Speculative syndicates · not a 10-K
Next 5 / 10 / 15 / 20 years
Four industrial desks that are not Tesla or SpaceX products. They assume Starship becomes a trucking line. Percentages and dollars are models, labeled as such. A render is not a constellation.
Reconstruction — LEO compute modules with solar wings. Not a Tesla/SpaceX product. Not on orbit. 
Reconstruction — vacuum radiator and unlabeled racks. Cryogenic cooling in vacuum is the cited physics, not a shipped SKU. Reconstruction · Orbital Compute Network — Reconstruction clip — orbital dolly past unlabeled solar wings. A render in motion is still a render. Y+5 · space-based AI infrastructure
Y+5 horizon · 2026–2031
Orbital Compute Network
Solar-powered, high-density AI clusters in LEO. Vacuum is the radiator. Unobstructed sun is the grid. The pitch is to get training and inference off terrestrial water and peaker plants.
- Cited capex
- $15.0B–$30.0B
- US GDP lift (model, $B/yr at horizon)
- $18.0B–$55.0B
- Low / High
- US jobs (thousands, model)
- 40–95k
- Cited complete-by
- 2031
- Y+5
Feasibility heat
34
Starship dependency
92
Grid relief
48
Water relief
62
Cited human benefit
Relieves grid and water load from Earth-side training clusters. Medical and science models keep scaling without a corresponding spike in consumer kWh prices — if the constellation actually flies.
Forensic cut
Model, not a program. Starship cadence and V3 mass-to-orbit are the gate. Colossus on the ground is the thing that exists. An orbital rack is a reconstruction. Heat 34/100 of a flying datacenter business, not of a render.
What has not shipped
No public constellation SKU. No FCC orbital-compute shell. No disclosed offtake from xAI/Colossus. Starship is not a scheduled trucking line.
Open a living dependency
Reconstruction — unmarked autonomous capture aircraft. A vapor trail is not a gigaton. 
Reconstruction — orbital sun-shade lattice. A sail between Earth and Sun is a treaty, then a product. Reconstruction · Planetary Geo-Engineering — Reconstruction clip — unmarked aircraft over dark water. Mist is not a gigaton. Y+10 · atmospheric scrubbing + thermal management
Y+10 horizon · 2026–2036
Planetary Geo-Engineering
Gigaton-scale CO₂ capture with autonomous aerial and maritime fleets, plus high-altitude aerosol or orbital sun-shade options to manage thermal equilibrium. Climate as an industrial utility, not a pledge.
- Cited capex
- $50.0B–$100.0B
- US GDP lift (model, $B/yr at horizon)
- $12.0B–$40.0B
- Low / High
- US jobs (thousands, model)
- 80–180k
- Cited complete-by
- 2036
- Y+10
Feasibility heat
22
Starship dependency
40
Grid relief
18
Water relief
28
Cited human benefit
Directly attacks climate destabilization and agricultural-belt stress — if capture is real gigatons, not a pilot, and if sun-shades do not become a weapons-grade argument at the UN.
Forensic cut
The farthest from anything Tesla or SpaceX files today. Optimus-class autonomy is a cited labor input, not a climate P&L. Heat 22/100. Carbon-credit math is not physics. Do not launder this with Megapack GWh.
What has not shipped
No gigaton fleet. No sovereign climate contract on this desk. Sun-shades are a treaty fight, not a SKU. Optimus useful labor is still ~0.
Open a living dependency
Reconstruction — unmarked craft on a dark NEA. A grapple is not a mine. 
Reconstruction — empty deep-space refinery. Ice tanks and ingots. No yield on this card. Reconstruction · Asteroid Resource Mining — Reconstruction clip — grapple closing on a dark NEA. Ice glint is not a PGM offtake. Y+15 · deep-space extraction + refining
Y+15 horizon · 2026–2041
Asteroid Resource Mining
Intercept, capture, and refine near-Earth asteroids for platinum-group metals, rare earths, water ice, and construction-grade iron/silicon. Propellant from ice. Hulls from the rock. Earth mines become optional.
- Cited capex
- $100.0B–$250.0B
- US GDP lift (model, $B/yr at horizon)
- $25.0B–$90.0B
- Low / High
- US jobs (thousands, model)
- 30–70k
- Cited complete-by
- 2041
- Y+15
Feasibility heat
14
Starship dependency
96
Grid relief
8
Water relief
12
Cited human benefit
Ends the terrestrial mining footprint and the wars that sit on rare-earth maps — if transit is cheap and the first rock actually pays. Until then it is a slide.
Forensic cut
Requires mature, low-cost interplanetary transit. That is Starship operational reuse, not a catch photo. Heat 14/100. A grappled rock in a render is 0% of a refining P&L. Do not confuse HLS with a mine.
What has not shipped
No capture. No ISRU refinery. No offtake for asteroid PGM. Starship has not closed orbital reuse. Mars architecture is still a stacked slide.
Open a living dependency
Reconstruction — cyanobacteria ponds under an empty greenhouse. A pond is not an atmosphere. 
Reconstruction — seed vaults and pressure drums. Logistics for a biosphere that has not started. Reconstruction · Interplanetary Terraforming — Reconstruction clip — dust over empty ponds. A greenhouse is not a biosphere. Y+20 · biosphere logistics on Mars and moons
Y+20 horizon · 2026–2046
Interplanetary Terraforming
Mass deployment of engineered extremophiles, cyanobacteria, and localized pressure generators to kickstart habitation cycles. A utility for a second biosphere, not a flag-planting ceremony.
- Cited capex
- $500.0B–$2.00T
- US GDP lift (model, $B/yr at horizon)
- $8.0B–$35.0B
- Low / High
- US jobs (thousands, model)
- 20–60k
- Cited complete-by
- 2046
- Y+20
Feasibility heat
6
Starship dependency
99
Grid relief
4
Water relief
6
Cited human benefit
Multiplanetary redundancy for terrestrial DNA. A self-sustaining off-world biosphere is the cited hedge against impact and super-volcanism. A city it is not. A greenhouse still is not a biosphere.
Forensic cut
The Master Plan slide that keeps getting confused with a city. Heat 6/100 of a living world, ~0% of a 2046 metro. Optimus-as-Mars-labor is the crossover story. Multiplanetary city stays 0% on the board. Do not score a pond as an atmosphere.
What has not shipped
No cyanobacteria program of record. No pressure-generator SKU. No independent biosphere. Starship has not landed a city kit. HLS is a lander trench, not a climate.
Open a living dependency
Capex path, $B cumulative
Stacked model, not a budget. Orbital compute spends first. Terraforming is a second-generation bill.
US economy — low vs high lift
Annual $B at each syndicate’s horizon year. Low / high band. Not a CBO score.
US industrial lanes — model $B/yr
Where the cited US dollars would land at each horizon. Not a BEA table. Not a 10-K.
Feasibility heat
Starship dependency
Nearest vs farthest
Orbital compute versus terraforming on the same axes. Feasibility is not Starship dependency. GDP is scaled to 100. Model, not a scorecard.
Gates that have to be real
None of these four fly if Starship is still a test article. Click a gate to open the living trench.