SpaceX IPO Deep Dive: Key S-1 Takeaways, Traps, and How Investors Can Access It by Region

In May 2026, SpaceX filed its IPO registration statement (S-1), with the proposed ticker SPCX. Headlines erupted, and most people saw some version of: “the rocket company is going public at a valuation as high as US$1.75 trillion.”

But if you actually open the filing, you find that SpaceX has long since moved beyond a simple rocket company. It ties rockets, satellites, and AI into one package. This piece reads the S-1 directly: first the key points you need to understand, then the traps most likely to trip people up, and finally how investors in different regions can actually get access. To be clear upfront, this article does not call any direction on the stock; it only helps you understand the document. If you want to get to know SpaceX as a company first, read What Is SpaceX?.

The figures in this article mainly come from the S-1 SpaceX filed with the SEC (accession 0001628280-26-036936, 2026-05-20). Items marked [S-1] are from the filing text, items marked [media] are from media reports, and items marked [calculation] are calculated from S-1 figures. These three kinds of source should not be mixed.

---

Part One. Key Points: Understand These First

The IPO itself

Start with the skeleton:

• Issuer: Space Exploration Technologies Corp.
• Ticker / exchange: SPCX, proposed dual listing on Nasdaq and Nasdaq Texas [S-1]
• Filing status: preliminary S-1; share count, price range, and fundraising size are all still blank [S-1]
• Underwriting syndicate: Goldman Sachs leads, followed by a long list of major investment banks including Morgan Stanley, BofA, Citi, JPMorgan, Barclays, and others [S-1]
• Retail access channel: the S-1 explicitly lists participation through platforms such as Schwab, Fidelity, Robinhood, SoFi, and E*TRADE [S-1]

Here is the first point to remember: the S-1 fee table contains a “maximum aggregate offering price” of US$1 billion, but that is the basis for calculating registration fees, not the amount SpaceX is actually trying to raise, and not its valuation [S-1]. The US$1.75 trillion target valuation reported by media is an underwriting and market negotiation number [media], not a market price at which shares have already traded. We will come back to both numbers later.

It is really “three major segments,” not four lines

Many reports describe SpaceX as a four-in-one mix of rockets, satellites, AI, and social media. But in the S-1 financials, after absorbing xAI and X, the company began reporting under three major segments from the first quarter of 2026 [S-1]:

• Space: rocket launches, Starship, government and commercial launches
• Connectivity: the Starlink satellite network (consumer + enterprise + government)
• AI: Grok, Colossus compute capacity, plus X advertising, subscriptions, data licensing, and APIs

That last point matters: the AI segment’s revenue includes X advertising revenue. So the question “how much pure Grok monetization is there?” cannot be separately answered from the S-1 [S-1]. When you see “AI revenue of US$3.201 billion,” do not treat it directly as Grok’s earning power.

Revenue by the three major segments in 2025 looked like this:

Source: [S-1]; shares are [calculation]. The picture is obvious at a glance: the main revenue pillar is Connectivity, where Starlink sits, not the AI business that excites people most.

The real cash-flow engine: who is funding whom

Revenue alone is not enough. You need to see which piece earns and which piece burns. Spread out operating income and loss across the three segments, and the story becomes clear:

Source: [S-1]. This is the one table in the whole S-1 worth remembering most: Starlink’s profits are funding AI’s massive losses. Connectivity made US$4.423 billion in a year; AI lost US$6.355 billion in a year. The AI buildout is being supported by satellite-network cash flow, financing, and balance-sheet flexibility.

Why can Starlink carry so much? Because it is growing: subscribers rose from about 5 million in March 2025 to about 10.3 million in March 2026 [S-1]. That is a doubling in one year. It is also why SpaceX dares to burn cash on AI this way.

Overall financial condition

Pull the lens back to the whole company:

• Full-year 2025: revenue of US$18.674 billion, operating loss of US$2.589 billion, net loss of US$4.937 billion; gross margin of about 49.4% [S-1/calculation]
• Q1 2026: revenue of US$4.694 billion, operating loss of US$1.943 billion, net loss of US$4.276 billion [S-1]
• Cash on hand (2026/3/31): cash of US$15.852 billion + marketable securities of US$7.823 billion; plus about US$29.1 billion in debt principal, though the nearest principal maturity is not until August 2027 [S-1]

In other words, this is a company whose revenue is growing but whose books show large losses because it is investing aggressively. It is not out of cash (2025 operating cash inflow was US$6.785 billion [S-1]); it is pouring cash and borrowed money heavily into capital expenditure.

