SpaceX IPO: What the S-1 Actually Says

Friday’s offering is one of the most consequential IPOs in a generation. This piece goes beyond the headline — segment by segment, so you understand what you’re actually buying before trading begins.

In This Piece

1. What You’re Actually Buying: Three Companies in One
2. Connectivity — The Cash Machine That Funds Everything Else
3. Space — The Infrastructure Bet Hidden Inside an R&D Loss
4. AI — The CapEx Story: xAI, Grok, X, and the Physical Stack
5. Consolidated Financials: Reading the Numbers Honestly
6. Liquidity, Debt Structure, and the Bridge Loan
7. Valuation: What 94x Revenue Requires You to Believe
8. Governance and the Musk Question
9. The Lock-Up Schedule and What It Means for Near-Term Volatility
10. Bottom Line

I. What You’re Actually Buying: Three Companies in One

The S-1 is unambiguous: SpaceX is three distinct operating segments — Space, Connectivity, and AI — with varying maturity, different financial profiles, and in some respects limited immediate operational overlap. Understanding what you own requires treating each independently before considering the long-term thesis that ties them together.

The company’s stated mission is to make life multiplanetary, understand the nature of the universe, and extend consciousness to the stars. That language sounds abstract until you recognize that every one of its current business decisions traces back to a physical infrastructure thesis: the belief that whoever controls the hardware of the future — rockets, satellites, data centers, chips — controls the economic value that flows through it.

The combined entity as of Q1 2026 includes the original SpaceX rocket and Starlink businesses, xAI (acquired February 2, 2026), and X Holdings, which was acquired by xAI in March 2025. The financial statements have been retrospectively recast to consolidate all three as if they were always one company. That matters when you look at the income statement: the AI segment’s substantial losses are now embedded in the historical record alongside Starlink’s gains.

Key context: Only approximately 5% of the company will be available to trade on Friday. All publicly listed shares are Class A, carrying one vote each. Class B shares carry 10 votes. Post-offering, Musk is expected to retain approximately 85% of total voting power. You are buying a minority economic interest in a founder-controlled enterprise. That is neither inherently good nor bad — but it is the precise nature of what you own.

II. Connectivity — The Cash Machine That Funds Everything Else

Starlink is the business that makes the rest of the SpaceX story financially viable. There is no softening that. The Connectivity segment generated $11.4 billion in revenue in 2025, $7.6 billion in 2024, and $3.9 billion in 2023 — a compound growth rate that would be exceptional for any business, anywhere.

What is driving this? Starlink subscriber growth has been extraordinary — 96.5% growth in 2024 and 104.7% growth through Q1 2026 year over year. The S-1 is clear that average revenue per subscriber (ARPU) has been declining in parallel — down 8.1% in 2024 and 22.9% in the most recent quarter — as the company expands internationally into lower-priced markets and introduces tiered plans. This is the deliberate strategic trade: sacrifice unit economics to capture scale, then rebuild margins as costs come down with the launch of the next-generation V3 satellite via Starship.

Why the Margin Trajectory Matters

The key mechanism here is cost structure. Connectivity cost of revenue is heavily weighted toward satellite depreciation and kit delivery costs — both of which scale with volume but decline per unit as Starship’s economics materialize. Each Starship launch can deploy 60 V3 satellites versus Falcon 9’s current capacity, and each V3 satellite offers one terabit per second of downlink capacity. A single Starship launch could provide a 20-fold increase in downlink capacity versus a Falcon 9 mission. When that happens, the cost per megabit of Starlink capacity falls dramatically — and operating margins expand without a price increase.

Enterprise and government revenue is also growing alongside consumer. Aviation, maritime, land mobility, and government connectivity all contributed in 2025 and Q1 2026. The Starshield program — a dedicated, classified constellation for U.S. national security — is a separate but meaningful contributor to the government line. SpaceX is the primary launch provider for the U.S. government and executed 11 of 12 National Security Space Launch missions in 2025.

The bottom line on Connectivity: This is a business with genuine, durable competitive advantages — low latency, global coverage, and a satellite constellation that represents approximately 75% of all active maneuverable satellites in orbit as of Q1 2026. It is already generating the kind of operating leverage that most growth companies spend a decade chasing. It is also the segment that is currently funding everything else.

III. Space — The Infrastructure Bet Hidden Inside an R&D Loss

The Space segment is the public face of SpaceX, and its financials require the most careful reading. The segment posted a $657 million operating loss in 2025, up from essentially breakeven ($21 million income) in 2024. On the surface, that looks like deterioration. It is not. It is the consequence of $3 billion in Starship research and development expense flowing through the income statement in a single year.

