SpaceX prices the largest IPO in history tonight. Here's what the S-1 says — and what it changes for the supply-chain map

Tonight, after the close, SpaceX prices its IPO. Tomorrow — June 12 — it lists on Nasdaq as SPCX: roughly 555.6 million shares at an indicated $135, a $75 billion raise at a $1.75 trillion valuation. The largest initial public offering in stock market history, and demand reportedly reached 2–4× oversubscribed within days of the roadshow, with up to 30% of shares discussed for retail allocation — several times the usual 5–10%.

A week ago we published a data-driven map of SpaceX's supply chain and the Starlink economy, built entirely from what other companies disclose in SEC filings — because SpaceX itself filed nothing. As of tomorrow, that asymmetry ends. This post does two things: reads the S-1 numbers now on the table, and updates the map with what changes when the center of the graph starts filing.

(Financial research, not investment advice. S-1 figures below are as reported in the public prospectus and press coverage of it; evidence labels at the end.)

The S-1 numbers, in one table

For a company that spent two decades private, the S-1 is the first audited look inside. The headline figures:

The loss is new — 2024 was reportedly profitable on $15–16B of revenue. The swing is not the rocket business deteriorating; it's the February 2026 all-stock acquisition of xAI (at a reported $250B valuation) consolidating an AI lab's burn onto SpaceX's income statement.

The segment split is where the story is:

Three observations:

1. Starlink is now the company. It crossed 9 million subscribers by end-2025 and 10.3 million across 164 countries by early 2026 — up from 4.5 million a year earlier. A 63% EBITDA margin on $11.4B of connectivity revenue is satellite-telco economics nobody else has achieved. The pressure point is ARPU: reported down ~18% to ~$81/month as the mix shifts toward consumer and emerging markets.
2. xAI turns SpaceX into an AI-infrastructure buyer. In Q1 2026, xAI's capex was $7.7B — 76% of the entire group's capital spending. The rocket company is now, by dollars deployed, primarily a data-center builder. That matters for valuation (Goldman's underwriter forecast has AI at $322B of a projected $474B 2030 revenue) and, as we'll get to, for the supply-chain map.
3. Starship is the open-ended line. $15B+ cumulative development spend through 2025, against a future of higher-cadence, lower-cost launch that the S-1 asks investors to underwrite.

And the multiple: at $135/share, SPCX prices at roughly 94× 2025 revenue and 266× adjusted EBITDA. Morningstar's pre-IPO fair-value range is $600–800B — the market is paying more than twice the high end. Musk retains ~82.4% voting power through the dual-class structure, with compensation reportedly tied to market cap and Mars-mission milestones. Whatever else this is, it is not priced on trailing numbers.

We searched our own index for the S-1. What came back is its own datapoint

Habit check: every EvidInvest claim should trace to a primary source, so we ran Aether — our SEC-native search engine — for SpaceX's registration statement.

Two honest findings.

First: SpaceX's S-1 (publicly filed May 20, 2026) is not yet in Aether's index, which covers ongoing SEC reporters — 10-K/10-Q/8-K filers and their exhibits. New-registrant S-1s enter the pipeline at listing. From tomorrow, SPCX becomes a reporting company and its filings flow in like any other ticker. We flag this because pretending otherwise is exactly the kind of overclaim this blog exists to avoid.

Second: what the search did return is a snapshot of market froth. The top hits for "SpaceX S-1 Starlink" were blank-check companies that have draped themselves in the name: "Starlink AI Acquisition Corporation" ($OTAI — a Cayman SPAC, $100M trust, whose own auditor flags going-concern doubt) and "Space Asset Acquisition Corp." ($SAAQ), whose prospectus pitches the space economy growing from $613B (2024) to $1.8T by 2035, leaning on SpaceX's reusability story. Neither has any relationship to SpaceX. When shell companies are named to ride a ticker that doesn't trade yet, that is sentiment data — the kind that historically clusters near local tops in a theme.

What the IPO changes for the supply-chain map

Our June 4 map had a structural constraint: SpaceX filed nothing, so every edge in the graph came from the other side — AMD's SEC-filed exhibit naming Starlink satellites on Versal SoCs, United and T-Mobile describing Starlink deals in their own earnings materials, teardown-level component evidence in the terminal BOM.

Public listing inverts that. Three concrete upgrades:

1. The center of the graph starts disclosing. From its first 10-Q, SPCX must report segment results, purchase obligations, supplier concentration where material, customer concentration (NASA and the U.S. government will be quantified), and risk factors in its own words — every quarter. Claims that lived as "component presence" or "capability proxy" in our labeling system can graduate to SEC-filed, or die. The map stops being one-sided.

