SpaceX (SPCX) S-1 Analysis: Largest IPO in History at $1.75T — Initiating at Hold on Sum-of-Parts Math vs. Indicated Valuation

Filing Snapshot

SpaceX's public S-1 arrived today. It presents SpaceX, xAI and X as one company for every period shown, which the S-1 calls a retrospective combination of entities under common control. This is restated history, not a pro forma. FY2025 revenue was $18,674M, up 33.2%. Adjusted EBITDA was $6,584M, a 35.3% margin. The operating loss was $(2,589)M and the net loss $(4,937)M. FY2024 had been profitable, with operating income of $466M and net income of $791M, so the 2025 loss is a reversal rather than a continuation. Q1 2026 revenue was $4,694M, up 15.4%, with a net loss of $(4,276)M. The accumulated deficit reached $41,311M at March 31, 2026.

Implied Valuation Multiples at IPO Target

Revenue Assessment

FY2025 revenue of $18,674M grew 33.2%. Space and Connectivity together produced $15,473M, up 35.8% from $11,395M. The AI segment added $3,201M, up 22.2%, but most of that was X advertising ($1,844M); AI Solutions & Infrastructure revenue was $1,357M, up from $892M. Connectivity alone was $11,387M, up 49.8%, and 61.0% of the total. Within Space, Launch Services revenue was flat at $2,576M against $2,584M: the S-1 says customer launches and the average price per launch were roughly flat. Launch & Development revenue rose to $1,510M on NASA cargo missions and a U.S. Department of War contract. Q1 2026 revenue of $4,694M grew 15.4%, slower than 2025, as Connectivity's $782M gain was partly offset by a $246M decline in Space revenue. We model FY2026 revenue at $25-27B (+34-45% YoY), which requires growth well above Q1's pace. That estimate was set without the Anthropic compute contract, which ramps in May and June 2026: if it reached full rate by July, it alone would add about $7.5B in the second half and put FY2026 above our range.

Profitability Assessment

The segment results show where the profit is made. Connectivity earned $4,423M of operating income in 2025, a 38.8% margin, and $7,168M of Segment Adjusted EBITDA. Space lost $(657)M at the operating line after $3,004M of R&D that the S-1 says mainly relates to Starship; before all Space R&D, the segment earned roughly $2.3B, and its Segment Adjusted EBITDA was $653M. The AI segment lost $(6,355)M at the operating line and $(1,237)M at the Segment Adjusted EBITDA line. Space and Connectivity together earned $3,766M of operating income, so the consolidated operating loss is an AI story. The trend is mixed. Connectivity's operating income more than doubled in 2025, but the consolidated Adjusted EBITDA margin fell from 38.2% to 35.3%, and in Q1 2026 it fell to 24.0% from 42.5% a year earlier. The path to consolidated GAAP profitability runs through (a) continued Starlink scaling, (b) AI revenue catching up with its cost base, and (c) Starship moving from R&D to paying launches.

Cash Burn Assessment

Operating cash flow was $6,785M in 2025 against capex of $20,737M, a gap of $(13,952)M. The gap widened in Q1 2026 to $(9,060)M in a single quarter, as capex reached $10,107M and AI took $7,723M of it. At March 31, 2026 the company held $15,852M of cash and $7,823M of short-term marketable securities, and it had drawn the full $20,000M bridge loan. The indicated $75B raise is therefore not $75B of new money: an amount equal to the net proceeds must be applied to the bridge loan within six months, which leaves roughly $55B. Together with cash and securities, that is about $79B, enough for roughly five and a half years at 2025's burn rate, or a little over two years at Q1 2026's pace. We model consolidated free cash flow reaching break-even by FY2028-2029, but that estimate assumed capex near 2025's level, and Q1 2026 ran at about twice that rate.

Corporate Structure & xAI Consolidation

The most consequential corporate event before the IPO is SpaceX's acquisition of xAI, which had itself acquired X in March 2025. The accounting for these deals shapes every number in the S-1.

