SpaceX (SPCX)

Executive Summary & Action Plan

Verdict

SELL (cut from HOLD, 7/25) — the first print delivered the P&L and detonated the cash-flow line: a +14.5% revenue beat with a 45% EBITDA margin came bundled with $18.4B of quarterly capex guided flat for two more quarters. With a real 10-Q finally on file, we retire the S-1-era SOTP ($124) and re-mark fair value to $45.60 on three methods that now cluster within $5.50 — spot sits 146% above the mark even after the −10.5% post-print break.

Price $112.14 | Market cap $1.48T | Target $45.60 | Upside −59.3% IWANNAVY Fair Value (log-median of 3 methods, re-underwritten on the Q2'26 10-Q) · price as of 2026-08-05 intraday, cross-checked same-day · fully-diluted 13.746B shares · Street $223 (n=35, range $62–$800, reference) — street targets extrapolate the $100B run-rate guide toward the $1T path; we credit only signed-contract evidence (~17% of the 2030 target)

Abstract

SpaceX's first public quarter was a clean beat and a cash-flow shock in one release: revenue of $7.81B (+92% YoY) topped consensus by 14.5%, adjusted EBITDA of $3.54B nearly tripled QoQ to a 45.3% margin, and the AI segment turned EBITDA-positive on $14.1B of newly contracted cloud sales — yet capex of $18.4B ran 39% above expectations and management guided two more quarters at that pace, ~$65B annualized against a $100B post-IPO war chest. The 10-Q supplies what the July SOTP lacked — real statements, segment EBITDA, and a capex plan — so the mark drops from $124 to a three-method cluster: DCF $42.6, EV/EBITDA $48.1, EV/FCF $45.6, log-median $45.60. A reverse DCF shows $112 still prices ~$335B of 2030 revenue, twice the consensus path. Rating cut to SELL; even the bull case ($97.9) sits below spot. Principal risks to the call: squeeze convexity on a ~26% short float and a faster-than-modeled cloud ramp toward the December $100B run-rate checkpoint.

Forecast Path

The path prices the unlock calendar against the fair-value gap. Weights are judgmental and non-default (Bear 25 / Base 55 / Bull 20): the print resolved the first-report information vacuum (base up from 48), the capex guide and ~$102B of August supply keep the left tail fat, and short-float convexity plus the October cloud-contract ramp keep the bull tail real. Note the house asymmetry — even the bull path converges to $97.9, below today's tape.

Action Plan

Exit into any bounce; the unlock calendar does the compression work from here — no re-entry until deep fair-value territory, and the sell call itself stops out on a sustained re-rate above $150.

• Risk/Reward on the exit call 1.8:1 (−59% to base FV avoided vs +34% to the $150 invalidation) | Prob-weighted 12M return −57.3% | Confidence: Medium — three methods converge within $5.50, but two public quarters and a 26% short float make timing treacherous

---

Last Four Quarters

Four recast quarters are now public (Q3–Q4'25 standalone splits were never filed; the xAI merger recast all periods). The pattern is a barbell: gross margin and EBITDA inflected sharply up in Q2'26 as compute revenue landed on already-depreciating infrastructure, while the FCF line went the other way — capex nearly doubled QoQ and the FCF margin reached −204%.

---

What Happened — The First Print (8/4) and the −10.5% Answer (8/5)

The quarter beat on every P&L line: revenue $7.81B vs $6.7–6.9B consensus (+13–16%, source-dependent), EPS −$0.09 vs −$0.26 street (~+65%; a −$0.23 aggregator variant implies +61%), adjusted EBITDA $3.54B vs ~$2.0–2.1B expected. Operating loss narrowed 93% QoQ to −$143M — within sight of GAAP breakeven. The sell-off decomposes into two verified drivers: the capex shock — $18.4B vs ~$13.2B expected, $15.8B of it AI compute, with CFO Bret Johnsen guiding the next two quarters "similar" and defending the spend as "almost like a COGS item" with sub-1-year payback — and the supply clock: the first unconditional unlock tranche (~911.5M shares, ~$102B at spot) becomes sellable August 6, and was front-run a day ahead. The stock fell as much as ~8–9% after hours and closed −10.5% at $112.14.

