| Phase | Deliverable | Evidence to advance |
| --- | --- | --- |
| Define | Public task and measurement protocol | Clear quality gates and boundaries. |
| Measure | Baseline and optimized logs | Repeatable comparative results. |
| Coordinate | Verification and scheduling prototype | Benefit including system overhead. |
| Validate | Independent reproducibility review | Resolved discrepancies. |
| Extend | Additional-resource feasibility study | A complete use-case model. |

---

# Glossary

A shared vocabulary across energy, computation, space and coordination.

Type: Reference · Reviewed: 4 October 2026

## Energy

| Term | Meaning |
| --- | --- |
| Watt (W) | Power: one joule per second. |
| Kilowatt-hour (kWh) | Energy: 1 kW sustained for an hour, equal to 3.6 million joules. |
| Terawatt-hour (TWh) | One billion kilowatt-hours. |
| Energy intensity | Energy per defined unit of useful output. |
| PUE | Facility energy divided by IT equipment energy. |
| Carbon intensity | Emissions attributed per electricity unit, dependent on time, place and method. |
| Embodied impact | Impact from producing and deploying hardware, separate from operation. |

Source: [Power Usage Effectiveness](https://www.thegreengrid.org/node/372) — The Green Grid. The definition of facility energy divided by IT equipment energy.

## Computation

| Term | Meaning |
| --- | --- |
| SI | The intelligence theme in SpaceXSI, explored through computation and optimization. |
| Inference | Applying a trained model to new inputs. |
| Quantization | Representing values at reduced precision. |
| Batching | Processing requests together to share execution resources. |
| Utilization | Used capacity as a fraction of available capacity. |
| Quality gate | The acceptance test for an output to count as useful. |
| Attestation | A verifiable statement about an execution or measurement. |

Source: [Quantization schemes](https://docs.nvidia.com/deeplearning/tensorrt/latest/inference-library/quantized-types-schemes.html) — NVIDIA · TensorRT documentation. Lower-precision inference formats and their implementation constraints.

## Space systems

| Term | Meaning |
| --- | --- |
| LEO | Low Earth orbit; behavior depends on altitude, inclination and geometry. |
| Sun-synchronous orbit | An orbit maintaining approximately consistent local solar time at crossings through precession. |
| Eclipse | A period shadowed from direct sunlight. |
| Radiator | A surface engineered to emit heat as radiation. |
| Downlink | Data transfer from spacecraft to Earth. |
| End-of-life performance | Capability expected after mission aging. |
| Lifecycle assessment | Evaluation across production, deployment, operation and disposal. |

Source: [Small spacecraft: power subsystems](https://www.nasa.gov/smallsat-institute/sst-soa/power-subsystems/) — NASA. Solar arrays, storage, eclipses and end-of-life power budgeting.

Source: [Small spacecraft: thermal control](https://www.nasa.gov/smallsat-institute/sst-soa/thermal-control/) — NASA. Heat transfer in vacuum and spacecraft thermal management.

## Coordination

| Term | Meaning |
| --- | --- |
| Proof-of-stake | Consensus whose primary mechanism uses staked assets rather than competitive mining. |
| Onchain | Recorded or executed in shared blockchain state. |
| Offchain | Performed outside that shared state. |
| Commitment | A compact reference binding a participant to data. |
| Settlement | Finalizing an accepted transfer or obligation. |
| Oracle | A mechanism for introducing external information into a blockchain. |

Source: [Ethereum energy consumption](https://ethereum.org/energy-consumption/) — ethereum.org. The energy implications of proof-of-stake; reported estimates have specific dates and boundaries.

---

# Sources & methodology

Primary references, explicit assumptions and traceable research. Reviewed on 4 October 2026.

Type: Source library · Reviewed: 4 October 2026

## Research approach

This edition uses energy-agency analysis, spacecraft engineering references, benchmark documentation, technical documentation and research papers. References appear beside the sections they support.