What Starship Actually Changes

The economic argument for Starship is straightforward. In 2011, putting one kilogram into Low-Earth Orbit cost approximately $54,500. Falcon 9 brought that to $2,720/kg through reusability. Starship, at scale, is designed to bring that number to $100–200/kg — a roughly 14x reduction from today. The payload capacity increases from Falcon 9’s 23 metric tons to an expected 100 metric tons for Starship V3, with future generations targeting 200 metric tons.

The S-1 discloses that Starship R&D is currently expensed as incurred — because the vehicle is still in development. At commercialization, that changes: Starship costs will be capitalized and then depreciated into the cost of revenue of whichever segment’s payload was launched. The operating loss in the Space segment today is, in economic substance, a capital investment in a technology that transforms every other part of the business. That does not eliminate the execution risk. SpaceX had completed 11 Starship flight tests as of the filing date; the 12th was scheduled. The vehicle has not yet entered commercial operational service.

The Space segment’s revenue mix — 63% Launch Services, 37% Launch and Development in 2025 — reflects its current government-contract-heavy profile. SpaceX services NASA CRS missions, Department of War contracts, and NSSL launches. These are long-duration, cost-plus-adjacent contracts. The revenue base is stable but not fast-growing on its own. Growth in this segment is a Starship story.

IV. AI — The CapEx Story: xAI, Grok, X, and the Physical Stack

This is the segment that makes SpaceX’s IPO genuinely different from anything that has come to market in recent memory. And it is also the segment that demands the most honest accounting of what you are being asked to believe.

xAI was acquired in February 2026 — four months before this IPO. The AI segment includes the COLOSSUS and COLOSSUS II data centers in Memphis and Southaven, Mississippi; the Grok family of large language models; the X social platform; Grok Voice, Grok Business, Grok Enterprise; and the Imagine image and video generation system. As of March 31, 2026, the combined Grok and X platform had approximately 550 million MAUs, with about 117 million using Grok’s AI features.

The Revenue Story Inside the Loss

The AI operating loss is real, but it obscures two things worth separating. First, a significant portion of the loss is GPU depreciation — not cash outflow in the current period. The company acquired and depreciated an extraordinary amount of GPU hardware to build its compute clusters, and that depreciation flows through R&D expense. Second, revenue in Q1 2026 reflects a meaningful shift in mix: AI solutions and infrastructure revenue (Grok subscriptions, data licensing) grew $191 million year over year, while advertising revenue declined $100 million due to a platform advertising rebuild. Advertising headwinds are cyclical. Grok subscription growth is structural.

The Physical Stack Thesis

SpaceX’s S-1 makes an argument that is unlike anything in a typical tech IPO. The claim is that the binding constraint on AI growth is not software or algorithms — it is the physical infrastructure: chip manufacturing, data center capacity, and power generation. The company that controls the physical stack controls the AI era. SpaceX frames itself as uniquely positioned to own that stack from chips (Terafab, a chip manufacturing initiative with Tesla and Intel) through compute (COLOSSUS) through distribution (Starlink, orbital data centers).

The Terafab initiative is worth flagging. It is described in the S-1 as a general framework — specific projects, timelines, and capital expenditures have not yet been determined. The orbital data center concept is similarly aspirational. These are genuinely compelling long-term vectors, but they are not near-term financial contributors. Investing on the basis of Terafab or orbital compute is a 5-to-10-year thesis, not a 2026 catalyst.

Grok and the X Competitive Advantage

The most defensible near-term differentiator in AI is Grok’s access to approximately 350 million daily posts on X — a real-time data feed that no other frontier model possesses natively. While Grok currently lags behind Claude, ChatGPT, and Gemini on many general benchmarks, the S-1 notes it reached frontier-level performance in scientific reasoning (GPQA Diamond benchmark) within two years of the initial release — faster than any competitor has publicly disclosed. The iteration pace is the key variable. An AI company with 1 gigawatt of compute, proprietary real-time training data, and a rapidly advancing model is a different kind of competitor than its current benchmark position suggests. SpaceX also disclosed a compute agreement with Cursor — the AI-native IDE — and holds an option to acquire it outright in the 30-day window following the IPO, which would add $3.1 billion in assets (primarily $2.7 billion in cash) and a major developer-facing product to the platform.

