Initiating Coverage — SpaceX (Nasdaq: SPCX)

Space Exploration Technologies Corp. · Coverage desk · 2026-07-15

Rating: Underweight · 12-month price target: $75–$115 (base case $94) · Last close: $136.12 (2026-07-14, Nasdaq)

Implied downside to the base-case target of ~31%; working range implies 15%–45% downside from the last close.

This is a Tier 1 initiation, re-tiered from Tier 2 (private) on the 2026-06-12 listing date. Actuals are anchored to SpaceX's S-1 (filed 2026-05-20, amended 2026-06-01/03) and 424B4 (2026-06-12); no 10-K or 10-Q has yet been filed (first 10-Q, for the quarter ended 2026-06-30, expected ~August 2026). See the Appendix for sourcing status and open disclosure gaps.

---

Executive Summary — The best company we have ever been unable to recommend

SpaceX is, by a wide margin, the most operationally dominant company in the aerospace complex, and possibly the most dominant private-turned-public franchise to list in a generation. We initiate anyway at Underweight, because the call on SPCX is not a call on the business — it is a call on the multiple. The stock came public at a valuation that already discounts a second monopoly (Starlink) and a third, unproven one (AI/orbital compute) on top of the launch monopoly that is actually earning money today. Our five-point thesis:

1. The stock is priced for a future that has to go nearly perfectly, in the one segment with the least evidence. At $136.12 (2026-07-14), SPCX carries an implied market cap near $1.77T on FY2025 revenue of $18.67B (S-1, 2026-05-20) — roughly 95x trailing revenue, and still ~50x our base-case FY27E revenue and ~39x FY28E (model.md §6a). No public comparable is within an order of magnitude. The entire premium above a rich aerospace/telecom multiple is being paid for an AI segment that lost $6.355B on $3.2B of revenue in FY2025 — a loss larger than the whole company's consolidated net loss of $4.9B (S-1 via filings.md §2a).

2. The core is genuinely excellent, and we model it that way. The launch franchise (165 orbital missions in 2025, >60% of global launches by count, >80% of mass to orbit — BryceTech via Via Satellite, 2026-04-10) funds everything else through a reusability cost advantage no competitor is within a decade of matching. Starlink is the only profitable segment — $11.39B FY2025 revenue, +48% YoY, $4.4B operating profit (S-1, 2026-05-20). Under every scenario we run, launch and Starlink revenue grow through FY30E and neither turns loss-making at the segment level. The bear case is not that these businesses are weak; it is that they cannot, by themselves, justify the price.

3. The AI segment is where the thesis is decided, and it is genuinely unresolved. Our base/bear/bull spread on FY30E AI revenue alone — $8.0B / $17.5B / $35.0B (model.md §5) — is wider than the spread on Space and Connectivity combined. The bull's fact (AI is already at a ~$3.3B annualized run rate in its first disclosed year) and the bear's fact (it is burning more than the entire company lost) are both true. On one disclosed year of segment data, we are not willing to underwrite the AI outcome the current price requires at even 50/50 odds.

4. Our target-multiple bridge puts fair value below the market. Applying scenario-appropriate EV/Revenue multiples (20x / 35x / 55x) to FY27E revenue yields implied per-share values of $43 / $94 / $177 (model.md §6b). We set a 12-month target range of $75–$115, bracketing the $94 base case and widened for two-sided AI uncertainty; we treat the $43 and $177 model tails as scenarios, not the working range. The current price sits 31% above our base case and requires results closer to the bull scenario just to be fair today.

5. This is a valuation call, not a fundamentals call — and it is falsifiable. Base-case FY27E revenue of $34.7B is +86% above FY2025 actual; we are not forecasting a business in trouble. We are saying the market has already paid for the good outcome. The single disclosure that would move us most is the first 10-Q (~August 2026): AI-segment revenue accelerating past a ~$5B annualized run rate with narrowing losses would pull us toward the bull case and compress our discount; a flat or decelerating print reinforces Underweight.

We would be buyers of this business at a price. This is not that price.

---

Business overview — three segments, one cost curve

SpaceX reports three segments, a structure that dates only to the all-stock merger with xAI Corp. that closed 2026-02-02/03 at an implied combined value of $1.25T (SpaceX $1T + xAI $250B — CNBC, 2026-02-03). That merger is the reason an AI line exists in the P&L at all; before it, SpaceX was a launch-and-broadband company.