The call itself was peak-Musk: the $1T revenue target pulled forward to 2030 ("non-zero chance" of 2029), $100B annualized run-rate promised by December, $6.7B of new cloud contracts signed in the first weeks of Q3, a Starship tower-catch attempt targeted for late August (mgmt) – September (E, trackers), and the reveal that Starlink's terrestrial wireless buildout will run on the EchoStar 65 MHz spectrum via femtocells co-sited with Starlink dishes — aimed at the ~$600B/yr US carrier market. Executives repeatedly hedged what Musk promised; the CFO's numbers, not the vision, moved the tape.

The QoQ/YoY bridge shows where the quarter's quality actually sits — a $2.4B QoQ swing in EBITDA-relevant lines with the net loss shrinking 87% QoQ:

Revenue & Profit Mix

Connectivity is the profit engine, AI is now the growth engine and nearly all of the capex, Space is the strategic cost curve. The mix shifted 11 points toward AI in a single quarter.

• Connectivity (55% of revenue, 60.5% segment EBITDA margin): 12.0M subs (2x YoY, +1.7M QoQ), ARPU stabilized at $66 flat QoQ — the July thesis-breaker condition that did trigger. Enterprise & government +108% YoY on airline wins and $6B+ Starshield awards. Must know — this segment's $2.6B quarterly EBITDA is what funds everything else.
• AI (33% of revenue, first positive EBITDA +$1.15B): compute revenue $2.19B vs $0.48B in Q1 as $14.1B of Cloud Services Agreements (reported to include Google and Anthropic) began converting; 1.4 GW nameplate live vs a 15 GW end-2027 target. Must know — $15.8B of Q2's capex sits here; the entire rating hinges on whether this spend clears its claimed sub-1-year payback.
• Space (12% of revenue): 38 launches, 485t to orbit in Q2; Starship V3 flights 12–13 succeeded; Falcon is being wound down in favor of a vehicle that must now work on schedule. Must know — Starship cadence gates Starlink V3 capacity, the wireless timeline, and the moon narrative all at once.

---

IWANNAVY Fair Value

The July mark ($124) was a sleeve-level SOTP built on S-1 disclosure with Starship and xAI option-priced, held deliberately unchanged through the pre-print washout. The 10-Q replaces option-pricing with evidence, and the evidence prices lower: a consolidated 10-yr DCF crediting 17% of the $1T-by-2030 target (WACC 10.5%, g 3%) gives $42.6; 22x FY27E EBITDA of $27.3B — a premium to every verified infrastructure peer — gives $48.1; 28x normalized FY31E FCF discounted back gives $45.6. Log-median $45.60. The revision decomposes into the capex guide (~$65B FY26E vs the ~$36B annualized burn underwritten in July), the retirement of sleeve WACCs (8–9.5% → 10.5% consolidated), and consensus-anchored revenue replacing option value. The three methods now agree within $5.50 where July's spread was $46–124 — a tighter, humbler, and much lower mark.

• Thesis breaker (for the SELL): a Starship tower catch plus Q3 evidence the $100B December run-rate is real (≥$80B exit pace) with capex held — not raised — re-opens the bull path ($97.9) and forces a cover above the $150 invalidation line.

---

Catalysts & Risks

---

References

• SpaceX Q2'26 earnings release (8-K EX-99.1, SEC EDGAR, 2026-08-04) · Form 10-Q, period ended 2026-06-30 (SEC EDGAR, 2026-08-04) · 424(b)(4) IPO prospectus (2026-06-12)
• CNBC — SpaceX dives 10% after AI spending surge (2026-08-05) · Fortune — capex tanks SpaceX on debut earnings (2026-08-04) · TechCrunch — Musk one-upped his execs on the call (2026-08-04)
• Fierce Network — the femtocell plan vs the Big 3 (2026-08-05) · Investing.com — the August unlock (2026-08-02 (E)) · Seeking Alpha — spending surge overshadows the beat (2026-08-05)
• Companion house reports: SpaceX pre-print refresh (2026-07-25) · SpaceX full workup (2026-07-12)