SpaceX Deep-Value Research Report

Another point that deserves close attention is customer concentration. The S-1 summary shows that over the past three years, an unnamed "Customer A" contributed 25.2% / 24.2% / 20.9% of revenue in 2023/2024/2025 respectively; in addition, roughly one-fifth of revenue is tied to U.S. federal agencies. For a capital-intensive company, this concentration is both a moat and a meaningful vulnerability: if a single large customer or a key government relationship changes, the impact will be very large.

At the industry level, SpaceX is not in a "slow-growth, mature industry" but in two arenas that are expanding rapidly while carrying very high barriers and risks: space launch/defense aerospace and LEO satellite connectivity. The U.S. FAA has already approved SpaceX to raise its Starbase cadence to 25 Starship launches per year; the official Starship page and update page further define it as a future fully reusable platform that will take on larger Starlink deployment and lunar/Mars missions. This shows that demand on the industry side is strong, but it also shows that the industry remains fundamentally dependent on technological breakthroughs, permitting, and capital investment.

The competitive landscape is not simple. On the connectivity side, Amazon Leo states in its official shareholder letter a mid-2026 commercial launch, and has already secured commitments from enterprises and governments including Delta, JetBlue, AT&T, Vodafone, DIRECTV Latin America, NBN, and NASA; Europe's leading alternative is OneWeb/Eutelsat, whose official materials claim 600+ LEO satellites with continued satellite orders to fill out the constellation; on the direct-to-cell front, although AST SpaceMobile had only 70.9 million dollars of revenue in 2025, it is already deeply tied to partners such as Verizon and still represents a potentially disruptive force. On the launch side, Blue Origin, ULA, and Rocket Lab are all catching up in different niches. In short, SpaceX is currently first in the field, but the field is not without challengers.

Taken together, if the question "is this a business I can understand" refers to the core SpaceX/Starlink, my answer is yes, though no longer as cleanly as a few years ago; if it refers to the entire consolidated entity about to go public, the answer is only partially understandable. My scores are therefore:

The core judgment behind this is: if the stock market closed for five years, I would be willing to own the core SpaceX/Starlink; but I would not be willing to casually hold this "consolidated public composite" at today's reference valuation.

Moat, Management, and Capital Allocation

First, the moat. On brand strength, SpaceX is no longer merely a "famous rocket company" but an infrastructure-grade player in U.S. aerospace and commercial satellite internet. The official Falcon User's Guide shows that by early 2025, the Falcon family had completed 430+ launches and 384 booster reflights, with a very strong reuse reliability record; the official update page also describes SpaceX as "the world's leading launch services provider." This brand was not built by advertising but accumulated through sustained mission success and critical national-level programs.

Cost advantage and scale advantage are its hardest moat. Falcon's reuse, vertical integration, and high launch frequency let SpaceX both drive down unit launch costs and quickly send its own Starlink satellites to orbit at near-internal cost, forming a flywheel of "launch capability → satellite deployment → connectivity revenue → reinvestment in launch." Starlink in turn supports the capital needs of the rockets and Starship. This flywheel is precisely the hardest thing for most competitors to replicate: launch companies lack Starlink's cash flow, satellite companies lack SpaceX's launch costs, and telecom companies lack SpaceX's orbital assets.

Network effects and switching costs exist for Starlink, but they are not the classic internet-platform network effect; rather, they are an infrastructure network effect of "scale → coverage → experience → more users → stronger scale." More satellites, more spectrum, more ground stations, and more enterprise partnerships improve coverage and availability, which in turn brings more customers. Official materials show that in 2025 it added 4.6 million+ active customers, completed deployment of its first-generation Direct-to-Cell constellation, and served 155+ markets. Once enterprise, government, aviation, and maritime customers complete terminal deployment, network integration, and operational changeover, switching is not easy either.

Licensing, regulatory, and credential barriers are equally deep. Commercial launch licenses, spectrum coordination, FAA/FCC approvals, NASA commercial crew certification, and Space Force/NRO security credentials cannot be instantly replicated by throwing capital at them. The Space Force awarding the larger share of NSSL Phase 3 Lane 2 to SpaceX shows that it holds an institutionalized advantage on the hardest national-security missions; and the FAA's approval to raise Starbase to 25 launches per year reinforces its future capacity ceiling.