2. Aether already found a node we missed: EchoStar ($SATS). Searching the supply-chain domain for SpaceX counterparties, the strongest new evidence wasn't a parts supplier — it's a multi-billion-dollar spectrum transaction. Per EchoStar's 10-K (filed 2026-03-02) and Q3-2025 10-Q (SEC source):

• EchoStar agreed to sell AWS-4, AWS-3 and H-Block spectrum licenses to SpaceX, with $9.82B of Seller Notes outstanding against the licenses as of year-end 2025.
• SpaceX committed ~$2B of interim debt service on those notes through November 2027 via a credit agreement, and EchoStar's $414M of cash interest payments to date are reimbursable by SpaceX at closing.
• The deal bundles long-term commercial agreements: EchoStar's wireless subscribers get access to Starlink Direct-to-Cell, a fee-based referral program routes HughesNet customers to Starlink, and EchoStar has already begun performing installation services for new Starlink customers.

That last line is the interesting one: a legacy satellite competitor converting into a Starlink channel and service layer, documented in its own filings. Spectrum is the raw material of Direct-to-Cell — this is supply chain in the truest sense, and it carries dollar figures, dates and enforcement terms, which almost nothing else in the SpaceX map does. (Run the SATS valuation →)

How big is the whole spectrum trade? A third party put a number on it: Crown Castle's January 2026 press exhibit (SEC source) describes EchoStar's licenses "being sold for more than $40 billion" to AT&T and SpaceX — in the same release where Crown Castle terminates its DISH agreement and pursues $3.5B in defaulted payments after DISH discontinued its network business. The SpaceX spectrum grab is large enough to produce litigation-grade collateral damage two counterparties away. That's what a supply-chain graph is for.

3. xAI's consolidation plugs SPCX into the AI-infrastructure series. With 76% of group capex flowing to AI data centers, SPCX joins the demand side of the build-out we've been documenting ticker by ticker — NVDA, AVGO, MU, the optics and interconnect names. The suppliers in those posts already disclose hyperscaler concentration; a $30B+/year capex program at xAI is the kind of customer that eventually shows up, named or unnamed, in their filings. Cisco's latest 10-Q already describes new purchase commitments for Silicon One and memory components "currently constrained" to meet hyperscaler demand — the same shortage driving the memory names this month. One more deep-pocketed buyer does not loosen that market.

And the most SEC-concrete new edge in the graph runs through Tesla. Tesla's Q1 2026 earnings exhibit (SEC source) states it plainly: "Our partnership with SpaceX aims to build the largest chip fab ever: vertically integrating logic, memory and advanced packaging," beginning with a Tesla-owned Research Fab at the Gigafactory Texas campus, with Tesla's AI5 inference processor design completed in April. Read that against the xAI capex line: the Musk complex is responding to the same constrained chip and memory market our AI-infrastructure series tracks by attempting to vertically integrate out of it — a structural demand signal for semicap equipment, and a long-run structural threat to the foundry and memory incumbents it currently buys from.

How to trade the information (not necessarily the stock)

We'll keep this in the house style — what the evidence supports, not a price target:

• The S-1 supports the Starlink thesis and prices the Mars thesis. $11.4B revenue at 63% EBITDA margin is a real, audited, compounding asset. It is not, at $1.75T, the thing being paid for — the premium over Morningstar's $600–800B range is the market underwriting Starship, xAI, and Musk execution simultaneously.
• The proxy basket logic from our last post survives the IPO — and the demand side keeps compounding on schedule. Fresh from Aether's transcript index this quarter: United has completed Starlink installs on 327 United Express aircraft with fleet-wide coverage expected by end-2027 (Q1-2026 release, SEC source), and Southwest announced fleet-wide Starlink deployment with the first aircraft in service summer 2026 and at least 300 aircraft by year-end (Q1-2026 release, SEC source). Each quarter, more of Starlink's revenue base is visible in other companies' filings — positions valued on their own income statements.
• Watch the first 10-Q, not the first tick. Day-one pops on 4×-oversubscribed deals tell you about allocation, not value. The first quarterly filing — segment detail, government concentration, xAI burn trajectory, any named suppliers — is when the evidence map actually updates. Aether will have it indexed the day it hits EDGAR.

Monitor it yourself

• Search the filings with Aether — try "SpaceX" license purchase agreement spectrum to pull the EchoStar evidence chain, or Starlink agreement aircraft for the demand side.
• Run the financials on the map: SATS, AMD, UAL, TMUS — and SPCX itself once it has a trading history.

Evidence labels: SEC-filed — EchoStar transaction terms (SATS 10-K 2026-03-02, 10-Q 2025-11-06, linked); Tesla–SpaceX chip fab partnership (TSLA Q1-2026 exhibit, linked); Crown Castle $40B spectrum figure and DISH default (CCI exhibit 2026-01-12, linked); United and Southwest Starlink rollouts (Q1-2026 exhibits, linked); SPAC filings (OTAI 424B4 2026-05-08, SAAQ 424B4 2026-01-28); Cisco supply commitments (CSCO 10-Q 2026-02-17). Prospectus-reported — SpaceX S-1 financials as carried in the public prospectus and press coverage (Reuters, Morningstar, TradingKey); Aether has not yet independently indexed the S-1. Press-reported — oversubscription levels, retail allocation discussions, Goldman 2030 forecast. Not investment advice.