How to think about valuation: what kind of number is US$1.75 trillion?

Media reports put the target valuation at about US$1.75 trillion [media]. How large is that? Divide it by 2025 revenue of US$18.674 billion, and it equals a price-to-sales ratio (P/S) of about 93.7 times [calculation].

This is not a traditional aerospace multiple, where single digits to the low teens are more typical. The only way to make it remotely coherent is to read it as a kind of “sum-of-the-parts (SOTP) + option value” pricing: Starlink as a high-growth global broadband platform, AI as a future bet far beyond current revenue (Colossus, Grok, X data, orbital compute, and other options), and rockets as long-term strategic value.

The point to remember: US$1.75 trillion is an underwriting and market negotiation number, not a market transaction price [media]. The actual price will not be known until the listing prices and the book is built. Some analysts are bullish (for example, Reuters quoted Wedbush’s Dan Ives describing SpaceX as central to multiple long-term themes [media]), but that is an analyst view, not a buy or sell recommendation from this article.

---

Part Two. Traps: Where Readers Are Most Easily Misled

Once the key points are clear, read it from the other side: where are people most likely to get it wrong?

Trap: treating “total revenue” as “AI revenue”

This is the most common misreading. People see “SpaceX revenue of US$18.7 billion” and assume AI is already a strong earner. In reality, the AI segment is only about 17% of revenue, and even that 17% includes X advertising. Treating the whole company as AI will seriously overstate its AI monetization.

Trap: treating “total capital expenditure” as “AI capital expenditure”

Media often writes that “SpaceX is spending US$20 billion on AI.” The S-1 text shows why that is imprecise:

Source: [S-1]; shares are [calculation]. US$20.737 billion is the total; AI accounts for about US$12.727 billion of it. The right reading is “AI consumed most of capital expenditure,” not “everything went into AI.” It is worth noting that AI’s share jumped to 76% in Q1, meaning the money is tilting toward AI quickly.

Trap: treating a “valuation number” as a “market price”

As noted earlier: the US$1 billion in the fee table is a fee-calculation basis, not the amount being raised; US$1.75 trillion is a negotiation number, not a transaction price; and the preliminary S-1 has not even filled in the share count or price. Retail investors are especially prone to anchoring on one dramatic big number while missing that the actual terms are not set.

Trap: treating “economic ownership” as “voting control”

This stock has a dual-class structure, which matters a lot for minority shareholders:

• Class A: 1 vote per share
• Class B: 10 votes per share, and Class B holders elect 51% of board seats [S-1]
• Class C: no voting rights

Before the offering, Musk held about 85.1% of the voting power [S-1] (the exact post-offering percentage is left blank in the S-1, but the filing states that he will retain a majority). The company also plans to rely on Nasdaq’s “controlled company” exemption, so it does not need a fully independent compensation committee [S-1].

Meaning: even if the Class A shares you buy represent meaningful economic ownership, your real voice in board elections, mergers, compensation governance, and strategic shifts is extremely limited. Buying this stock means accepting the premise that the company is firmly controlled by Musk.

Trap: assuming AI cash burn can always be supported by satellites

Part One showed that Starlink is funding AI. Now flip that around: this cross-subsidy has conditions. If Starlink’s cash flow reverses one day because of competition, price cuts, or rising satellite replacement costs, the financial foundation for the AI buildout will shake with it. The fact that it can support AI now does not mean it can do so forever.

Trap: treating “vision” as something that has already happened

The S-1 contains a reading technique worth learning: look at what the company itself says in the risk factors. In the filing, SpaceX explicitly describes its in-house chip plan (TeraFab), orbital compute, and related projects as early-stage, capital-intensive, and possibly never commercialized [S-1]. The specific TeraFab project, timeline, and capital expenditure are even “not yet determined” [S-1].

When a company says in its own prospectus that a plan may not succeed, that is exactly the signal to take seriously. Read the story-stock narrative separately from the filing’s own warnings.

Other items, such as the fact that 2024 net income included an unrealized gain on digital assets (lower-quality earnings), adjusted EBITDA making AI losses look much smaller, and post-listing lockup expirations plus dilution from Musk’s huge performance stock awards, are all points advanced readers can dig into further. For space, this article only flags them.

---

Part Three. How Investors in Different Regions Can Get Access

This final section is about the mechanism: whether you can actually buy it from where you live. This explains channels; it is not a recommendation to buy.

Start with one general rule: IPO subscription (getting an allocation before listing) and buying in the market after listing are two different things. The former has high thresholds and scarce allocations; most non-U.S. investors in practice use the latter.