Looking at the moat alone, I would give 4.5/5. But once the question expands to "is management trustworthy and is capital allocation rational," the score must drop noticeably. At the operating level, Gwynne Shotwell's team has performed exceptionally well over a long period; but as the governance object of a public company, Musk's control arrangements are clearly unfriendly to outside shareholders. Reuters reported that after listing Musk will retain 85.1% of voting power; SpaceX will also hold "controlled company" status and will not need a majority of independent directors; the charter and Texas organizational arrangements further layer on mandatory arbitration, litigation restrictions, and similar provisions. More extreme still, Reuters states directly: only Musk himself can remove Musk as CEO/Chairman. This is no longer an ordinary "dual-class structure" but an extreme minority-shareholder-disadvantaged structure.

On capital allocation, my view is: excellent in the early and middle stages, clearly worse of late. The excellence lies in the launch business's cash flow long serving reusable rockets, the Starlink constellation, and critical government programs, all in tight strategic alignment; this is very much what a truly long-term-minded owner would do. The deterioration lies in capital being drawn at scale toward AI compute and related infrastructure after the absorption of xAI / X. In 2025 total capital expenditure was about 20.74 billion dollars, of which the AI segment took 12.727 billion dollars, about 61%; in Q1 2026 total capital expenditure was about 10.1 billion dollars, of which AI took roughly 7.7 billion dollars, about 76%. This means most of the money Starlink earns must now first feed AI. For ordinary shareholders, this is not a free "option" but a high-intensity, low-certainty reinvestment.

In addition, Musk's cross-company resource shuffling further discounts the transparency of capital allocation. Reuters reported that SpaceX/xAI bought roughly 650 million dollars of goods from Tesla in 2025, and Tesla once participated in an investment with close to 19 million Class A shares; meanwhile, the S-1 disclosed more than 20 billion dollars of debt/obligations related to AI infrastructure leases. Add the Anthropic compute contract of 1.25 billion dollars per month, which sounds enticing but which the contract allows to be terminated on 90 days' notice, and that is not the same as stable, permanent cash flow. In other words, capital allocation is not only aggressive but also highly cross-dependent with Musk's entire business empire.

My scores for this part are therefore:

My overall judgment can be put in one sentence: SpaceX is a first-rate operating entity, but for outside shareholders it is not a first-rate governance asset.

Financial Quality and Owner Earnings

First, the high-certainty, already-public key numbers. In 2025, on a consolidated basis SpaceX had revenue of 18.674 billion dollars, adjusted EBITDA of 6.584 billion dollars, and a net loss of 4.94 billion dollars; Q1 2026 revenue was 4.694 billion dollars, with an operating loss of 1.943 billion dollars and a net loss of roughly 4.276 billion dollars. As of the end of March 2026, the company's principal debt was about 29.1 billion dollars; as of the end of 2025, a Reuters summary shows company assets of about 92 billion dollars and liabilities of about 50.8 billion dollars, while by March 31, 2026, the accumulated deficit had grown to 41.31 billion dollars. Together these numbers point to one conclusion: it does not lack scale or financing capacity, but it is still a long way from being a "stable, distributable free-cash-flow machine."

Key Financial Overview

The numbers in the table are compiled from sources including the public S-1 summary, Reuters, Barron's, and MarketWatch; among them, the full 2025 cash flow statement, depreciation and amortization, interest expense, working capital changes, and precise net debt cannot be reliably reconstructed from currently accessible public summaries, so I do not fabricate them.

If you look only at 2025's "revenue growth" and "adjusted EBITDA," it is easy to reach an overly optimistic conclusion. But capital expenditure quickly distorts that optimism. In 2025, Capex / Revenue was about 111%; in Q1 2026 this ratio was roughly above 215%. This is not a company with "high-quality profit and stable cash flow" in the traditional sense, but a company still trading astonishing capital intensity for future share and future optionality.

More importantly, the sources of profit are deeply split. In 2025 Connectivity EBITDA was about 7.168 billion dollars, for a rough EBITDA margin of about 63%; the Space segment's EBITDA was -351 million dollars, a rough margin of about -8.6%; the AI segment's EBITDA was -6.355 billion dollars, a rough margin of about -199%. This means the group's "accounting loss" is not because Starlink is failing, but because Starlink is currently transfusing blood to the entire composite. If you believe AI/X can ultimately become high-return, this can still be defended; but if you are a conservative long-term owner, you must acknowledge: today's stock does not equal today's Starlink.