The Deal Mechanics

SpaceX completed its acquisition of X.AI Holdings Corp. on February 2, 2026, and xAI became a wholly-owned subsidiary. Each xAI common share converted into 0.1433 SpaceX shares on a pre-split basis, and eligible service providers could elect $75.46 per xAI share in cash instead. In total, SpaceX issued 321.7M Class A shares and 121.7M Class B shares before its five-for-one stock split of May 4, 2026, and paid $2,947M in cash. The deal was mostly stock, but not entirely.

The widely reported marks for the deal, about $250B for xAI and about $1T for SpaceX, do not appear in the S-1. We use the $250B figure as a reference point for the AI segment, not as a filed number. The S-1 does give one price marker: the cash alternative of $75.46 per xAI share, against 0.1433 SpaceX shares, implies about $527 per pre-split SpaceX share, or about $105 after the split. The $1.75T indication is about $140 per share on the S-1's share count, a step-up of about a third in three and a half months. Against the combined reported marks of about $1.25T, the step-up is about 40%. Public investors will need to validate it.

Accounting Treatment in the S-1

• Restated history, not a pro forma. The mergers were accounted for as reorganizations of entities under common control, because Musk held a controlling financial interest in SpaceX, xAI and X. All periods combine the three companies at historical carrying amounts, so FY2025 and earlier years include xAI and X in full.
• The $(4,937)M FY2025 net loss is driven by the AI segment. The AI segment's operating loss was $(6,355)M, while Space and Connectivity together earned $3,766M of operating income.
• Capex splits the same way. The AI segment spent $12,727M of the $20,737M total in 2025 (61.4%) and $7,723M of $10,107M in Q1 2026 (76.4%). Space and Connectivity together spent $8,010M in 2025.
• No new goodwill. Because the deals were common-control reorganizations, the S-1 states that no new goodwill or other intangible assets were recorded. Goodwill was $11,809M at December 31, 2025, of which $11,296M sat in the AI segment. Impairment risk is limited to that balance.

Class A / Class B / Class C Share Structure

Musk Voting Power

As of May 1, 2026, before the offering, Elon Musk beneficially owned 849,494,440 Class A shares (12.3% of the class) and 5,569,053,075 Class B shares (93.6%), for 85.1% of total voting power. These figures include 350,000,000 Class B shares issuable on exercise of options and 1,302,072,285 restricted Class B shares. The restricted shares are 2026 performance awards. 1,000M of them vest on 15 market-value milestones from $500B to $7.5T and on the establishment of a permanent human colony on Mars with at least one million inhabitants. Another 302.1M, which replaced his xAI award, vest on milestones from $1.065T to $6.565T and on non-Earth-based data centers delivering 100 terawatts of compute a year. The S-1 leaves the post-offering columns blank, but an offering of about 4% of the company cannot bring Musk's voting power anywhere near 50%. Class B holders, voting as a class, also elect 51% of the board, and removing Musk from his board and executive roles requires a majority of the Class B vote.

That makes SPCX a "controlled company" under Nasdaq and Nasdaq Texas rules, and it may opt out of the requirements for:

• A majority-independent board of directors
• A nominating committee composed entirely of independent directors
• A compensation committee composed entirely of independent directors
• Annual performance evaluations of the compensation and nominating committees

The S-1 says the company intends to use "certain of these exemptions" and does not expect its compensation and nominating committee to be composed entirely of independent directors. The board will have eight directors, five of whom it has determined to be independent. The practical consequence: public shareholders of SPCX will own economic claims on the business but will have little ability to influence board composition, executive compensation, or strategic direction.

Segment Deep-Dive

The S-1 reports three segments: Space, Connectivity and AI. For each it gives revenue, operating income, Segment Adjusted EBITDA and capex for 2023 through 2025 and for Q1 2025 and Q1 2026, plus key operating metrics. Starlink sits in Connectivity. Starship sits in Space, where it earns no revenue: all Starship launches to date are classified as internal, and its cost runs through Space R&D. We discuss Starship separately below because it is a separate piece of our valuation.