Connectivity (Starlink) — the profit engine. 69% of FY2025 revenue at $11.39B and the only segment in the black at +$4.4B operating profit (S-1, 2026-05-20). It makes money three ways: recurring consumer/enterprise subscriptions (12M+ subscribers across ~160 countries as of 2026-06-04, Yahoo Finance), one-time hardware/kit sales at each gross add, and a growing wholesale/enterprise/Starshield book (maritime, aviation, direct-to-cell via T-Mobile, government broadband). Revenue is fundamentally subscribers × ARPU plus a hardware and enterprise uplift.

Space — the franchise that funds the others. ~22% of FY2025 revenue at $4.08B, loss-making because Starship R&D is absorbed into segment cost (S-1 via filings.md §2a). The critical structural point, which the model is built around: the 165-flight Falcon cadence is dominated by SpaceX's own Starlink deployment missions, which carry no external revenue. Only externally contracted launches — commercial and government — are revenue-bearing, which is why segment revenue is a fraction of what cadence × list price would imply. Space makes money on external launch (Falcon 9 dedicated ~$74M late-Feb 2026), NASA and Space Force programs, and crew/cargo (Dragon).

AI (SpaceXAI / xAI / Grok / X) — the swing factor. ~9% of FY2025 revenue at ~$3.2B, and a –$6.355B operating loss — the dominant drag on consolidated profitability (S-1 via filings.md §2a). It monetizes through Grok subscriptions and API, and X advertising off a depressed base. In Q1 2026 the segment did $818M of revenue against a –$2.469B operating loss (S-1/FWP via filings.md §2a) — roughly a $3.3B annualized run rate, but with unit economics that are entirely unproven against hyperscaler incumbents.

The through-line: cheap, high-cadence reusable launch is the cost structure that makes the Starlink build-out economic, that makes the government satellite-prime pivot possible, and that — the bull argues — makes megawatt-class orbital compute conceivable. Every segment sits downstream of the launch cost curve.

---

The moat — reusability, vertical integration, scale

SpaceX's advantage is not a single technology; it is a compounding cost structure that no competitor can replicate on the current timeline.

Reusability that keeps compounding. SpaceX flew 165 orbital missions in 2025, up from 134 (2024) and 96 (2023) — more than the rest of the world combined, and ~85% of U.S. launches (SpaceNews, accessed 2026-07-15). The reuse record keeps extending, not plateauing: booster B1067 flew its 36th mission on 2026-07-09 (AIAA), within three flights of Space Shuttle Discovery's all-time total, and the fleet approached its 600th cumulative reflight days later (TechTimes, 2026-07-14). Each additional reflight lowers marginal launch cost further. This is the balance sheet that funds every other segment: Starlink's economics do not work without near-free internal launch.

Vertical integration. SpaceX builds its own engines, airframes, avionics, satellites, and ground infrastructure, and operates its own constellation as the anchor customer for its own rockets. That internalizes the margin a conventional launch provider pays away to suppliers and to a fragmented customer base, and it lets the company deploy Starlink capacity at a cost no reseller of third-party launch could match.

Scale that competitors are subsidizing. SpaceX delivered >80% of global mass to orbit (2,213 tonnes) in 2025 (BryceTech via Via Satellite, 2026-04-10). The tell on competitive distance is Amazon: its Leo (Kuiper) constellation is buying launches from SpaceX itself to hit its FCC buildout deadlines (CNBC, 2026-01-30). When your nearest full-stack rival is a customer, the moat is real.

The honest caveat: the next leg of the cost curve — sub-$500/kg — depends on Starship, which is not yet operational. The moat that exists today is Falcon-9-based and fully proven. The moat the valuation is partly paying for is Starship-based and is not.

---

Starlink deep-dive — the best business in the company, growing lower-quality revenue

Starlink is the clearest, largest, most durable growth engine SpaceX owns, and it is also the segment where bull and bear read the same facts in opposite directions.