V. Consolidated Financials: Reading the Numbers Honestly

The consolidated income statement tells a story that requires context. Total revenue grew 33% to $18.7 billion in 2025, from $14.0 billion in 2024. The company swung from $791 million in net income in 2024 to a $4.9 billion net loss in 2025. That swing is driven almost entirely by two items: the AI segment’s accelerated compute build (including $3.9 billion in additional GPU depreciation and cloud computing costs), and a $1.3 billion tax provision reversal tied to the One Big Beautiful Bill Act that unwound a deferred tax benefit recognized in 2024.

 depreciation, share-based comp, restructuring, and impairment. It reflects the underlying cash-generative capacity of the operating segments before the AI build-out absorbs it. The Connectivity segment alone generated $7.2 billion in Adjusted EBITDA. The Space segment generated $653 million. The AI segment was negative $1.24 billion. The math shows a profitable, compounding core business funding a high-conviction capital expenditure program.

VI. Liquidity, Debt Structure, and the Bridge Loan

This section of the S-1 deserves more attention than it typically receives in pre-IPO coverage. SpaceX held $15.9 billion in cash and equivalents plus $7.8 billion in short-term marketable securities as of March 31, 2026 — a total liquidity position of approximately $23.7 billion. That is a strong position. The complexity is in the debt stack.

In March 2026, SpaceX drew a $20 billion unsecured bridge term loan to retire the legacy X and xAI debt that came with the acquisition. The bridge loan matures September 2027, with two optional three-month extensions. The company simultaneously amended its existing revolving credit facility to $5 billion (from $1.5 billion), maturing May 2031. The bridge loan creates a near-term refinancing obligation — the company will need to either replace it with permanent capital (likely senior notes or term loans from IPO proceeds or follow-on activity) or extend it. That is a known and manageable task for a company with $23.7 billion in liquidity and investment-grade ambitions, but it is a real item on the calendar.

On the debt: Interest expense rose 23% to $1.9 billion in 2025, driven by GPU financing arrangements and the debt inherited from the xAI acquisition. This is not distressed leverage — but it is meaningful, and the cost of the AI build-out extends into the financing lines, not just the income statement.

VII. Valuation: What 94x Revenue Requires You to Believe

At a $1.77 trillion expected valuation and $18.7 billion in 2025 revenue, SPCX enters trading at approximately 94x trailing revenue. That is not a valuation multiple with historical precedent for a business of this size. It is, instead, a forward-looking investment in what the business becomes. The question worth asking directly: what does the S-1’s own financial history imply about the return available at this price?

If you use the Connectivity segment’s $7.2 billion in 2025 Adjusted EBITDA as the baseline, and apply a 25x EBITDA multiple — generous but not extreme for a business with this growth profile — you get a Connectivity-only value of approximately $180 billion. The Space segment, at $653 million in EBITDA with a significant development option in Starship, might command $50–100 billion at a reasonable growth multiple. That leaves approximately $1.5 trillion of the expected valuation attributable to the AI segment, which generated negative $1.2 billion in Adjusted EBITDA in 2025.

At $1.77 trillion, roughly 85 cents of every dollar of valuation rests on what xAI, Grok, the physical stack thesis, Terafab, and orbital infrastructure become — not what they are today

That is not a reason not to invest. It is the precise nature of the investment thesis. Investors in SPCX at IPO are not primarily buying Starlink’s cash flows — though those are real and valuable. They are primarily acquiring optionality on a set of long-duration technological outcomes that, if realized, could be worth multiples of the current price. If Starship delivers its cost-curve promise, the orbital economy is transformed. If Grok achieves frontier-model status on a 1-GW compute cluster with proprietary real-time data, the AI economics change. If Terafab produces a vertically integrated chip and compute stack, the physical infrastructure thesis is complete. Each of those outcomes requires meaningful execution in an environment where SpaceX is facing well-capitalized competitors in every domain.

The IPO is expected to be oversubscribed. Nasdaq 100 inclusion — which typically occurs 15 days post-offering — will mechanically force index-tracking ETFs to buy shares at whatever the market price is. That technical demand is real and will support the stock near-term. It is not the same as the business being worth the price.

VIII. Governance and the Musk Question

Governance in a dual-class IPO is always a negotiation between investor protection and founder execution. In SpaceX’s case, the negotiation has essentially been resolved in favor of the founder. Musk will control approximately 85% of voting power post-IPO. The Class B structure means Class A shareholders — the public — have limited ability to influence board composition or major decisions. The S-1 explicitly notes that Class B shareholders elect a majority of the board.

This cuts both ways, and the analysis depends on your view of Musk’s track record and incentives. The case for accepting the structure: Musk’s dual roles at Tesla and SpaceX have driven extraordinary value creation. His vision for what SpaceX could become — and his willingness to absorb risk that conventional management would not accept — is what produced Starlink, Falcon reusability, and the Starship program. Investors who are uncomfortable with concentrated control have that concern priced into the offering in the form of the Class A discount that typically applies to such structures.