Now the balance sheet. What can be publicly confirmed is: as of the end of March 2026, the company's principal debt was about 29.1 billion dollars; Reuters also reported that, for pre-IPO consolidation and refinancing, the company once swapped multiple older debts with a bridge loan, and that AI-related infrastructure lease obligations are large. In other words, the company's financing flexibility remains strong, but the financial structure has begun to grow complex, and it depends heavily on external capital markets continuing to believe Musk's long-dated story. A balanced, conservative-leaning investor must stay alert here: solvency risk may not be high, but the risk of returns being diluted by capital structure and devoured by future financing is not low.

On equity and shareholder-return policy, at least for now there is no typical mature-company framework of "dividends, buybacks, deleveraging." On the contrary, what the S-1 exposes is continued large-scale reinvestment, continued concentrated control, and highly aggressive incentive arrangements. Business Insider reported that Musk could also receive up to 1 billion shares in award shares after meeting extremely ambitious targets. Since the final number of shares offered has not been disclosed, the scale of dilution cannot be precisely quantified for now, but the direction is already clear: minority shareholders now play the role of "passively co-investing in Musk's grand reinvestment plan."

Owner Earnings Assessment

Following Buffett's owner-earnings approach, the formula is not complex: net profit + non-cash charges − maintenance capital expenditure − additional working capital absorbed. The problem is that in SpaceX's current public summaries, non-cash charges and working capital data are not fully available, and maintenance capital expenditure is extremely hard to split out precisely, because Starlink constellation refresh, AI data centers, Starship R&D, and ground infrastructure are all intertwined. Precisely because of this, I would rather be conservative than pretend to great precision.

My conservative assessment is: the consolidated SpaceX's owner earnings in 2025 were negative, or at least not clearly positive. The reason is direct: net profit was already -4.94 billion dollars; even adding back a portion of non-cash charges, in the face of 20.74 billion dollars of capital expenditure, especially with AI taking the bulk, the final distributable cash is most likely still negative. In other words, in discussing this company today, you cannot use "adjusted EBITDA is very high" as a substitute for "owner earnings have been established." At this stage the two are not the same thing.

If you look only at the core Connectivity business, you can reach a friendlier conclusion. In 2025 Connectivity EBITDA was 7.168 billion dollars; on a conservative assumption that its maintenance capital expenditure and maintenance working capital needs together total about 3 billion to 5 billion dollars, its normalized owner earnings could be on the order of 2 billion to 4 billion dollars. But I must stress: this is only an inference about "core Starlink economics," not an owner-earnings assessment of the "common stock of the public composite." Because public shareholders must simultaneously carry the Space and AI portions.

This produces a glaring valuation fact: if you compare core Starlink's 2 billion to 5 billion dollars of normalized owner earnings against the 1.25–1.75 trillion dollar reference valuation, the corresponding multiple is probably still somewhere between 250x and 583x; if you use the consolidated group's owner earnings, it is currently even hard to assign a meaningful positive multiple. This is generally not a value investor's comfort zone.

My conclusion for this section is simple: today's SpaceX derives its real economic value mainly from Starlink; yet today's SpaceX stock has to pay for a large amount of high-uncertainty capital investment beyond Starlink.

Valuation, Intrinsic Value, and Margin of Safety

First, a premise that must be stated clearly: the S-1 has not yet disclosed the final number of shares offered or the offering price, so at this stage a "fair buy price" can only be expressed as a total company equity valuation and cannot be responsibly converted into a per-share price. This is not evasion but being faithful to the facts.

Discounted Owner Earnings Method

Since current consolidated owner earnings are most likely still negative, I cannot run a DCF by simply extrapolating 2025 free cash flow. What I use is a two-stage, conservative model based on normalized owner earnings: first assume that AI/X losses narrow over 3 to 5 years and that Space/Starship commercialization improves, then discount the owner earnings that enter a stable phase around 2030 back to today. The key here is not computing to the decimal but seeing under what assumptions 1.25–1.75 trillion dollars would be reasonable.