The growth is real and the profitability is real. Subscribers crossed 12 million across 160+ countries by 2026-06-04, roughly doubling year-over-year, on a trajectory from 1M (Dec-2022) to 10M (Feb-2026) to 12M (Jun-2026) (Yahoo Finance, 2026-06-04; Motley Fool, 2026-07-02). The constellation stood at 10,799 satellites in orbit, 10,783 operational as of 2026-07-14 (KeepTrack). Management guides Starlink Mobile toward 25 million active users by end-2026 (Basenor, 2026). And it is profitable while doing this: $4.4B of operating profit on $11.39B of revenue (+48% YoY) in FY2025 (S-1, 2026-05-20).

The quality of that growth is deteriorating, by design. Blended ARPU fell from $99/month in 2023 to $66/month in Q1 2026 — down a third — as the mix shifts into lower-priced African, Southeast Asian and Latin American markets, a volume-over-price strategy that S-1 language guides to continue (The Information via Roic.ai, 2026-04-29). The bull reads this as proof the unit economics work even under deliberate price compression against a >3.1B-person addressable footprint. The bear reads it as an increasingly low-yield growth algorithm: the highest-ARPU markets (rural/suburban North America) are closer to saturated, so incremental subscribers come from structurally lower-yield geographies — the same geographies Chinese state constellations (Guowang, Qianfan) are building toward.

Our model. We take average subscribers from ~7.05M (FY25E) to ~15.75M (FY26E) — haircutting management's 25M year-end target to ~22M — and continue the ARPU decline at a decelerating rate to a ~$52/mo floor by FY30E as enterprise/direct-to-cell mix offsets consumer dilution (model.md §1b). That drives Connectivity revenue from $11.39B (FY25) to $16.5B (FY26E) to $36.5B (FY30E), decelerating from +45% to +16% YoY. Crucially, this is not a bear assumption — we bake in the ARPU decline the bear worries about and the segment still nearly quadruples. The risk we watch is not the level of ARPU but its trajectory: if Q2/Q3 2026 prints show the decline reaccelerating rather than finding a floor, our base-case Connectivity margin comes under pressure.

---

Starship — the bet inside the company

Starship is the load-bearing asset for the long-run thesis and the single largest technical swing factor for the entire name. It is also the asset whose flight record most clearly does not yet support the premium being paid for it.

The record, flight by flight. Across six integrated flight tests from January 2025 to May 2026: IFT-7 (2025-01-16, ship lost), IFT-8 (2025-03-06, ship lost), IFT-9 (2025-05-27, both stages lost), IFT-10 (2025-08-26, success), IFT-11 (2025-10-13, success), IFT-12 (2026-05-22, ship succeeded, booster B19 lost) — two clean successes, three failures, one partial across six flights (Wikipedia/Spaceflight Now, accessed 2026-07-15). That is a 50% failure-or-partial rate across the two most recent test years, against a cadence Musk himself floated at up to 25 Starship launches in 2025 — five actually flew. IFT-12's booster loss triggered the second FAA-mandated mishap investigation and grounding inside eighteen months, delaying Flight 13 to 2026-07-16 (Spaceflight Now, SpaceNews, 2026-05-27).

Why it matters so much. Every element of the long-run story is downstream of Starship: the sub-$500/kg cost target, the Starlink V3 satellite generation (which requires Starship to deploy), the NASA-funded lunar program, and Mars. The S-1 discloses cumulative Starship program spend exceeding $15B (Reuters via BNN Bloomberg, 2026-05-01), with FY2025 Space-segment R&D of $3.0B going "the entirety" to Starship (model.md §2) — capital committed at a scale that presumes the vehicle graduates from experimental to operational on a timeline the flight record does not yet support.

The bull's rebuttal, fairly stated. The trend line inside the record points toward success: three consecutive failures were followed by two clean flights and then a flight where the harder, higher-value half of the stack (the ship) succeeded. Every iteration has narrowed the failure mode. A vehicle that achieves even a fraction of its designed reusability collapses launch cost again, exactly as Falcon 9 did — and SpaceX has already proven the reusability thesis once. That is genuine asymmetric optionality.

How we model it. As optionality, not a base-case dependency. Our base case grows external launch revenue only modestly, with Starship-dedicated commercial missions layering in from FY28E — so we do not require Starship success to hit our numbers, and Starship risk sits in the swing factors rather than as a markdown of the core franchises. Flight 13 (2026-07-16) is the next data point; three consecutive clean flights would be the evidence to promote Starship from optionality to a base-case driver, and would remove one argument for the discount we currently apply.

