Title: How SpaceX Makes Money
Show: App Economy Insights (How They Make Money)
Author: Bertrand Hartman
Date: 2026-05-26
URL: https://www.appeconomyinsights.com/p/how-spacex-makes-money
Note: Premium/paid Substack post, saved verbatim for personal study. Prose article — no (mm:ss) timestamps. Body reproduced as published.

---

🚀 How SpaceX Makes Money — Starlink funds the moonshots. (May 26, 2026)

SpaceX is going public; the rocket company is now an AI company too. The IPO filing landed last week, with xAI, Grok, and X folded inside. SpaceX aims for a $1.5–$2 trillion valuation with up to $75 billion raised, which would make it the largest IPO in history by both measures. The pitch: Starlink generates the cash; rockets bend the cost curve; AI and Mars are the moonshots.

1. Overview — SpaceX is the world's dominant launch provider, operator of the largest satellite network ever built, the Pentagon's primary access to orbit, and now the corporate home of xAI, Grok, and X following a February 2026 merger. The S-1 financials are retroactively rewritten so every prior period includes xAI and X. HQ Starbase, Texas; founded 2002 by Elon Musk; ticker SPCX (Nasdaq); mission "make life multiplanetary." Falcon 9 is the medium-lift workhorse; Starship is the next-gen fully reusable super-heavy vehicle (~5x more payload), still in development. "Railroad to space": reusability cut launch costs — NASA estimates Falcon 9 brought launch to ~$2,700/kg vs a historical ~$18,500 (85% reduction); one booster flew 34 times; Starship designed to take another 99% off. Three segments: 🚀 Space (Falcon, Dragon, Starship; NASA, Pentagon, intel agencies, commercial); 🛰️ Connectivity (Starlink consumer/enterprise/government/direct-to-mobile); 🤖 AI (xAI, Grok, X + compute). Scale: ~650 orbital launches, >80% of everything launched since 2023 by weight, 165 Falcon launches in 2025, >99% success; ~9,600 Starlink satellites (~75% of active maneuverable satellites), 10.3M subscribers across 164 countries. Financials: $18.7B FY25 revenue (+33% Y/Y), $20.7B FY25 CapEx (~5x in two years). Spends more than it makes.

2. Business Model — Space lowers cost to orbit (Falcon 9/Heavy, Dragon to ISS, Starship; also Starshield defense network; vertical integration; flew 11 of 12 NSSL missions and all 5 NASA crew/cargo flights in 2025; $0.7B FY25 operating loss only because it funded $3.0B Starship R&D in 2025 + $930M in Q1 2026; Starship pre-revenue until late 2026 — strip it out and launch is profitable). Connectivity = Starlink: $11.4B 2025 revenue (+50%), $4.4B operating income (+120%), $7.2B segment EBITDA (+86%) = 63% EBITDA margin; ARPU compressed $99 (2023) → ~$66 (Q1 2026) on emerging-markets + mobile; Starlink Mobile 7.4M devices, ~30 countries, partners T-Mobile and Vodafone; Sept 2025 EchoStar spectrum purchase (FCC-approved May 12, 2026) makes it vertically integrated. AI (messiest, most expensive): xAI + Grok + X + COLOSSUS/COLOSSUS II in Memphis (~1.0 GW training, "first coherent gigawatt-scale AI training cluster"); Grok trains on X's ~350M daily posts; a $1.25B/month Anthropic compute deal (May 2026–May 2029, ~$45B contracted revenue from a Grok competitor); Terafab — proposed $20B+ chip JV with Tesla and Intel (no binding terms; targets 1 TW annual compute); option to acquire Anysphere (Cursor) for ~$60B in Class A stock or a $10B breakup fee. AI is a $3.2B revenue business losing $6.4B from operations in FY25; most of the valuation premium above $1T sits here. Flywheel: rockets → Starlink cash → Starship/AI/Mars.

3. Financial Highlights — FY25: revenue $18.7B (+33%), operating loss ($2.6B), net loss ($4.8B), adj EBITDA $6.6B, OCF $6.8B, CapEx $20.7B, FCF ($13.9B). Segments: Space $4.1B rev / $0.7B EBITDA / ($0.7B) op loss (Starship R&D); Connectivity $11.4B / $7.2B EBITDA / 63% margin; AI $3.2B rev / ($6.4B) op loss / ($1.2B) EBITDA (losses grow as depreciation on $12.7B FY25 AI CapEx flows through). Q1 FY26: revenue $4.7B, op loss ($1.9B), CapEx $10.1B (AI $7.7B, Space $1.1B, Connectivity $1.3B), FCF ($9.1B); cash $24.7B (FY25) → $15.9B (Mar 31), ~$29B debt, ~$1.9B annual interest. Backlog $28.4B + $12.1B deferred revenue; Anthropic deal adds ~$45B not in backlog.

