Give each process
its own environment.
Cluster thermal, solution-growth and particle-assembly campaigns where they share conditions. Reduce changeovers and protect sensitive processes.
03 / BEYOND ONE FOUNDRY
From a factory.
To a frontier.
A network of specialized foundries. Shared standards. Coordinated capacity. One mission: make orbit a useful part of Earth’s industrial system.
One foundry proves a process and its return path. Concept illustration; no operating fleet is claimed.
FUTURE-STATE CONCEPTORBITAL PATHS & SCALE ARE ILLUSTRATIVE
03.1 / WHY A NETWORK?
Scale the system.
Specialize the work.
A fleet could let each foundry do what it does best, while a shared operating layer coordinates the whole production journey.
Give each process
Match capacity
to actual demand.
Route jobs among qualified vehicles as queues and resource availability change. Capacity is useful only if the right process can run there.
Plan backwards
from the return.
Coordinate growth, preservation and return windows so an accepted batch reaches Earth when it is needed.
03.2 / THE SCALE IS ARRIVING
The ride is getting cheaper.
The demand is getting real.
A network makes sense only if transport scales, products can come home often enough, and customers need what orbit makes. We measured all three from public launch logs, return records, clinical-trial registries and the research literature.
192 GBpublic data processed
256Mrecords analysed
64Slurm jobs on MIT Engaging
36public datasets
How we did it ↗03.3 / TEST THE BENEFIT OF COORDINATION
More intelligent.
Only if it performs better.
We compare four strategies on identical synthetic jobs. Change the workload to see when coordination helps—and when simpler dedicated operation is stronger.
SYNTHETIC SIMULATION
3 capsules · 120-hour horizon
100 paired random seeds per scenario
ON-TIME DELIVERY / % OF ALL JOBSHIGHER IS BETTER
Whiskers: bootstrap 95% intervals over seed-level results. Real prices, orbital mechanics and vehicle return cycles are not modeled.
24operating scenarios
100paired seeds per scenario
4strategies compared
0claimed flight operations
REALITY CHECK FROM THE RETURN RECORD
This model assumes a return opportunity every 12–24 hours. The public landing record says otherwise: since 2021 a capsule that a US commercial payload could ride has landed every 45 days at the median. Return, not launch, is the bottleneck, and it is why a network adds value: each foundry that carries its own return capsule adds return opportunities. Read these on-time rates as a test of scheduling logic, not as deliveries today's vehicles could make. See the return record ↗
In the initial benchmark, coordinated v0 delivered 25.0% on time versus 44.6% for dedicated operation. Across the sensitivity sweep, dedicated leads 12 scenarios by mean, coordinated 6, earliest due 4 and FIFO 2. Small differences are not necessarily statistically resolved. Full benchmark and uncertainty ↗
03.4 / AN INDUSTRY, BUILT IN STAGES
Think in decades.
Prove the next experiment.
01NOW
THE SCIENCE CAMPAIGN
Find a material worth flying.
Benchmark Earth controls, validate image measurements and test a functional hypothesis. Pick one leading candidate rather than spreading the first mission across all five.
DECISION GATE / REPRODUCIBLE APPLICATION BENEFIT02NEXT
THE FIRST FOUNDRY
Turn a recipe into a reliable service.
Qualify a cartridge and its return path. Add a second compatible process only after shared-resource and disturbance tests pass.
DECISION GATE / ACCEPTED YIELD + MEASURED COMPATIBILITY03EXPAND
THE SPECIALIZED FLEET
Let different foundries do different work.
Separate high-temperature work from quiet growth. Standardize interfaces, batch histories and qualification while adapting each vehicle to its demand.
DECISION GATE / DEMAND + DELIVERY + COST PER ACCEPTED BATCH04HORIZON
THE ORBITAL INDUSTRIAL NETWORK
Make orbit part of the supply chain.
Connect research labs, orbital production, Earth finishing and customers. Explore on-orbit servicing and replenishment when transport and economics make them credible.
DECISION GATE / RELIABLE LOGISTICS + SUSTAINABLE UNIT ECONOMICS03.5 / BIG DATA, NEXT
Run the network
on live data.
At fleet scale, the operating system is a data system. Demand, capacity, orbits and returns become streams that a scheduler can act on, tested first against simple policies.
Read demand from registries.
Track subcutaneous-biologic trials, drug approvals and battery-materials markets to forecast which cartridges the fleet should carry.
Calibrate returns from real flights.
Replace synthetic return cadence with live launch and splashdown manifests, then re-run the fleet benchmark on fresh seeds.
Plan shells, not just satellites.
Use the full tracking catalogue and operator ephemerides to place foundries where drag, sunlight, crowding and return geometry balance.
Learn across foundries.
Pool batch records from every foundry so each process family improves from the whole fleet's experience, not one vehicle's.
WHAT EACH ANALYSIS ASKS FOR NEXT
Loading the extension plan from each analysis…
03.6 / WHO WE CALL FIRST
Partners for
an orbital industry.
Transport at scale, return, stations and first customers. Proposed collaborations only; no partnership is claimed.
THE LONG VIEW
The value of space
is what we bring
back to Earth.
We’re building the evidence, the architecture and the operating logic for a new kind of factory.
Follow the work ↗