I handed over SpaceX's 300-page IPO document to Claude and uncovered 5 numbers that the officials deliberately kept low-key.

Author X (@DamiDefi) handed Claude Opus 4.8 the 300-page SpaceX IPO prospectus to find the "digit that is deliberately understated" in the document. Claude identified 5 things. The compilation is as follows.
(Background: Anthropic will pay SpaceX $1.25 billion monthly in exchange for xAI surplus computing power)
(Additional context: Musk announced that xAI is no longer an independent company, renamed "SpaceXAI," positioned as a subsidiary of SpaceX)

Table of Contents

Toggle

• Finding 1: Valuation multiples are not supported by "current" conditions
  • The TAM segmentation in S1
  • Claude’s specific annotations
• Finding 2: Turning profitable into unprofitable in just one year
  • Claude’s annotations
• Finding 3: Starlink subscriptions doubled, but ARPU fell 23%
  • Claude’s annotations
• Finding 4: Retail investor quota 30%—three times the usual
  • Claude’s observations
• Finding 5: Musk holds 85% voting rights + collective lawsuit waiver
  • Claude’s specific notes
• How Polymarket views it
• A summary paragraph from Claude
• This is not investment advice to buy or avoid
• Run it once yourself with Claude
• Timeline

What stops me isn’t the $2 trillion valuation, but $791 million—SpaceX’s net profit after taxes in 2024. A real aerospace company with launch services and satellite internet moat, profitable, and growing.

And then I keep reading.

After completing the 2025 acquisition, the same company reports a $4.94 billion net loss. In Q1 2026 alone, burning $4.28B.

This merger did not combine the "profitable AI business" with the "profitable rocket company." It did the opposite. Then the IPO price is set so you pay up to $2 trillion to buy the entire conglomerate.

I downloaded the full 300-page S1 and handed it to Claude Opus 4.8, giving it one task: find the content "deliberately understated" in the document—not the headline risks in the executive summary, but buried in comparison tables, footnotes, structural disclosures, the numbers that would change the entire picture if you actually read through.

Below are 5 things Claude identified that news reports are not mentioning.

Finding 1: Valuation multiples are not supported by "current" conditions

SpaceX’s full-year revenue in 2025 is $18.7 billion—this number is real, up 33% annually, mainly driven by Starlink’s $11.4 billion (about 61% of total revenue).

At a $1.75 trillion valuation, that’s about 94x revenue. At a $2 trillion valuation, about 107x.

This multiple isn’t supported by SpaceX’s "current" state. It’s supported by the TAM (Total Addressable Market) claimed in the S1.

TAM segmentation in S1

| Business | | --- | | Claimed TAM | | --- | --- | | Space (Space) | $370 billion | | Connectivity (Internet) | $1.6 trillion | | AI | $26.5 trillion |

In reality, Space and Connectivity, which are profitable, combined account for less than 7% of the total TAM. The remaining 93% is in AI. Just the "enterprise applications" within AI is listed as $22.7 trillion.

You are asked to use 2026 dollars to buy a company that needs to capture a meaningful share of the $26.5 trillion market to justify the valuation—yet every major tech company on Earth is competing in that market.

Claude’s specific annotation

S1 does not explain what mechanism SpaceX uses to compete for the AI TAM dominated by Anthropic, OpenAI, Google, Microsoft. TAM is written, but the competitive moat is not.

Finding 2: Turning profitable into unprofitable in just one year

Many people missed this number because it requires comparing with historical data.

| Period | | --- | | Result | | --- | --- | | Full year 2024 (pre-merger) | Net profit $791 million | | Full year 2025 (post-merger) | Net loss $4.94 billion | | Q1 2026 (single quarter) | Net loss $4.28B | | Cumulative deficit | $41.3 billion |

The merger did not combine the profitable AI business with the profitable rocket company.

It attached the unprofitable AI infrastructure business (including the X platform) to the rocket company, then sold it at an "AI as main asset" valuation.

AI-related losses are about $2.5 billion per quarter. This is not R&D spending that decreases over time. AI infrastructure at the scale of xAI requires ongoing, and in the short term, larger, not smaller, capital expenditures.

Claude’s annotation

The shift from profit to loss happening within a single year coincides exactly with the xAI merger. S1 presents this as a "strategic investment phase," but does not model when this investment phase ends.

Finding 3: Starlink subscriptions doubled, but ARPU fell 23%

Starlink is the most investable part of this company, and the only unit generating real revenue.

Subscription growth is impressive:

• Q1 2025: 5 million
• Q1 2026: 10.3 million (doubling in a year)

But S1 buries this in the segment financials:

ARPU (average revenue per user) declines 23% annually.

Revenue growth is driven by more subscribers. Revenue per subscriber is declining—because Starlink is pushing into lower-priced international and consumer markets to scale.