Note: the most optimistic scenario in this table is already quite aggressive, because it implies that AI/X no longer merely devours Starlink's profits but begins to contribute high-quality cash flow, and that Starship commercialization achieves a significant breakthrough. Even so, the upper bound of the optimistic intrinsic value still falls below most of the reference points within 1.25–1.75 trillion dollars. In other words, to rationalize the current market target valuation, what you need is not "normal excellent execution" but "near-perfect execution + high returns delivered by new businesses."

Relative Valuation Method

On the public reference valuation, SpaceX's valuation is very aggressive. At a 1.75 trillion dollar valuation, the P/S against 2025 revenue is about 93.7x, and the multiple against 2025 consolidated adjusted EBITDA is about 265.8x; even at a 1.25 trillion dollar valuation, P/S is still 67x and EV/EBITDA is nearly 190x. This is a classic "dream price," not a price anchored to current cash flow.

A rough comparison with public peers makes this more intuitive. Rocket Lab's current market cap is about 81.298 billion dollars, with 2025 revenue of 601.8 million dollars, for a trailing P/S of about 135x; AST SpaceMobile's market cap is about 26.04 billion dollars, with 2025 revenue of 70.9 million dollars, for a trailing P/S of about 367x; Iridium's market cap is about 4.825 billion dollars, with 2025 revenue of 871.7 million dollars and OEBITDA of 495.3 million dollars, for a P/S of about 5.5x and a market cap / OEBITDA of about 9.7x. These comparables tell us two things: first, aerospace growth stocks are inherently expensive; second, everyone in the peer group being expensive does not prove SpaceX is cheap. In particular, Iridium represents the pricing floor of an "already-commercialized, already-stably-profitable satellite connectivity asset," while Rocket Lab / ASTS represent the pricing ceiling of "high-volatility, highly speculative growth assets." SpaceX's pricing today is closer to the latter than the former.

Asset Value and Liquidation Value Method

Viewed through liquidation or book assets, SpaceX is even further from cheap. The Reuters summary shows end-2025 assets of about 92 billion dollars and liabilities of about 50.8 billion dollars, implying book equity of roughly 41.2 billion dollars; by March 31, 2026, the accumulated deficit had reached 41.31 billion dollars. At the same time, principal debt at the end of March 2026 was about 29.1 billion dollars, plus large AI infrastructure-related leases/obligations. Given that much of its assets consist of satellites, rockets, test facilities, specialized equipment, software, and highly customized/classified programs, liquidation value is neither high nor easy to realize. This means: book assets offer almost no protection against the current trillion-dollar narrative.

My Valuation Conclusion

If I translate the "margin of safety" you asked for into value-investing language, then my answer is: this is not "a good company on sale" but "a good company priced at its long-dated ideal state, or even higher." In this situation, the most fragile assumption is not "whether revenue can keep growing," but: whether AI investment can ultimately produce high-return cash flow; whether Starship can be delivered on a commercial cadence rather than a test cadence; and whether minority shareholders can share in the results. If any one of these three assumptions fails, the valuation cannot stand.

Risks, Counterarguments, and Comparison Benchmarks

Risks and the Strongest Bear Case

If I take the strongest bear's stance, the opposing logic is in fact very powerful: you think you are buying "the strongest commercial space company," but what you actually buy is "a composite supported by Starlink's cash flow and used to finance Musk's grander AI/Mars narrative." This composite has an extremely concentrated governance structure with constrained minority rights; its most profitable segment requires continuous heavy-capital maintenance; its newest business burns the most cash and has the least provable returns; and yet the valuation has already priced in the best long-term vision.

The specific risks can be grouped into several categories. On competitive risk, Amazon Leo plans to launch in mid-2026 and has already secured multiple large enterprise/government commitments; OneWeb/Eutelsat is Europe's explicit LEO alternative; and AST SpaceMobile offers a different path in direct-to-cell. On technology substitution and execution risk, Starship is still in testing and on the eve of ramp-up, and whether it can truly cut the cost curve further depends on the next few years rather than the next few test flights. On regulatory risk, launch licensing, environmental litigation, spectrum coordination, and AI content regulation are not low-probability noise. On financial risk, debt, lease obligations, and ultra-high capital expenditure are all raising the return hurdle. On customer concentration risk, Customer A accounts for more than 20% of revenue, and government-related revenue is a relatively high share. On governance risk, this is the point almost every traditional institutional investor will worry about: outside shareholders have almost no ability to correct the board, the CEO, or capital allocation.

