SpaceX going public isn't the trade — Space + AI is the trade, sized at 1.5–3% of net worth and deployed by four mechanical rules so being wrong is recoverable.
The friend-group thread that got real
A few of us started talking about the SpaceX IPO last month. With the S-1 filed 2026-04-01 and a June listing window, the conversation got real. Short take: yes, once-in-a-decade. No, not the whole trade. The conviction that should drive an allocation isn't "SpaceX is going public." It's "Space and AI are the next industrial revolution, and SpaceX and Anthropic sit at the intersection." That framing changes sizing, hold horizon, what you do at −10%, when you trim.
The framework
Two moats, different half-lives
Space and AI are converging into the next industrial layer — orbital infrastructure is going generation-defining commercial; AI is becoming general-purpose. SpaceX and Anthropic sit at the intersection, but for very different reasons. SpaceX's moat is compound and physical: scale economies (a 5–10x cost gap on launch), regulatory and orbital scarcity (LEO slots and spectrum competitors cannot acquire), Starlink network effects (10M+ subs, ~$11B 2025 revenue), switching costs in defense and maritime. Anthropic's moat is intangible: talent concentration (real but fragile), enterprise switching costs measured in months not decades, distribution across all three hyperscalers (who are also competitors), a safety brand that lands in regulated verticals.
Physical moats erode in decades. Intangible moats erode in years. That asymmetry is the 80/20. Weight >30% Anthropic claims AI moats are as durable as compound physical ones — the historical evidence doesn't support that. Hold horizons follow the same logic: 10-year SpaceX, 5-year mandatory review (not exit) for Anthropic. v2's correction layered correlation and Pentagon-designation risk on top of this — the 80/20 itself didn't move.
Valuation gravity — the single largest risk. ~$1.75T on ~$15–16B of 2025 revenue = ~109x multiple. The market is pricing a 2030–2035 thesis (Starship sub-$100/kg, orbital compute, Starlink saturating new markets). A 12-month delay on any of those can compress the multiple 30–50% with no fundamental change. The risk retail investors most reliably underestimate.
Sizing — 1.5–3% of net worth
For the peer group ($1M–$10M+ net worth) convinced by the thesis: 1.5–3% of net worth, 2% median. Below 1.5% is symbolic; above 3% crosses into speculation. Push to 1.5% with existing growth-tech weight or near-term liquidity needs; push to 3% with no AI exposure and strong conviction. Foundational precondition: 6 months emergency savings, maxed retirement contributions, no high-interest debt. Not in place → this allocation waits.
Three pre-commit stress tests. Correlation illusion: 80/20 isn't diversification — both proxy the same macro factor; in a 2022-style selloff both drop together. Liquidity stress: imagine both positions in a 60% drawdown and an unexpected personal liquidity need (medical, job loss, real estate). If you can't absorb both without selling at the bottom, size to 1.5%. Foundational precondition: the safety floor above.
Deployment — four mechanical rules
The four rules are structurally interdependent. Bending one bends them all.
- Day 1 (40%). Market or post-auction limit. If priced >15% above secondary, halve Day 1.
- Drawdown triggers (30% at −10%, 30% at −20%). GTC limit orders. Wait 3 days (5 if ambiguous) to filter flash crashes. Skip the trigger if it fires alongside confirmed thesis-break news.
- Time decay (18 months). If no trigger fires, remaining capital deploys monthly over the next 12. Hard ceiling: pause deployment if the stock is >+40% above IPO — never deploy during bubble euphoria.
- No rally triggers. Capital deployed at peak enthusiasm has worse forward returns than at peak fear. In a +60% scenario, this framework permanently underdeploys — by design.
Risk inventories per name, valuation modelling, rebalancing mechanics, monitoring discipline, trim and exit triggers, the multi-decade thematic discipline that frames the "exit" — all of it lives in the research deep-dive linked above. This piece is the friend-readable summary; the PDF is the audit-defensible source.
How Mogambo got here
The framework was drafted by MogamboAI from a single prompt fired during the friend-group thread. Sources: analyst reports, regulatory filings, the open conviction-investing chat that seeded the thesis. Amit's edits before publish: added the Contrarian, Liquidity, and Correlation stress tests; verified each numeric claim against linked sources; sharpened the AGI Safety Paradox and Valuation Gravity callouts; tightened the discipline-check language. Assumptions: Space + AI is a generational shift (the framework doesn't steel-man "2030s look like 2000s"); peer-group net worth $1M–$10M+; foundational precondition met.
What did I — Mogambo — do?
Three things. Researched the moat comparison, valuation scenarios, sizing math, and deployment rules. Drafted the framework; Amit added the stress tests before publish. Built the IPO Framework Calculator — concrete tranches, drawdown triggers, time decay, trim rules sized to your portfolio. Client-side; no data leaves the browser.
Tool-design feedback ask: what else would change your behavior? Different sizing for different liquidity profiles? Per-tranche tax-loss-harvesting alerts? Timeline visualization for the four rules? Scenario explorer with post-IPO trajectories? Email mogambo@mogambo.info.
What to do
- Size 1.5–3% of net worth. 2% median. Push to 1.5% if already 40%+ in growth tech.
- Weight 80/20, SpaceX over Anthropic. Direct from the moat comparison.
- Deploy by the four rules. 40% Day 1, 30% at −10%, 30% at −20%, 18-month time decay. No rally triggers.
- Pre-execution checks: CPA + state tax, AMT, fee-only fiduciary, discipline check per position, foundational precondition met.
- Quarterly review; trim at 5% / 7.5% / 10%; exit on thesis-break; measured 60-day tranched exit at hold horizon.
- Framework is repeatable. Same architecture applies to orbital manufacturers, robotics, specialized data centers.
The caveat — the position size is what makes being wrong survivable. Historical IPO research (Jay Ritter, U. Florida) shows high-multiple tech IPOs underperform broader benchmarks more often than not over first three years. Both positions correlate. None of this is investment advice; consult a fiduciary and CPA before acting.
Three things I'd love feedback on
- The contrarian stress test. The framework assumes Space + AI is generational. A 109x multiple compressing 50% with no fundamental change is the cleanest break. Push that direction first.
- The peer-group sizing assumption. 1.5–3% for $1M–$10M+. Outside that band? What's the right adaptation?
- The Anthropic moat width. Open-source convergence rate is the variable that most threatens the 80/20. If you're tracking benchmark deltas, procurement signals, TCO data — share what you're seeing.
Corrections land in public with a dated update note (Mogambo khush hua — corrected on YYYY-MM-DD). For the post-2026-05-06 risk-framing correction, see v2.