The key point: The bullish narrative for Starlink is built on "large subscriber base × healthy gross margin → sustained cash flow."
The actual trend in the document is "large subscriber base × shrinking gross margin."
These are two completely different businesses.

Claude’s annotations

Both subscriber growth and ARPU decline are in the document and are true. The executive summary mentions only the former; the latter appears only in the segment financials.
If you only use subscriber growth to estimate Starlink’s value, you are only seeing half the picture.

Finding 4: Retail investor quota 30%—three times the usual

The norm for large IPOs is about 10% allocated to retail investors, with the rest to institutions.

SpaceX’s IPO allocates 30% to retail—three times the usual.

The sales channels listed include: Schwab, Fidelity, Robinhood, SoFi, ETRADE—all retail platforms. The entire issuance process is designed for individual investors to buy at IPO price, not in the secondary market after listing.

The marketing pitch is "democratization"—allowing ordinary investors to participate at the same price as institutions.

Less salesy phrasing: Venture capital and private equity investors who have held SpaceX shares at much lower valuations for years are now exiting—their shares need new owners.
And those new owners, at 3 times the normal allotment, are retail investors.

Claude’s observation

S1 does not hide this. The 30% retail quota is clearly disclosed. It’s in the document, but the framing is not.

Finding 5: Musk holds 85% voting rights + collective lawsuit waiver

This part deserves the most attention but gets the least coverage.

• Musk owns about 42% of the shares
• Voting rights about 85%

This means: Every other shareholder, combined, holds only 15% voting rights in this $2 trillion valuation company.

S1 also includes:

• Arbitration clause
• Class action waiver

Shareholders with claims cannot file class actions, only individual arbitration.

Together, this creates:

• A voting structure where no major decision can be overturned without Musk’s approval
• A legal structure where shareholders cannot act collectively if things go wrong

Claude’s specific notes

This governance structure is disclosed in the risk factors section of the S1. But it is presented as a "feature of the company’s operation," not as a "risk to investor returns."
The real risk—"if capital allocation decisions destroy value, there is no mechanism for shareholders to hold management accountable"—is not framed as such.

How Polymarket views it

Your readers think in terms of "positions," not just "opinions."

Polymarket’s SpaceX market this week shows:

| Range | | --- | Probability | | --- | --- | | Over $2 trillion | 47% (leading option) | | $1.8–2 trillion | 18% |

The prediction market believes IPO will close at or above the top of the range.

The S1 financials tell a different story: Falcon launch services and Starlink, which are actually profitable, if valued at current earnings multiples, would support a valuation well below $1.75 trillion.
The premium is entirely driven by the TAM claims for xAI and AI divisions.

Two perspectives:

• Either the market is correct—SpaceX will capture a meaningful share of the $26.5 trillion AI market, and the premium is justified
• Or the market is pricing in a narrative that even the S1 cannot support with a competitive differentiation argument

Claude’s summary

After reading the entire document, I asked Claude: If you had to describe in one sentence what this prospectus is really selling, what would you say?

It responded:

SpaceX is selling three businesses under one ticker: a profitable launch services company (Falcon 9, Falcon Heavy), a growing but margin-compressed satellite internet business (Starlink), and a loss-making AI and social infrastructure business (xAI, including X).

The valuation is based on the third business as the main asset, using a TAM of $26.5 trillion, but the S1 does not explain why it would capture that share. The first two businesses have actual revenue and real competitive moats, but their standalone valuation would be significantly lower than the current asking price.

The IPO price is essentially what buyers are paying for—"the scale of the third business"—before that scale even exists.

That’s what this document, condensed into its core message, is really saying.

This is not investment advice to buy or avoid

The launch services business is truly exceptional. Starlink’s subscriber count doubling from 5 million to 10.3 million in a year is real. The moat of reusable rockets is rare in tech.

The question isn’t "Is SpaceX a good company?" The question is:

Is $1.75–2 trillion a reasonable price for "this company now," under this governance structure, during this xAI merger phase?

That’s a different question. The S1 answers the former; it’s less useful for the latter.

Run it yourself with Claude

The document is public and free. Download from SEC EDGAR, case number 333-296070.

Upload it to Claude Opus 4.8’s project, with this prompt:

I uploaded SpaceX’s S1 prospectus. Before producing any content, please read the entire document.

Then produce:

1. Three financial disclosures revealed in the document but not mentioned in the executive summary, which would materially change investment judgments
2. Any "headline figures" that contradict underlying trends
3. A single risk factor disclosed but underestimated in importance due to framing
4. Using current multiples, how much revenue xAI must generate in the future to justify the valuation—please give specific numbers

Do not summarize SpaceX’s strengths from the S1.
Focus entirely on what the document reveals when used as a reference for self-assessment.

Timeline

• Roadshow: starting June 4
• Pricing: around June 11

After institutional bookbuilding ends, what you buy is momentum in the secondary market, not fundamentals.