Which facts would I see as overturning the current judgment? First, if over the next two to three years the AI segment can rapidly turn from huge losses into high-quality, repeatable positive free cash flow, my valuation midpoint would move up noticeably. Second, if Starship can truly achieve high-frequency, reliable, low-cost commercial reuse, turning the Space segment from negative EBITDA into a strong cash-flow platform, the trillion-dollar story would rest on firmer ground. Third, if the company makes substantive improvements to outside-shareholder rights, such as reducing the concentration of control, raising board independence, and softening mandatory arbitration and restrictive provisions, the governance discount would shrink. Conversely, the moment any combination of an unstable Anthropic compute contract, continued out-of-control AI cash burn, significant Starship delays, the loss of Customer A, or escalating regulatory shocks appears, I would lean toward "avoid rather than watch."

The biggest scenario for permanent capital loss is also clear: investors enter at the 1.25–1.75 trillion dollar story valuation, but 5 to 10 years later the market finds that the group's real owner earnings never catch up with the imagination, and the valuation midpoint falls back to 300 billion to 500 billion dollars. In that case, even if the enterprise itself does not fail, shareholders could endure a long-term capital drawdown of 60% to 80%+. This matters more than short-term volatility.

Comparison with Other Opportunities

Compared with its strongest rivals, SpaceX is still the strongest on current operating quality. Amazon Leo has cloud, channel, and enterprise-customer synergies, OneWeb has government-enterprise and European-sovereignty logic, and AST has direct-to-cell imagination, but in launch capability, in-orbit scale, depth of government programs, and the overall industrial loop, SpaceX remains the strongest asset for now. I do not dodge this point.

But compared with a broad-based index, the conclusion is entirely different. The S&P 500 represents 500 large U.S. companies and covers roughly 80% of the investable market cap; and SPY is currently priced at about $741.25. The index's advantage is not that it is sexier, but that its diversification, cash-flow visibility, and governance repairability are far higher than for a founder-extreme-control asset like SpaceX. For a balanced, conservative-leaning investor, buying SpaceX today is not "clearly superior" to buying the index.

Now compare it with the risk-free yield. FRED shows that the U.S. 10-year Treasury yield was about 4.67% on May 19, 2026. To make a conservative investor give up this risk-free yield and instead bear SpaceX's execution, regulatory, capital-expenditure, and governance risks, the reasonable required expected annualized return should be at least a double-digit range clearly above 4.67%. The trouble is: if you buy in at a 1.25–1.75 trillion dollar reference valuation, then even if the company's market cap reaches 1.6 trillion to 3.0 trillion dollars over the next 10 years, the annualized return is only low-single-digit to high-single-digit; and if 10 years later it can only land near 0.8 trillion, the annualized return is negative. As risk compensation, this is not worthwhile.

If I could hold only 5 assets, I would not include it today. It is very much worth continuing to study, but it is not worth rushing to tie up capital at the current reference valuation.

Checklist and Final Conclusion

Investment Checklist

Final Investment Conclusion

【Final Rating】 Watch

【One-Sentence Investment Thesis】 SpaceX's core business is extraordinarily strong, but at the current public reference valuation what ordinary shareholders buy is a combination of "high-quality assets + high-intensity cash burn + extreme control rights," rather than a value opportunity with a margin of safety.

【Core Bull Case】 SpaceX still has the world's strongest integrated commercial space capability, with a moat that comes from reusable launch, orbital scale, Starlink's recurring revenue, and deep government ties. Starlink has become the group's economic engine, contributing 11.387 billion dollars of revenue and 7.168 billion dollars of EBITDA in 2025. The company simultaneously holds positional advantages in NASA, the Space Force, the NRO, and commercial connectivity, which no single competitor can easily replicate. If Starship is successfully commercialized and AI/X ultimately turns into a high-return business, the long-term upside remains enormous.

