31 Verified Sources

Primary source — TERAfab event, AI Sat Mini dimensions (170m+), D3 chip, Terafab announcement, mass driver on Moon.

TERAfab $20–25B, AI Sat Mini 100kW, future satellites megawatt range, SpaceX IPO $50B summer 2026.

FCC filing details: 500–2000km altitude, Ka-band, 100kW/tonne target, 1M tonnes/year = 100 GW.

SpaceX+xAI merger. Deutsche Bank: parity well into 2030s. Musk claim: 2–3 years.

10,000 launches/year target. Musk: orbital AI = all terrestrial AI annually within 5 years.

80% D3 space chips, 20% AI5 terrestrial. Tesla CFO: $20–25B not in 2026 capex plan.

Prof. Hugh Lewis: 40,000 maneuvers/day baseline, 100,000 upper bound, 14.5M–36.5M/year. Prof. Lawler: constellation size underestimated.

1 GW orbital DC = 4,300 satellites, 30M kg. Cost >$50B over 5 years = 3× terrestrial equivalent.

GPU failure rate 9%/year (Meta Llama 3 basis). Kessler above 700km already active. Prof. Aaron Boley: cascading collision risk.

ISS EATCS: 70kW across 422m² radiators. 1 GW orbital DC would need 834,000m² radiators. 1M sats by late 2030s risk.

Thermal wall analysis. Obsolescence trap: GPU doubles every 2yr, satellite lasts 5–6yr. Google TPU radiation test: irregularities <3yr.

Rad-hard chips: 10–15 years behind current tech. Thermal cycling: solder joint fatigue. SEU: bit flips corrupt AI training.

MRAM vs Flash vs ReRAM for LEO. Avalanche MRAM: gigabits now available. Flash insufficient for long-term radiation. Bob Patti (NHanced).

Space-grade MRAM: SEU threshold 84× higher than Flash. Expected time to first event in LEO: 500 years. TMR+EDAC integrated on-die.

No standardised BoM for orbital compute. No rad-tolerant CPU/GPU/memory sourcing framework. Lack of interoperability = several times more expensive than terrestrial.

MoffettNathanson: 100 GW orbital = $4–5 trillion capex in Nvidia equipment. Top 5 hyperscalers combined: $400B capex 2025.

Google Suncatcher: launch cost must fall below $200/kg by 2035. Saarland University: orbital DC = 10× more emissions than terrestrial (Dirty Bits in LEO).

Satellite cost: $100K–$200K/kg. Launch cost: $1,000–$10,000/kg. Fuel: ~$10/kg. 5–10% annual cost decline since 2000.

Market: $1.77B by 2029, $39B by 2035 (67.4% CAGR). Google TPU v6e: confirmed 5-year LEO radiation tolerance. Five-year radiation lifespan = continuous hardware refresh demand.

Version 1 baseline: 70kW/sat, 5–10 MW per Starship launch. Texas Starbase: 25/yr. Florida LC-39A: 44/yr approved. IPO $50B closes summer 2026.

TCO dominated by component qualification, in-orbit servicing, radiation lifespan, ground-segment integration and replacement cadence.

1967 Outer Space Treaty liability framework. Google study: $200/kg breakeven ~2035 at 180 Starship launches/year. Sun-synchronous orbit resource constraints.

60m underground limestone mine. 35-acre water reservoir for cooling. Year-round stable temperature. Tier 3 standards. Reference operational case.

EGS could provide 120+ GW baseload in continental US. Capacity factor >90%. Rhodium Group: $50/MWh at geothermal-sited DC vs $75/MWh grid-sited.

>10cm: 40,000+ tracked. 1cm–10cm: 1,000,000+. 1mm–1cm: 130,000,000+. 1cm at 7km/s = grenade energy. Kessler active 700–1500km.

LEO investment: $45B in 2025 (up from $25B in 2024). Nvidia GTC 2026: Space-1 Vera Rubin Module, 25× AI compute vs H100. SpaceX IPO = 'Netscape moment'.

Starlink as of March 23, 2026: 11,641 launched, 10,116 operational. TERAFAB $25B confirmed. Astronomers: 'completely destroy the night sky'.

Jonathan McDowell: 1.7M proposed satellites total worldwide. 'Extremely difficult to operate such a huge number safely.' FCC public notice February 4.

Launch/Costs Economics

Key facts (expert validation of white paper Starship Math & IPO narrative arguments):
• SpaceX announced up to 75 Starship flights for 2026 — as of late March 2026, zero have taken place.
• Last Starship flight: nearly 6 months ago. No Starship has yet completed a full orbit.
• Next flights remain suborbital, without ship recovery. Starship V3 Super Heavy static fire expected from April 9.
• K2 Space (betting on mass abundance in orbit): >$3B valuation — unicorn status — despite the premise unproven.
• Buffenoir conclusion: "Cette hypothèse bénéficie déjà d'un soutien massif… dans un contexte où sa démonstration reste encore incomplète."

Follow-up comments (same thread):
• Buffenoir: "Ce qui est marquant, c'est moins de savoir si Starship va réussir… que le fait que certains acteurs industriels prennent déjà des décisions comme si c'était acquis." — This is the white paper's IPO narrative argument formulated by an independent space industry expert.
• Exchange with Hernandez (Devil-Hop) on profitable niches: Buffenoir confirms that even "niche profitable use cases" (messaging, GSM terminals) depend on launch cost evolution — validating the white paper's stratified conclusion between niche orbital edge and mass-market datacenter.

Kessler/Debris Physics

Mathematical foundation for the conjunction/manoeuvre problem at 1M satellite scale.
• Conjunctions grow exponentially (not linearly) with satellite count — catastrophic at 1M
• "Replacement Flux": constant launch/deorbit transit permanently congests LEO
• Zero failure rate statistically impossible: 99.9% reliability → thousands of dead satellites
• Kessler Syndrome described as "mathematical near-certainty" at this density
• Data basis: SpaceX's own FCC semi-annual reports — Starlink conjunction manoeuvres doubled every 6 months, reaching 100,000+ in 2024 for 10,000 satellites. Projection to 1M = billions/year.
Context: Lewis updated his Kessler analysis with Donald Kessler himself, finding 800–1,000km already above runaway threshold (9th European Conference on Space Debris, April 2025).

Kessler/Debris SLA & Legal Economics

Formal legal objections filed with the FCC on record — three structural impossibilities:
• Section VI-F — "The 0.8% Operational Tax": satellites constantly maneuvering burn propellant and duty-cycle time that should be spent on data processing. Compute capacity permanently degraded by avoidance burden.
• Section VI-H — LEO Monopolisation: SpaceX's "Replacement Flux" (constant launch + deorbit) essentially claims the entirety of LEO for itself, making safe launch impossible for any other operator.
• Section VI-I — "The Debris Factory": at 1M satellites, even a 99.9% success rate leaves thousands of dead, unguided projectiles in orbit. A zero-failure rate is statistically impossible at this scale.
Stewart coined the terms "Impossible Cadence" and "Maneuver Paradox" for the logical contradictions at the core of the SpaceX filing. Both are validated by Prof. Lewis's mathematical analysis [source 30]. These are formal filings, not commentary — the FCC must address them.