Starlink (Connectivity): The Economic Engine ($11.4B Revenue, $4.42B Op Income)

Connectivity, which the S-1 describes as "primarily driven by Starlink", was the only segment with operating income in 2025, and it earns enough to fund much of the rest. Revenue of $11,387M grew 49.8% from $7,599M, which had itself grown 96.4% from $3,869M in 2023. Operating income of $4,423M grew 120.4%, and the operating margin reached 38.8%, up from 26.4% in 2024 and 12.1% in 2023. Segment Adjusted EBITDA grew 86.2% to $7,168M. In Q1 2026, revenue grew 31.6% to $3,257M, but operating income grew only 15.0% to $1,188M.

Unit economics moved in two directions. Subscribers rose from 2.3M at the end of 2023 to 10.3M at March 31, 2026, about 4.5 times. Subscriber ARPU fell from $99 a month in 2023 to $91 in 2024 and $81 in 2025, and to $66 in Q1 2026 from $86 a year earlier. The S-1 attributes the declines to "international expansion and the addition of lower priced service plans." Margin expanded anyway. The S-1 credits subscriber growth, enterprise adoption and "continued improvement in network efficiency". We think scale in satellite production and internal Falcon launches also helped, although the S-1 does not quantify either.

Two definitions matter. A Starlink Subscriber is a service line on a Starlink.com account, and the count covers both Personal and Business plans. Managed enterprise and government customers, such as aviation and maritime contracts, are not counted. So Subscriber ARPU is not a pure consumer figure, and the S-1 gives no enterprise ARPU. Consumer subscribers produced over 60% of Connectivity revenue in 2025: $7,208M, against $4,179M from Enterprise & Government.

The forward growth drivers fall into three buckets. Consumer broadband: the S-1 sizes the Starlink Broadband market at about $870B, and says its approximately 9,600 satellites were about 75% of all active maneuverable satellites in orbit at March 31, 2026, serving 164 countries, territories and other markets. Enterprise and government: aviation, maritime, land mobility and government contracts sit outside the subscriber count, and Enterprise & Government revenue grew from $2,769M to $4,179M in 2025. Named airline partners include United Airlines, Southwest Airlines, Qatar Airways, Lufthansa Group, British Airways, Alaska Airlines and Hawaiian Airlines. Starlink Mobile: the satellite-to-mobile service reaches about 7.4 million monthly unique devices in about 30 countries through roughly 30 carrier partners, including T-Mobile in the United States. The S-1 sizes this market at $740B, and the FCC approved SpaceX's roughly $19.6B purchase of EchoStar spectrum on May 12, 2026.

Assessment: Starlink is operating cleanly at scale, with one of the best growth-and-margin profiles in technology. The subscriber runway is long (we model 10M growing to 25-40M by FY2030 in the base and bull cases), and V3 satellites launched on Starship, which the S-1 expects to begin deploying in the second half of 2026, will add capacity per launch. The weak point is ARPU. The company expects it to keep falling, and Q1 2026 fell 23%. Our DCF below holds consumer ARPU near $80 before it recovers; that is our view, not the company's. Our standalone Starlink valuation anchors at $1.0-1.1T, which makes Starlink the majority of consolidated equity value.

Space: Launch Reliability, Flat Launch Revenue, Starship Spending

The Space segment is operationally dominant but, as reported, not profitable. In 2025 SpaceX launched 165 Falcon 9 rockets, 157 of them on flight-proven boosters. The S-1 cites an over-99% mission success rate across Falcon rockets and a first-stage booster reflown 34 times. Most launches carry SpaceX's own satellites: of the 165 Falcon launches in 2025, 122 were internal and 43 were for customers.

The launch economics are harder to read than the reliability. The S-1 gives no current price or cost per launch; its per-kilogram figures are historical: about $2,700 for the first Falcon 9 in 2010 and about $1,400 for the first Falcon Heavy in 2018. It says customer launches and the average price per launch were roughly flat in 2025, which is why Launch Services revenue did not grow. The segment's $(657)M operating loss comes after $3,004M of R&D. Before that R&D the segment earned roughly $2.3B, so Space is profitable before its development spending, and the reported loss is largely the cost of Starship.