4. Risks — Starship execution (first V3 test flight succeeded May 22; operational payload not expected until H2 2026; V3 Starlink needs Starship not Falcon); AI cash burn/integration ("recently formed, still being integrated," $6.4B losses, $12.7B CapEx); Starlink pricing pressure (ARPU declining; Amazon Leo/Kuiper begins US service mid-2026 with 300+ satellites + Amazon's balance sheet); US government concentration (single-customer exposure); spectrum/regulatory dependence (FCC/FAA); Musk control + related-party complexity (Nasdaq "controlled company"; Tesla/xAI/X/Boring Company web); unproven moonshots (orbital AI compute, lunar, Mars, asteroid mining). $75B ≈ two years of runway at current burn; likely needs to raise again if timelines slip.

5. Management — Musk is CEO, CTO, Chairman; majority voting via Class B (10 votes vs 1); Nasdaq "controlled company." Gwynne Shotwell President/COO (operational architect since 2008); Bret Johnsen CFO; Antonio Gracias director (Valor Equity). Pay: Jan 2026 board granted Musk 1B performance RSUs (15 tranches) on top of ~5B-share stake; each tranche needs a market-cap milestone (up to $7.5T) AND the "Mars Colony Milestone" (a permanent ≥1M-inhabitant Mars colony) — Mars award ~$583B at top tranche; a second ~302M-share award tied to 100 TW orbital compute + a $6.56T market-cap milestone (~$154B). Combined ~$737B. Compensation is the IPO thesis in plain sight: not a 5-year DCF business.

6. Use of Proceeds — $75B would be the largest IPO ever (3x Saudi Aramco). Priorities: AI infrastructure (COLOSSUS expansion, Grok, early Terafab); Space (Starship to operational); Connectivity (V3 satellites, Starlink Mobile spectrum). Going public because the next chapter needs balance-sheet capacity Starlink's cash can't fund alone.

7. Future Outlook — S-1 claims a $28.5T TAM ($26.5T AI, $1.6T Connectivity, $370B Space) = ~27% of global GDP, 93% of it AI; "an AI company that happens to own rockets." xAI revenue ($3.2B 2025) is a fraction of OpenAI/Anthropic; Grok's perception lags. Catalysts: Starship operational debut (H2 2026, first payload after 12 test flights); V3 deployment cadence (one Starship carries 60 V3 satellites at 1 Tbps each, ~20x a Falcon 9); orbital compute (100 GW would need thousands of launches/yr + ~1M metric tons/yr vs ~7,400 tons cumulative ever); AI unit economics (Anthropic deal covers part of burn); the IPO price itself (>$2T = buying moonshots; <$1.5T = paying for Starlink+launch with AI as a free option).

8. Personal Take — "IPO = It's Probably Overpriced." SpaceX ~$350B late 2024 → ~$1T at the xAI merger (xAI $250B) → IPO pitched $1.5–2T (midpoint ~$1.75T) ≈ 90x 2025 revenue / 265x adj EBITDA. Sum-of-the-parts: Connectivity $500–700B (70–100x EBITDA); Space $100–200B (10–20x fwd rev); AI $200–500B placeholder (Anthropic crossed $30B annualized in Apr 2026 ahead of OpenAI ~$24–25B; ChatGPT ~50M paying subs vs SuperGrok 1.9M); Moonshots ~$200–500B optionality. Range $1.0–1.9T. "Musk premium": Tesla gets value for Optimus pre-revenue, xAI valued $250B pre-merger, SpaceX gets credit for orbital data centers/Starship/Mars before they exist — while Amazon Leo doesn't move AMZN and Google's quantum barely registers in GOOG ("capitalized for one CEO and expensed for everyone else"). Verdict: belongs on every serious investor's watchlist; "I'd rather watch this one trade for a few quarters than chase it on day one." Disclosure: author owns AMZN, GOOG, NVDA, TSLA.