Two compounders and one $1.4T option, priced as if all three are proven.

The moated core. Starlink booked $11.4B of revenue at a 62.9% Segment EBITDA margin in FY25, up from 50.7% a year earlier; Falcon won 11 of 12 NSSL Phase 3 missions and put more than 80% of global mass to orbit in 2023–25. Apply defensible peer multiples — 25–30× on Starlink segment EBITDA, 10–15× on normalized launch — and the moated core fair-values at roughly $300–400B.

The unproven frontier. The other ~$1.4T of the IPO equity sits on xAI / Grok, the Starship cost-curve extension to sub-$500/kg, and the orbital AI compute thesis. xAI lost $6.4B on $3.2B of revenue in FY25 against hyperscalers running 10× the data-center footprint with installed enterprise distribution.

The frame that matters. Forcing one multiple across all three is the analytical error. Underwrite Starlink + Falcon on cash earnings; treat Starship and xAI as long-dated options whose strike prices are observable and whose time decay is real. The IPO is being priced as if those options are already in the money.

Starlink prints, Falcon reinvests, xAI burns — and consolidation hides all three.

FY25 consolidated revenue rose 33% to $18.7B, but the income statement is the sum of three different businesses. Starlink earned $4.4B of operating income on a constellation that already self-funds; Falcon-plus-Starship printed $3.7B of segment EBITDA before $3.0B of Starship R&D rolled through it; xAI consumed $12.7B of capex on $3.2B of revenue. The ~$75B of IPO proceeds bridge the $14B free-cash-flow hole until xAI Segment EBITDA inflects or capex tapers — neither dated in the prospectus.

Three forensic flags sit between the bull narrative and the cash.

• ARPU is falling before Kuiper has launched. Blended Starlink ARPU went $91 (FY24) → $81 (FY25) → $66 (Q1 FY26) — a 27% drop in 15 months. Management cites international mix; mature-geo ARPU is undisclosed. Amazon Kuiper expects first commercial service in 2026–27 with $10B+ already committed against Starlink.
• The $6.8B operating cash flow is a payables loan. Accounts payable jumped from $4.4B to $11.8B in FY25 — a +$7.4B swing tied to data-center construction — plus a separate $1.15B Q1 FY26 deferred-revenue inflow. Once construction stabilizes, AP normalizes and cash flow collapses toward the AI segment's underlying loss.
• The CFO's pay metric was switched two months before pricing. The CFO's 4M performance options originally vested on free cash flow above $2B per year. Actual FY25 FCF was −$14.0B. In January 2026 the board re-pegged vesting to Adjusted EBITDA. The bar was lowered and the metric was switched after the target was missed.

Founder-controlled, mission-aligned, minority-thin — and partly tied to a director's firm.

• Musk holds 85% of the votes. Ten-vote Class B super-shares give Elon Musk ~85.1% of voting power post-IPO and the unilateral right to elect 51% of directors. SPCX will list as a Nasdaq controlled company without a fully-independent compensation or nominating committee. The S-1 confirms there is no key-person insurance on him, while an April 2026 Pampena v. Musk partial 10(b) judgment (entered 2026-04-03 against Musk personally) remains open.
• Pay is conditioned on Mars. The January 2026 grant of 1.0B performance Class B shares vests on market-cap tranches from $500B to $7.5T and the establishment of a Mars colony of at least 1 million people. A separate March 2026 grant adds 302M shares tied to non-Earth data-center milestones. Combined: ~16% of basic post-IPO shares. What they vest against is what the IPO price is being asked to underwrite.
• $20.2B of equipment leases sit with a director's firm. Three xAI compute-equipment leases with Valor — controlled by board director Antonio Gracias — total $20.2B of scheduled cash payments and are guaranteed by SpaceX. The Jan–Feb 2026 run rate of $857M in two months implies $5B+ per year of cash outflow to a related party.

Three windows decide whether the $1.75T price has a cash anchor.

• 2026-06-12 — IPO pricing, then 40-day quiet period. The book test versus $1.75T. An Aramco-style downsize to $1.4–1.5T, or a first-week drawdown, would frame the July 22 sell-side initiations against a softer tape. Pricing at or above the reference sets the equity cost of capital for every dollar of follow-on capital through 2028.
• Q3 or Q4 2026 — First post-IPO 10-Q. The cleanest single forensic test on this name. Watch the AP balance against $11.8B and whether cash flow holds when payables stop growing. A mature-geo Starlink ARPU breakout, if disclosed, settles the margin debate in one print.
• 2026-12-09 — 180-day insider lock-up expiry, with Starship V3 in the same window. The largest known liquidity event of the first year. By then two quarterly prints exist, Starship V3 is either flying or slipped past its 2H 2026 target, and Founders Fund and early VC supply becomes free to sell.