United States

This is the most direct path. The S-1 explicitly lists IPO participation through platforms such as Schwab, Fidelity, Robinhood, SoFi, and E*TRADE (subject to account eligibility and platform rules) [S-1]. After listing, it is ordinary secondary-market trading. On taxes, the U.S. Internal Revenue Service (IRS) taxes capital gains, with the rate depending on holding period and income; for most individuals, the long-term capital-gains rate does not exceed 15%.

Taiwan

For Taiwan, the practical situation starts with a clear split between “IPO subscription” and “buying after listing.”

IPO subscription (getting an allocation before listing) is essentially unavailable to Taiwanese retail investors. We checked IBKR (Interactive Brokers), Schwab, and Tiger Brokers ourselves, and these firms currently do not provide a subscription/application channel for this SpaceX IPO. As for domestic sub-brokerage, there are community reports that investors may need proof of more than NT$30 million in assets to have a chance to participate; note that this is an online claim, not an official figure, and actual thresholds and conditions should be checked directly with your broker (for example Yuanta or Mega).

The more realistic route is buying after listing. Once SPCX formally lists and begins trading, investors can buy in the secondary market through two paths:

1. Sub-brokerage: a domestic broker is entrusted to trade foreign securities, the most common route.
2. Overseas broker: such as IBKR (Interactive Brokers), where you open a U.S. brokerage account yourself.

Costs and taxes require attention. Taiwan Stock Exchange investor-education materials remind investors that when buying U.S. stocks through sub-brokerage or overseas brokers, cross-border remittance fees, FX conversion fees, and total trading costs can be less transparent. For taxes, U.S. stock dividends and capital gains count as “overseas income” under Taiwan’s Alternative Minimum Tax framework (basic income): household overseas income of NT$1 million must be included, but tax is actually triggered only if basic income exceeds NT$7.5 million. Taiwan does not have a single one-line “U.S. stock capital-gains tax rate,” because it runs through this overseas-income framework.

Japan

Japanese investors generally use local brokers that support U.S. stocks (SBI, Rakuten, Monex, and others). One distinction matters: direct U.S. IPO subscription is not common. Monex’s own explanation, for example, says it does not handle U.S. IPO subscriptions, though post-listing trading is available. Most Japanese investors in practice buy after listing. On taxes, dividends from listed shares are generally taxed at 20.315% (national tax 15.315% + local tax 5%); exact capital-gains treatment depends on account type, withholding choice, and whether NISA is used, so confirm with a broker or tax adviser.

Mainland China

This is the hardest path. The key points:

• QDII is the only clearly compliant channel for offshore securities investment: exposure is indirect through QDII funds from domestic fund companies, but quota is limited and most QDII products will not make a single new stock their sole bet.
• Each person has an annual US$50,000 FX purchase quota, and the declared purpose of FX purchase may not be “offshore securities investment” (direct personal offshore securities investment is not currently open).
• Therefore, a path for “domestic individuals using domestic funds to directly subscribe to a U.S. IPO” is almost infeasible under current foreign-exchange and securities regulation.
• Taxes are also the most uncertain, and should be treated as a professional-adviser matter.

The pragmatic order is: for exposure, use QDII indirectly, or under a compliant setup wait until after listing and use existing compliant channels. Those are the relatively feasible choices for most domestic individuals.

Regional summary: for “who can most easily participate in the IPO subscription,” U.S. investors come first; for “who can realistically buy SPCX,” Taiwan (sub-brokerage / overseas brokers) and Japan (mostly after listing) are also feasible, while mainland China mainly uses indirect QDII exposure, and direct primary-market participation is the least practical.

---

Conclusion

At its core, the SpaceX IPO sells a package that bundles three business lines together. The key to reading it is repeatedly making three separations:

• Separate total revenue from AI revenue (AI is only about 17%, and it includes X advertising).
• Separate the negotiated valuation number (US$1.75 trillion) from the market transaction price (not priced yet).
• Separate economic ownership from voting control (Musk keeps firm control).

Add one more habit: take the company’s own “may not succeed” language in its risk factors more seriously than promotional tone.

To repeat one last time: this article breaks down the IPO to help you understand it; it is not advice on whether to subscribe or buy. Any investment decision should depend on your own risk tolerance, or on professional advice. To put these financials back into the context of the whole company, read What Is SpaceX?; to understand why even a rival like Anthropic is renting compute from SpaceX, read Why Even Anthropic Is Renting Colossus.