The case for caution: Key-person risk is not abstract. Musk’s attention is now divided across SpaceX, Tesla, DOGE, X, and xAI. The AI segment’s acquisition was executed four months before this IPO, which means public investors are buying into a strategic pivot that has not yet been tested at scale. And the bridge loan refinancing, the Cursor acquisition option, and the Terafab build are all items that require capital allocation decisions in the near term — decisions that will be made without meaningful public shareholder input.

IX. The Lock-Up Schedule and What It Means for Volatility

Lock-up provisions determine when pre-IPO shareholders can sell. SpaceX’s lock-up schedule is structured as a series of rolling tranches — not a single 180-day cliff — which distributes the supply pressure but also creates known dates of incremental selling.

The practical implication: supply will enter the market progressively beginning within weeks of Q2 earnings, which arrive roughly 60 days post-IPO. The Nasdaq 100 inclusion demand and the rolling lock-up structure create a specific kind of technical environment — strong mechanical buying early, followed by steady insider supply over the following six months. This is not a reason to avoid the stock, but it is a reason to think carefully about entry timing and position sizing in the first two quarters of trading.

X. Bottom Line

SpaceX’s IPO is one of the most interesting offerings in years, for reasons that go well beyond the hype. Starlink is a real, compounding business with extraordinary growth, improving margins, and a structural moat that no competitor can replicate on a short time horizon. The Space segment is a leveraged call option on Starship commercialization, with a stable government contract base underneath it. The AI segment is expensive, early, and enormous in ambition — and the combination of 1 gigawatt of owned compute, proprietary real-time training data, and a rapidly iterating model team is not trivially dismissible.

What the S-1 asks you to do is value two things simultaneously: a world-class operating business that is already generating $7+ billion in segment EBITDA, and a venture-scale portfolio of options on technologies that are still in development. The first part of that valuation is tractable. The second part is speculative by definition.

At approximately 94x 2025 revenue and a $1.77 trillion implied market cap, the market is pricing in significant execution on the venture-scale options. That may prove correct — the team that built Starlink from scratch has earned a higher prior probability of success than most. But investors should be clear about what they are paying for, what they own, and how much of the valuation rests on outcomes that have not yet occurred.

Bulls Say

• Starlink is the only low-latency global broadband network — a structural monopoly in 164 countries generating $7.2B in EBITDA and growing at 50%+ annually
• Starship V3 could reduce launch costs to $100–200/kg, transforming the economics of every other SpaceX business simultaneously
• 1 GW of owned compute, 350M daily posts of real-time training data, and an iterating frontier model is a durable AI competitive position
• The physical stack thesis — controlling chips, compute, and distribution — is a coherent and ambitious strategy unavailable to purely terrestrial AI companies
• Nasdaq 100 inclusion within 15 days creates mechanical demand that supports the stock near-term

Bears Say

• At 94x revenue, approximately $1.5 trillion of the valuation is attributable to an AI segment that lost $6.4B in 2025 and has no clear path to profitability in the near term
• Class A shareholders have no meaningful governance voice; 85% voting control rests with one person whose attention spans five organizations
• Starship has completed 11 flight tests but has not entered commercial service — the transformation thesis depends on execution that remains unproven at scale
• Starlink ARPU has declined consistently as the subscriber base expands internationally; margin recovery depends on V3 cost curves materializing on schedule
• Rolling lock-up tranches beginning ~60 days post-IPO create predictable supply pressure across the first six months of trading

The right question is not whether SpaceX is a great company — it is. The right question is whether it is a great investment at this price, on this timeline, with this governance structure. That is a question every investor has to answer based on their own risk tolerance, time horizon, and portfolio construction. If you are interested in a conversation about how SPCX would fit within your specific situation, I’m available to work through it.

— John McKay, CFA

This material is provided for informational and educational purposes only and does not constitute investment advice, a recommendation, or an offer or solicitation to buy or sell any security or investment strategy. The views expressed are those of the author as of the date of publication and are subject to change without notice. The author is a financial professional and may hold positions in, or manage client accounts that hold positions in, the securities discussed. Such holdings are subject to change at any time without notice. While the author strives to present information in a fair and balanced manner, no representation is made that this commentary is free from bias, and readers should be aware of potential conflicts of interest. The information presented is derived from publicly available sources believed to be reliable, but accuracy and completeness are not guaranteed. Past performance is not indicative of future results. All investing involves risk, including the possible loss of principal. This commentary does not take into account the investment objectives, financial situation, or particular needs of any specific person. No advisor-client relationship is created by the receipt or review of this material. Readers should consult with a qualified financial, legal, or tax professional before making any financial decisions. Reading this material does not create an advisor-client relationship.