Built by Claude Fable

SpaceX: 4.45/10

Deep dive pages: Overview | Business Model | Financial Trends | Thematic Exposure | Management Quality | Investor Sentiment | Concerns, Catalysts & Risks

Prepared 2026-10-29 and revised after the PM Review the same day. Market cap ($1,933.5B, FMP /stable quote x 13,581M shares from the 10-Q cover) and EV (about $1,921B, FMP net cash bridge) are used only as inputs to the forward multiples. Fundamentals from Daloopa (company_id 247912), cited inline. Fiscal year = calendar year. Last print Q3 2026 (2026-10-20); next print about 2027-02-25. SpaceX absorbed xAI (with X and SpaceXAI Corp.) on 2026-02-02 in a common control merger, so history is recast; Cursor closed mid-Q3 2026.

Composite 4.45/10 (was 4.85 before the PM Review). Gate flag: Below Quality Bar (Requires Exceptional Catalyst). Recommendation: AVOID. Revisit after Starship Flight 15 (Nov 2026) and the Q4 2026 print (about 2027-02-25).

Bottom line: SpaceX is #1 in the businesses that are shrinking as a share of revenue and about #6 in the business that is becoming the company. It flies about half of the world's orbital launches and more than 80% of mass to orbit, and Starlink has about two-thirds of all active satellites. But AI was 52% of Q3 2026 revenue and is roughly 80% of the Q1 2027 guide ($33B to $37B), and in AI compute SpaceX added only about 7% of new global AI IT capacity in 2026, renting it at spot prices to a customer that was 34.5% of revenue. The income statement is inflecting fast: revenue +161% YoY, Adjusted EBITDA +306% YoY to a 60% margin, and AI segment EBITDA swung from -$428M to +$4,953M. The cash flow statement is going the other way: TTM FCF is -$39B and Q4 capex is guided to $35B to $40B. Fundamentals are going from worse to better, but sentiment already got there (about 80% Buy, record retail allocation). On observed Q3 economics (about $16.5 of AI revenue per nameplate watt per year) FY2027 consensus of $151B looks roughly fair, and at the 13.5x peer EV/EBITDA the EV needs about $142B of FY2027 EBITDA, 2.4x consensus.

Brooker asked for the oligopoly framed two ways: (A) SpaceX's share of all space launches, any payload, and (B) its share of new AI compute measured by gigawatts added. Lens A is a clear oligopoly. Lens B is not, on anything energized so far.

Lens A: share of all orbital launches

• Count: about half of every orbital launch on Earth since 2024; with China's CASC (about 21%), two players hold about 70%.
• Mass, the economically relevant measure: more than 80% of world mass to orbit every year since 2023 (S-1, p.17 and p.127); Musk put it at almost 90% on the Q3 call, with China about 6%.
• Ex-Starlink: only about 20% to 26% of world launches, because much of the count is captive Starlink demand. Even so, SpaceX flies about half of the open (non-China, non-Russia) market.
• Trend: Falcon cadence fell in 2026 (113 launches in 9M vs 128 a year earlier) as internal Starlink flights shift to Starship V3, so count share dipped to about 49%. Mass share is unchanged at about 90%.

Lens B: share of new AI compute by GW added

Brooker's simple framing divides SpaceX's stated GW by a headline GW number, which mixes bases but is the right intuition:

• Energized reality: about 3% of 2025 and about 7% of 2026 global AI adds. 2026 builders on the house file (mostly facility basis): AWS about 6 GW, Microsoft about 4.3, Google about 3.5, Oracle about 2.9, Meta about 2.5, SpaceX/xAI about 2.0. SpaceX is about #6, and the top 3 hold only about 53%.
• 2027 target: +10 GW would make SpaceX a top-2 builder at about 22% to 25% of adds. That is a big share of new AI, but in a field of five to seven large builders it is not an oligopoly. The target rose about 50% in one quarter and needs roughly 37% to 50% of NVIDIA's 2027 output, so a chip-feasible 4 to 5 GW IT (about 12% to 16%) is more realistic.
• Basis: SpaceX "nameplate" includes facility overhead and margin (about 1.4x IT: 300 MW per 110k GB300). Denominators are the house view in research/ai-economics/debate/gw_added.md. Full detail on Thematic Exposure.

Gate flag: Below Quality Bar (Requires Exceptional Catalyst). Oligopoly YES, positive and growing FCF NO, management track record NO. Two NO answers cap the composite at 5.5; the raw 4.45 is already below that, so the cap does not bind.

What would change the score:

• Up (toward 5.5): quarterly FCF turns positive, which flips the FCF gate and lifts Financial Trends; the Q4 2026 and Q1 2027 prints confirm the $100B December ARR and the $33B to $37B guide with realized yield holding near the Q1-implied $35 per nameplate watt; Flight 15 catches both stages.
• Down (toward 4): compute spot pricing normalizes or Customer B shifts capacity, and AI revenue reprices within quarters; a Starship mishap brings an FAA grounding; a Tesla combination dilutes holders through the controlled-company process.

Cross-dimension reconciliation notes

• Per-watt units (the main correction). Management's $30 to $50 per watt is an annual revenue rate on rented IT watts on the newest contracts; it reconciles with the sub-1-year payback claim and the filed Google contract (about $51 per IT watt per year). The Q3 blended yield on all nameplate was $16.5 per watt per year ($7,022M x 4 / 1.7 GW), because about half of compute is used internally, nameplate is about 1.4x IT, and capacity bills only after it is energized. The first draft's "$4.1B per GW annualized" was per quarter.
• FY2027 consensus framing. Dims 4 and 5 now agree: on observed economics the $151B annual mean is roughly fair (bridge $146B to $167B); the premium sits in the multiple.
• Buybacks. Daloopa shows $4,346M in Q1 2026, $80M in Q2 and $0 in Q3, all pre-IPO tender activity.
• Q1 2026 net income. -$4,947M is attributable to common; -$4,276M is total net loss; the gap is a $671M deemed dividend.
• Q3 SBC. $2,363M total includes $1,217M of Cursor acquisition-related SBC; $1,146M excludes it.
• Data gaps. FMP /stable has no estimate-revision history, so FY2027 consensus before and after the Q3 call is not available; Bloomberg, Visible Alpha and SharePoint were not connected. Daloopa history starts in 2023 (annual) and Q1 2025 (quarterly), so only three YoY-comparable quarters exist.

The PM Review agreed with the AVOID call and asked for fixes before publishing: remove price talk, replace QoQ with YoY, add YoY rows and segment profitability, fix the per-watt unit error, re-score Thematic Exposure on the forward mix, add the Q2 capex miss to the promise audit, and complete the valuation framework with P/FCF and an EBITDA hurdle. All were applied except the items that need Bloomberg, Visible Alpha or SharePoint.

Data sourced from Daloopa.