Government exposure is large. The S-1 describes SpaceX as "the primary launch provider for the U.S. government." In 2025 it flew 11 of 12 National Security Space Launch medium and heavy lift missions and all five U.S. crew and cargo missions to the International Space Station. About one-fifth of 2025 revenue came from U.S. federal agencies, across all segments. Consolidated backlog was $28,377M at December 31, 2025, with about 32% expected to be recognized within a year. The S-1 does not split backlog by segment or by customer. Its largest customer, "Customer A", was 20.9% of 2025 revenue and spans all three segments; the S-1 does not name it.

Assessment: Launch is a reliable, contracted business with strategic national-security value. Its reported profit is suppressed by Starship development, and its revenue is capped by flat customer launch volume and pricing. The S-1 also expects Falcon 9 launches "to decrease over time" as Starship takes on more of the launch volume. If Starship works, Falcon ramps down. We value the launch business standalone at $40-50B, based on the present value of contracted work plus the ongoing launch business at a revenue multiple.

xAI and X (AI Segment): Capital Consumer, Strategic Hedge

xAI became a wholly-owned subsidiary of SpaceX on February 2, 2026. Its segment, which also contains X, is the structural counterweight to Starlink's profits. The S-1 reports it in full:

• FY2025 revenue: $3,201M, up 22.2%. Of this, $1,844M was X advertising and $1,357M was AI Solutions & Infrastructure, up from $892M.
• FY2025 operating loss: $(6,355)M. Segment Adjusted EBITDA was $(1,237)M.
• FY2025 capex: $12,727M, 61.4% of the consolidated total. AI capex was $7,723M in Q1 2026 alone, 76.4% of the quarter's total.
• Q1 2026: revenue $818M, operating loss $(2,469)M, Segment Adjusted EBITDA $(609)M.
• AI R&D: $5,064M in 2025, up from $1,176M in 2024, mainly for training Grok.
• Usage and compute: about 550 million monthly active users across Grok and X at March 31, 2026, of whom about 117 million used Grok's AI features. Nameplate compute draw reached 1 gigawatt.
• Anthropic: in May 2026 SpaceX entered Cloud Services Agreements with Anthropic for compute capacity across COLOSSUS and COLOSSUS II. Anthropic agreed to pay $1.25B per month through May 2029, with capacity ramping in May and June 2026 at a reduced fee. Either party may terminate on 90 days' notice.
• Cursor: an April 2026 compute and option agreement gives SpaceX the right to buy Cursor at an implied equity value of $60.0B, payable in Class A stock. If SpaceX terminates, Cursor is owed a $1.5B termination fee plus an $8.5B deferred services fee.

The strategic rationale for SpaceX absorbing xAI is debatable. The bull case: vertical AI integration, with Grok's models running on SpaceX compute, Starlink as a distribution layer, AI-driven operations, revenue diversification, and talent retention through the combined company. The bear case: xAI is structurally a different business (frontier AI versus space transportation and connectivity), the $250B mark is rich against realized economics, and the combination makes SPCX a more complex security that joins two distinct risk profiles.

The numbers are demanding. At the reported $250B mark, the AI segment is valued at about 78x its 2025 revenue, and about 184x the $1,357M of AI Solutions & Infrastructure revenue once X advertising is set aside. xAI competes directly with OpenAI ($500B mark) and Anthropic ($350B mark), both of which have stronger enterprise distribution, more diverse model families and lower regulatory overhangs. Against that, the Anthropic agreements alone would run at about $15B a year at full rate, more than four times 2025 AI revenue. That is real evidence that the compute can be sold externally, but a 90-day termination right makes it a weak anchor for valuation. xAI's main competitive advantage is Grok's integration with X; the practical question is whether Grok can close the capability gap with GPT-5 and Claude over the next 18 months.

Assessment: xAI is the most debatable component of the SPCX valuation. At the $250B mark we view it as fully valued in our base case. The bull case ($400-500B) requires Grok to reach capability parity with OpenAI and Anthropic; the bear case ($150-200B) reflects regulatory action and a continued capability lag. Our base case ascribes $250B to xAI in the SOTP. The swing variable is whether the Anthropic contract runs at full rate, and whether similar contracts follow, fast enough to justify capex that reached $7.7B in a single quarter.