---

Government & defense — becoming a satellite-system prime

The government book is shifting SpaceX's role from launch vendor to satellite-system prime — a higher-margin, stickier mix — and it is a demand pool where competitive contestability is unusually low.

The contract stack. NASA anchors the base: HLS ~$4.04B through 2028, Commercial Crew (CCtCap) $4.927B, CRS-1+CRS-2 $4.8B combined (NASA.gov, cross-checked). On launch, SpaceX holds the largest NSSL Phase 3 Lane 2 allocation by dollar value and mission count — ~$5.9B / 28 missions vs. ULA's ~$5.4B/19 and Blue Origin's ~$2.4B — and FY2026 task orders skewed 5-of-7 missions to SpaceX (Spaceflight Now, 2025-04-05). The newest and most striking leg is missile defense: the Golden Dome architecture generated ~$6.45B of SpaceX awards inside a single four-day span in May 2026 (Space Data Network Backbone $2.29B + SB-AMTI $4.16B — Bloomberg, 2026-05-29), reportedly exceeding the combined prototype awards to every other company in the program. Company-cited total federal backlog is ~$22B (directional; individual award lines do not fully reconcile to it without a USASpending.gov pull).

Why it is defensible. Blue Origin's New Glenn is uncertified for national-security missions and just lost its only pad in a May 28, 2026 static-fire explosion (Spaceflight Now, 2026-06-03); ULA cannot reuse hardware and cannot match Falcon 9 on price. The U.S. government needs resilient, mass-produced satellite architecture at cadence, and there is one domestic vendor that can build and launch it. That underlying demand does not depend on any single administration's mood.

The offsetting risk. ~20% of FY2025 revenue came from U.S. federal agencies (24/7 Wall St., 2026-05-21) — a concentration that is durable in the aggregate but exposed to political-relationship and budget-cycle shocks (see Risks). This is the leg of the business where the moat is strongest and the key-man risk is most acute at the same time.

---

Market & competition

Launch. SpaceX's near-term position faces no real competitive pressure. Both would-be Western heavy-lift rivals stumbled in 2025–2026: Blue Origin's New Glenn lost its pad in May 2026 (return-to-flight targeted YE2026); Rocket Lab's Neutron has not flown (Q4 2026 debut targeted after a redesign). ULA's Vulcan is executing (~18–22 launches targeted 2026) but is expendable and cannot match Falcon 9 on price. The one structural threat is China, which ran 92 launches in 2025 and targets ~140 in 2026 — the only cadence approaching SpaceX's own, but walled off from SpaceX's Western commercial market (SpaceNews; Space.com, accessed 2026-07-15).

Satellite broadband. Starlink holds a multi-year lead that is widening on a satellites-in-orbit basis. Amazon Leo (Kuiper) had 365+ satellites against a July 2026 FCC buildout deadline it could not meet, and is buying SpaceX launches to keep pace. Eutelsat OneWeb (~600 satellites) is enterprise-only and effectively concedes consumer. Chinese state constellations (Guowang, Qianfan; >350 combined) are geopolitically walled off but the only constellations with Starlink-scale ambition, and they contest the same low-ARPU emerging markets Starlink now depends on for growth. Direct-to-cell is the one sub-vertical genuinely contested: AST SpaceMobile (7 satellites, targeting 45–60 by YE2026; carrier-backed) is public markets' preferred direct-to-device proxy, versus SpaceX/T-Mobile's live-but-SMS-first service. Both are early-innings.

Government/defense. Covered above — SpaceX is the structural incumbent with limited near-term contestability.

The summary. SpaceX is the dominant player across launch, broadband, and national-security space simultaneously — a combination no competitor approaches. The nearest thing to a multi-vertical rival, Amazon, is years behind on satellites-in-orbit and partly dependent on SpaceX's own launch capacity to close the gap. The relevant public yardsticks — Rocket Lab ($50.6B), AST SpaceMobile ($26.2B), Planet Labs (~$9.6B), ~$86B combined (MacroTrends/companiesmarketcap.com, accessed 2026-07-15) — capitalize the entire investable commercial-space complex ex-SpaceX at under 5% of SPCX's market cap.