【Core Bear Case】 The current listing entity is not "pure SpaceX/pure Starlink" but a composite whose visible cash-flow quality has been heavily diluted by AI/X; the 2025 net loss was 4.94 billion dollars and the Q1 2026 net loss was about 4.276 billion dollars. Capital expenditure is too large, with 2025 Capex / Revenue above 111% and Q1 2026 above 215%. Musk will retain 85.1% of voting power, leaving minority-shareholder protection weak. The reference valuation of 1.25–1.75 trillion dollars leaves almost no room for error against current owner earnings.

【Key Assumptions】 For the investment to hold, at least four conditions must be met: first, Starlink keeps expanding scale while maintaining high gross margins and high cash conversion; second, Starship turns within the next 3 to 5 years from an R&D platform into a true commercial platform that lowers costs and lifts revenue; third, AI/X stops devouring Starlink's profits over the long run and begins contributing positive free cash flow; fourth, governance risk does not deteriorate further to the point of clearly harming outside shareholders.

【Fair Buy Price】 Since the number of shares offered and the offering price have not yet been disclosed, I can only give an equity valuation range rather than a per-share price. My ideal buy range is 300 billion to 500 billion dollars; at 500 billion to 800 billion dollars, it enters the "can hold but uncomfortable" range; above 1 trillion dollars, I would consider it clearly overvalued; and if the final pricing lands at 1.25–1.75 trillion dollars, I would choose not to chase the IPO price.

【Target Holding Period】 This can only be considered in units of 10 years or more, and only on the premise of a lower buy valuation and a clearer cash-return path. At today's price, it is not a question of holding period but a question of starting price.

【Expected Annualized Return】 If buying at the current reference valuation, I would see it this way: In the conservative scenario, 10 years out the company only proves itself worth about 0.8 trillion dollars, in which case the annualized return is most likely negative; In the neutral scenario, 10 years out it reaches about 1.6 trillion dollars, in which case buying at 1.25 trillion yields only roughly low-to-mid-single-digit annualized returns, and buying at 1.75 trillion may still be close to 0 or negative; In the optimistic scenario, 10 years out it reaches about 3.0 trillion dollars, and only then does the annualized return roughly enter the high single digits. For such a high-risk asset, this is not ample. The numbers here are scenario assumptions, not company guidance.

【Maximum Loss Risk】 In the worst case, if the market ultimately prices this company at only 300 billion to 500 billion dollars rather than a trillion dollars, then investors who bought at the 1.25–1.75 trillion dollar reference valuation could face a long-term market-cap loss of 60% to 80%+. This does not require the company to go bankrupt; it only requires "narrative valuation reverting to cash-flow valuation."

【Tracking Indicators】 The indicators most worth tracking going forward are: Connectivity segment revenue and EBITDA, Starlink active users and ARPU, changes in the Customer A share, Starship test-flight success rate and reuse cadence, total Capex and the AI Capex share, principal debt and lease obligations, AI segment revenue and the pace of loss narrowing, the renewal/termination status of large contracts such as Anthropic's, the final number of shares offered and potential dilution, and whether governance provisions are further tightened.

【Signals That Trigger Re-Evaluation】 If any of the following appears, I would re-evaluate immediately: AI cash-burn speed keeps rising rather than falling; Starship suffers repeated major delays or regulatory blocks; the Customer A or key government-contract share worsens; the final offering valuation is significantly above 1.75 trillion dollars; minority-shareholder rights are further weakened; or Starlink ARPU/retention worsens beyond expectations.

【Final Recommendation】 Put coolly, this is not "the company is bad" but "the price and structure are wrong." If you view this investment as a long-term business owner rather than an emotion-driven chaser, then the most rational action today is not to rush in but to keep tracking: wait for a lower valuation, wait for clearer free cash flow, wait for a more acceptable governance discount. A great enterprise does not automatically equal a great stock.

Open Questions and Limitations Currently accessible public summaries are insufficient to reliably reconstruct the full 2025–Q1 2026 cash flow statement, including depreciation and amortization, working capital changes, interest coverage, precise net debt, precise ROIC, and the complete equity-dilution path; this report therefore applies explicitly conservative normalized assumptions to Owner Earnings and DCF. If a later S-1/A, roadshow materials, or the formal offering range disclose this information, the valuation range should be updated again.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.