Starship: The Multi-Decade Option

Starship is the fully reusable heavy-lift system that, if it works, turns SpaceX from a launch and connectivity company into infrastructure for activity beyond Earth orbit. The S-1 says SpaceX has invested over $15 billion in Starship. Space R&D, mainly Starship, rose to $3,004M in 2025 from $1,835M in 2024 as the company accelerated the program, and was $930M in Q1 2026. SpaceX had executed 11 flight tests at filing, with a 12th scheduled to debut the next-generation vehicle, and it expects Starship to begin delivering payloads to orbit in the second half of 2026.

The economic stakes are large. Starship V3 is designed to carry 100 metric tons in a fully reusable configuration, and later versions could reach 200 metric tons. The S-1 expects a single Starship launch to deploy up to 60 V3 Starlink satellites, a potential twenty-fold increase in downlink capacity deployed per launch compared with a Falcon 9. Starship is also NASA's Human Landing System for the Artemis program, which aims to land humans on the Moon by the late 2020s; the S-1 does not state a contract value. Beyond that sit heavy commercial payloads, space-based defense systems, and ultimately Mars logistics.

The risks are equally large, and the S-1 is explicit about the most important one:

• In-orbit refueling is not yet proven. The S-1 states that in-orbit refueling of Starship "is essential to our lunar, Mars, asteroid mining, and other deep space ambitions beyond geostationary Earth orbit." It then states: "In-orbit refueling is complex, and we have not yet demonstrated or attempted it." The company has transferred about five metric tons of propellant between tanks in space, which it describes as key data for future full-scale transfers.
• Flight testing is iterative. The S-1 frames flight tests as data collection, and states that no result, even loss of a vehicle, is considered a failure. Investors should expect vehicle losses as part of the program.
• Capital intensity. Space R&D of about $3B a year is borne before Starship earns any revenue.

Assessment: Starship is a long-dated call option. Its value depends almost entirely on whether full reusability and in-orbit refueling become operational. We ascribe $50-100B of optionality value in the SOTP ($75B base), with a bull case of $300B+ if Starship works as designed and a bear case of $0-20B if it faces fundamental architectural problems. The probability-weighted value sits in the middle of this range.

Key Topics from the S-1

1. The Largest IPO on Record: Scale and Demand Dynamics

At the reported $75 billion raise on a $1.75 trillion valuation, SPCX would be the largest IPO ever, roughly 3x the size of Saudi Aramco ($25.6B raise) and Alibaba ($25B raise). The S-1 leaves the price, share count and offering size blank, so these are indicated terms, not filed ones. The S-1 does say that "a number of shares of our Class A common stock are expected to be allocated to retail investors in this offering", and it names Charles Schwab, Fidelity, Robinhood, SoFi and E*TRADE as retail channels, but it gives no percentage. A directed share program for employees and designated individuals is also planned, with its size left blank. The retail emphasis reflects SpaceX's public profile, demand from pre-IPO vehicles, and an apparent wish to broaden the shareholder base.

Assessment: A large retail allocation is double-edged. It broadens the shareholder base and creates a group of small holders who may hold through volatility. But retail-heavy deals have historically been more volatile around lock-up releases, and the S-1 itself warns that high retail interest may increase volatility. We expect the first 90-180 days of trading to be unusually volatile, with daily swings of 4-8% (vs. 1-2% for a typical large cap).

2. The $20B Bridge Loan

In March 2026 SpaceX drew a $20,000M unsecured bridge term loan. It used the proceeds to repay X and xAI debt (the X B-1 and B-3 term loans, two xAI term loans and xAI's 12.5% senior secured notes), a total of $18,905M including a $1,163M prepayment penalty. The loan matures on September 2, 2027, and can be extended to March 2028. The key term for IPO investors: the company must apply an amount equal to the net proceeds of a qualified IPO, including this one, to repay the loan within six months of receiving them. Net proceeds of certain debt financings must also go to repay it. The S-1's use-of-proceeds section does not mention the loan; the requirement sits in the debt disclosure.