---

Financial outlook

We build revenue bottoms-up per segment (external launches × price + programs; subscribers × ARPU + uplift; Grok/API + X ads), anchor FY23–FY25 actuals and FY25 adjusted EBITDA to the S-1, and calibrate FY26E-onward to the Q1 2026 print. FY26E-onward is entirely desk estimate. Full three statements, assumption stack, and tie-out checks are in model.md; the summary P&L follows.

Summary income statement ($M)

FY23A/FY24A/FY25A net income and FY25 adjusted EBITDA tie to S-1-disclosed actuals (–$4.6B / +$0.791B / –$4.9B net; $6.58B FY25 adj. EBITDA). FY2024 was the only GAAP-profitable year on record. The read-through for the forecast window: GAAP losses widen through FY27E-28E as AI-driven depreciation scales faster than EBITDA, then narrow toward EBIT breakeven by FY30E. It is the AI capex/depreciation cycle — not the core launch/Starlink business — that keeps the company GAAP-loss-making through most of the forecast even as adjusted EBITDA scales to $21.9B (34% margin) by FY30E.

Capital intensity is the other story. Consolidated capex quintupled to $20.7B in FY2025 (capex/revenue 111%), of which AI infrastructure was $12.7B — the one hard, sourced capex-by-segment figure (Yahoo Finance/AOL citing SpaceX disclosure) — more than Space ($3.0B E) and Connectivity ($5.0B E) combined per the desk's own model split (model.md §4; no clean public Space/Connectivity breakdown exists — see Appendix). Q1 2026 capex alone was ~$10.1B, ~$7.7B of it AI. Even after the $75B IPO raise, our base case shows sustained ~$29–33B/yr capex outrunning operating cash flow, forcing a return to debt issuance by FY28E–FY30E (to ~$23.0B). The self-funding question is live.

Scenario analysis (FY30E, $M)

Five swing assumptions drive the spread, ranked by materiality: (1) AI/xAI monetization — the dominant factor; (2) Starlink subscriber growth vs. ARPU; (3) AI capex intensity / self-funding; (4) Starship cadence/reusability; (5) government/defense execution.

The AI line does the work: its bear-to-bull spread ($8.0B → $35.0B) exceeds the spread on Space and Connectivity combined. That is by design, and it is the honest representation of where the uncertainty lives.

---

Valuation

Inputs. Shares outstanding ~13.0B, back-solved from $1.75T ÷ $135 IPO price — not a disclosed figure, flagged as pending confirmation against the 424B4 cover page (filings.md §2c). Last close $136.12 (2026-07-14); implied market cap ≈ $1,770B; estimated net cash post-IPO ≈ $50B, giving current EV ≈ $1,720B (model.md §6).

What the current price embeds. On base-case estimates, the stock trades at:

Even three years out, SPCX trades at ~39x forward revenue and ~140x forward EBITDA — a multiple no public comp approaches. Strip out the AI/xAI segment entirely and value only the profitable core (Space + Connectivity, $15.5B FY25 revenue) and the trailing multiple is still above 110x (research.md §6.2). The market is pricing SpaceX as an AI-infrastructure and orbital-compute platform layered on a dominant, profitable launch/broadband duopoly-of-one — not on conventional aerospace or telecom multiples.

The bridge to a price target. We apply scenario-appropriate EV/Revenue multiples to FY27E revenue (the nearest full fiscal year, 12–18 months forward), using each scenario's own FY27E revenue and net cash held at the FY27E base-case level ($11.9B) across scenarios for simplicity:

Even the base case assumes a compression from today's ~50x FY27E revenue multiple to 35x — still rich by any conventional standard, and not a bearish operating forecast: base-case FY27E revenue of $34.7B is +86% above FY2025 actual. The current price sits closer to the bull case in this framework, leaving little room for multiple expansion even if the AI segment executes in line with our base assumptions.