Forecasts retain their publication year and scenario context. A 2025 outlook is not relabeled as a 2026 measurement. Company feasibility summaries describe studies; preprints are identified as model-based research.

## Evidence types

| Type | Meaning |
| --- | --- |
| Research | Technical context attributed to an external source. |
| Worked example | Our calculation with selected assumptions. |
| Design direction | A proposed SpaceXSI principle or system. |
| Research synthesis | Our interpretation connecting published findings. |
| Project result | A documented implementation and measurement; no results are asserted here. |

## Scope of this edition

The docs form a research and design foundation covering energy, useful computation, space systems and crypto coordination. They are not a completed spacecraft specification or token-economics model.

Original source titles are preserved for traceability. Updates should identify changes, review dates and the evidence supporting new claims.

## Primary source library

Source: [Energy and AI: energy demand](https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai) — IEA · 2025. Data-centre demand estimates and scenario-based forecasts. Sector totals include workloads beyond intelligence.

Source: [Power Usage Effectiveness](https://www.thegreengrid.org/node/372) — The Green Grid. The definition of facility energy divided by IT equipment energy.

Source: [Power measurement for MLPerf](https://mlcommons.org/working-groups/benchmarks/power/) — MLCommons. Methods connecting measured power, energy and benchmark performance.

Source: [Carbon-Aware Computing for Datacenters](https://arxiv.org/abs/2106.11750) — Radovanović et al. · 2021 paper. Scheduling flexible compute around time-varying grid carbon intensity.

Source: [Quantization schemes](https://docs.nvidia.com/deeplearning/tensorrt/latest/inference-library/quantized-types-schemes.html) — NVIDIA · TensorRT documentation. Lower-precision inference formats and their implementation constraints.

Source: [Small spacecraft: power subsystems](https://www.nasa.gov/smallsat-institute/sst-soa/power-subsystems/) — NASA. Solar arrays, storage, eclipses and end-of-life power budgeting.

Source: [Small spacecraft: thermal control](https://www.nasa.gov/smallsat-institute/sst-soa/thermal-control/) — NASA. Heat transfer in vacuum and spacecraft thermal management.

Source: [Space-based solar power study](https://www.nasa.gov/organizations/otps/space-based-solar-power-report/) — NASA · 2024. A 2050-oriented study of power delivery to Earth, distinct from onboard electricity use.

Source: [ASCEND feasibility study results](https://www.thalesaleniaspace.com/en/press-releases/thales-alenia-space-reveals-results-ascend-feasibility-study-space-data-centers-0) — Thales Alenia Space · 2024. A consortium summary of orbital data-centre feasibility and lifecycle conditions.

Source: [Orbital Data Centers: Spacecraft Constraints and Economic Viability](https://arxiv.org/abs/2604.27197) — Slava G. Turyshev · April 2026 preprint. Model-dependent analysis of power, mass, communications, utilization and lifetime. A preprint, not demonstrated operation.

Source: [Earth Observing-1: Ten Years of Innovation](https://science.nasa.gov/earth/earth-observatory/earth-observing-1-ten-years-of-innovation/) — NASA. A historical example of onboard intelligent scheduling and autonomous science.

Source: [Space Environment Report 2026](https://www.esa.int/Space_Safety/Space_Debris/ESA_Space_Environment_Report_2026) — ESA · September 2026. Orbital sustainability, reentry and disposal trends.

Source: [Ethereum energy consumption](https://ethereum.org/energy-consumption/) — ethereum.org. The energy implications of proof-of-stake; reported estimates have specific dates and boundaries.

## Updating the research

- Recheck forecasts when newer editions are available.
- Publish benchmarks with identifiable protocols and measurement boundaries.
- Mark milestones complete only with linked deliverables.
- Revisit orbital conclusions as demonstrations and lifecycle assessments emerge.
- Retain assumptions so readers can reproduce calculations.