What the private mark embedded — and why it does not anchor the call. SpaceX's private valuation arc ran $350B (Dec-2024 tender) → $400B ($212/sh, Jul-2025) → $800B ($421/sh, Dec-2025) → ~$1.25T (Feb-2026 xAI merger) → $1.75T IPO (CNBC/TechCrunch/Bloomberg, dated inline) — a 5x move in eighteen months, coincident with the AI narrative entering the story. Those were non-dilutive secondary/tender events between existing shareholders, not primary capital, and they establish that the AI re-rating, not core-business growth, drove the last leg of the private mark. Now that there is a traded price, we anchor to it and express the call as downside to it — which is the discipline the private ladder never imposed.

Our target. We set a 12-month price-target range of $75–$115, bracketing the $94 base case and widened modestly for two-sided AI uncertainty. We treat the model's full $43 (bear) / $177 (bull) span as tail scenarios, not the working range: the bear requires AI monetization to essentially stall while Starlink/Starship also disappoint simultaneously; the bull requires the AI segment to ~4x our base-case FY30E revenue while sustaining today's rich multiple. Neither is the modal outcome. The working range implies 15%–45% downside from the last close.

---

Bear vs. bull

The bear, in one paragraph. This is a request to pay ~95–110x trailing revenue for a company whose only profitable segment faces deliberate margin dilution, whose most capital-intensive program (Starship) has a coin-flip flight-test record, whose founder has already once put a fifth of revenue in political jeopardy inside a single news cycle, and whose newest, most richly-valued segment is bleeding more money than the whole company lost last year. None of the underlying businesses need to fail for the stock to underperform — the multiple just needs to compress toward anything a public market has historically paid for infrastructure, telecom, or best-in-class AI compute.

The bull, in one paragraph. SpaceX is not one business at a stretched multiple — it is three call options (Starship-enabled cost curve, xAI/orbital-compute monetization, Mars) stacked on two already-profitable, competitively unassailable franchises (launch, Starlink) that alone would justify a premium aerospace/telecom multiple. The launch moat is widening, not static; Starlink is proving its unit economics hold even under aggressive price compression; the government book is becoming a satellite-prime franchise the U.S. cannot source elsewhere; and the AI segment is already at a multi-billion run rate in its first disclosed year, riding the one cost curve that makes orbital compute conceivable. Options are supposed to look expensive right up until one pays off.

Where we land. Both cases are honestly argued and the model vindicates neither fully. The launch and Starlink franchises are as strong as the bull claims, and we price them that way — every scenario grows them through FY30E. Starship risk is real and we side with the bear on near-term cadence, but we model Starship as optionality, so its risk lives in the swing factors, not in a markdown of the core. Key-man/political risk is real but we do not quantify it in the bridge, because it resolved once (June 2025) without moving realized contract dollars. The decisive, quantifiable finding is the valuation gap, and it favors the bear: the price embeds AI-segment success we are not willing to underwrite at even 50/50 odds on one year of data.

---

Risks (ranked)

1. AI-segment monetization and cash burn — the dominant risk in both directions. FY2025 AI operating loss (–$6.355B) exceeds the entire company's net loss; Q1 2026 burned –$2.469B on $818M of revenue. If losses do not narrow, they structurally offset Connectivity's profitability. This is also where the upside risk to our Underweight lives: a sharp acceleration would invalidate the call.

2. Valuation / multiple compression. At ~50x FY27E revenue, the stock needs the multiple to hold to avoid underperforming, independent of operating results. Any rotation away from AI-infrastructure multiples hits SPCX disproportionately given no comp support.

3. Starship program and regulatory risk. A 50% failure-or-partial rate across six flights and two FAA groundings in eighteen months. A third mishap within the window would reinforce the technical-risk discount; the cost-curve and Starlink-V3/Mars roadmap all depend on the vehicle graduating to operational.

4. Capex outrunning revenue / self-funding stress. Capex/revenue at 111% in FY2025, predominantly AI-driven. Even post-$75B IPO, our base case requires a return to debt issuance by FY28E–FY30E. A worse trajectory pressures the equity story and the credit profile at once.

5. Key-man and political-relationship risk — realized, not hypothetical. The June 2025 Trump–Musk feud put ~20% federal revenue in play within days (Trump threatened to terminate contracts; Musk threatened to decommission Dragon), with NASA/Pentagon publicly pushing competitors to accelerate alternatives, before a Vance-brokered reconciliation by May 2026 (NPR/Washington Post, 2025-06-07). A repeat would justify moving this from a qualitative overlay into the quantified bridge.

6. Starlink ARPU decline reaccelerating. Structural, management-guided decline from $99 (2023) to $66/mo (Q1 2026). Our base case assumes it decelerates to a ~$52 floor; a reacceleration would pressure base-case Connectivity margins directly.

7. Federal revenue concentration. ~20% of revenue from federal agencies — durable in aggregate but exposed to budget-cycle and re-competition risk, most acute for the already-loss-making Space segment.

8. Governance / control. SpaceX has historically maintained a founder-controlled voting structure; a super-voting arrangement preserving Musk's control post-IPO is our working assumption but is unconfirmed pending a direct 424B4 "Description of Capital Stock" pull (filings.md §2e).

---

What would change our view

• Toward the bull (would compress or close our discount): The first 10-Q (~August 2026) showing AI-segment revenue accelerating past a ~$5B annualized run rate with narrowing losses — the single most consequential near-term disclosure. Independent confirmation of a scaled compute/enterprise-API contract (the circulating, unverified Anthropic ~$1.25B/month claim, if confirmed via primary text or Anthropic's own disclosure, would be exactly this). Three consecutive clean Starship flights (from Flight 13, 2026-07-16), which would promote Starship from optionality to a base-case Space driver.
• Toward the bear (would reinforce or widen Underweight): A flat or decelerating AI-segment print in the first 10-Q. A third FAA-relevant Starship mishap inside 18 months. Starlink ARPU decline reaccelerating in Q2/Q3 2026. A repeat political-relationship episode or a Golden Dome/NSSL re-competition on the federal base.
• Mechanical restatement trigger: The confirmed share count from the 424B4 cover page. Our entire per-share arithmetic rests on a back-solved ~13.0B shares; a materially different filed count rescales every price-target figure proportionally and would be restated before the next coverage update per PROCESS.md §4.

We would revisit the rating on any of the above. Absent them, the price embeds more AI-segment success than one disclosed year of data supports.

---

Appendix — Sourcing & disclosure status

Tier migration. This name was scoped when SpaceX was assumed private (Tier 2). The desk's own disclosure sweep found the company had listed on Nasdaq as SPCX on 2026-06-12 (post the February 2026 xAI merger), and the initiation was rebuilt Tier 1 — anchored to the S-1/424B4 and 8-Ks, with the call expressed against the traded price. The tier-migration protocol (PROCESS.md §2/§4) was exercised on day one. Where the old Tier 2 framing survives in supporting notes, the underlying facts are good but the framing has been superseded here.

What anchors the actuals. No 10-K or 10-Q has been filed. The S-1 (filed 2026-05-20, amended 2026-06-01/03) and 424B4 (2026-06-12) are the only audited-adjacent, line-item financial disclosure that exists; all FY2023–FY2025 and Q1 2026 figures trace to that document. Per PROCESS.md §5, every specific figure carries an inline source and date on first use; desk estimates are labeled E in model.md.

Sourcing caveats carried forward from the sweep. Several S-1 figures were digested from secondary compilations (Via Satellite, HL.co.uk, KraneShares, Vested Finance, Morningstar) rather than a verbatim primary-text pull; those are flagged in filings.md §6 and should be closed against EDGAR primary text before the model locks its actuals. The ~13.0B share count is back-solved, not disclosed. The Anthropic ~$1.25B/month compute claim is unverified and is excluded from every model input. Post-IPO share count, exact vote structure, capex-by-segment split, and verbatim risk-factor text remain open until the first 10-Q/10-K.

First 10-Q (quarter ended 2026-06-30, expected ~August 2026) will be the first GAAP periodic filing — the first audited quarter-over-quarter trend on the AI segment that dominates our price target, the first confirmation of the over-allotment exercise and final share count, and the trigger for our next coverage update.

---

Disclaimer: This is an AI-generated research exercise produced by the chenzoo Coverage desk from public information. It is not investment advice, not a recommendation to buy or sell any security, and not a substitute for professional financial advice. Forward figures are desk estimates and may prove materially wrong; actuals are drawn from SpaceX's S-1/424B4 and secondary compilations thereof, some of which remain unconfirmed against primary text. The author holds no position. Do your own diligence.