Assessment: The bridge replaced X and xAI debt, including 12.5% notes, with a single SOFR-based SpaceX loan. That is a financing improvement, not a sign of distress. But it means about $20B (27%) of the indicated $75B raise is committed to debt repayment unless the loan is refinanced first, and the S-1 says the company may refinance it with notes or bank borrowings. Investors should treat the effective new capital from the IPO as closer to $55B than $75B.

3. Regulatory and Legal Exposure

The S-1 records a $530M accrual for litigation losses that are probable and reasonably estimable at December 31, 2025, falling to $399M at March 31, 2026. The accrual is company-wide and covers only losses that are probable and reasonably estimable; the S-1 does not allocate it to xAI or to any single matter. The disclosed matters are concentrated in X and the AI segment:

• EU Digital Services Act: in December 2025 the European Commission fined X entities, x.AI and Musk EUR 120 million over X's blue checkmark, advertisement repository and researcher data access. The parties challenged the decision in February 2026.
• Patent: a jury awarded Vidstream $105M against Twitter, and the court added $67M of prejudgment interest; both sides have appealed.
• Grok: in February 2026 the Irish Data Protection Commission opened a large-scale GDPR inquiry into generative AI functionality tied to Grok on X. The FTC is examining how the AI segment and other large technology companies evaluate the safety of chatbots acting as companions to children and teens. Several lawsuits over Grok's image-generation features have been filed since January 2026, including a suit by the City of Baltimore.
• Environmental: the NAACP has sued under the Clean Air Act over the turbines powering the COLOSSUS II data center, and on May 6, 2026 moved for a preliminary injunction.

Assessment: The accrual is a floor, not a ceiling: it covers only matters the company can estimate. The larger risk is regulatory action rather than direct liability. A GDPR or DSA finding could restrict Grok or X features in the EU, and the S-1 notes that loss of access to certain markets "has occurred in the past". The bull case treats this as ordinary AI and platform friction; the bear case treats it as an overhang specific to Grok and X. Our base case ascribes $1-3B of cumulative legal and regulatory cost over the next 24 months, which is manageable at SPCX's scale but a structural drag.

4. Musk Voting Concentration and the Controlled-Company Exemption

The dual-class structure gives Musk 85.1% of the voting power before the offering, and an offering of about 4% of the company will not change that materially. Class B holders elect 51% of the board, and only a Class B majority can remove Musk from his roles. The company will use controlled-company exemptions from the Nasdaq independence requirements for its compensation and nominating committee. Public Class A shareholders will have little influence on board composition, executive compensation or strategic direction. The S-1 also discloses related-party AI hardware financing from Valor, whose founder, director Antonio Gracias, is not among the independent directors. One Valor transaction was recorded in early 2026 as $5,365M of debt under failed sale-leaseback accounting, and an April 2026 lease with Valor carries $6,587M of payments.

Assessment: The governance concentration is among the most aggressive brought to US public markets at this scale. Two interpretations: (a) a "founder premium", where Musk's vision and execution justify single-leader control; (b) a "governance discount", where concentrated control weakens shareholder protections and adds key-person risk. The S-1 itself says the company is "highly dependent" on Musk, holds no key-person life insurance on him, and notes that he does not devote his full time to SpaceX. Our SOTP applies a governance discount of roughly 9-10% of pre-discount value. The market will likely price this discount unevenly: less in euphoric periods, more under stress.

5. Starlink ARPU Compression: Volume vs. Value

Starlink Subscriber ARPU fell 18% from $99 a month in 2023 to $81 in 2025, while subscribers rose from 2.3M to 8.9M. The decline then steepened: Q1 2026 ARPU was $66, down 23% from $86. The S-1 attributes the declines to international expansion and lower-priced plans. It says the company generally expects Subscriber ARPU "to continue to decline over the next few years" as more of its subscribers sit outside North America and it adds lower-priced plans. This is a deliberate volume strategy. The company says it prioritizes revenue growth and margins over ARPU.

Assessment: The compression is real and guided to continue, but volume has more than offset it: Connectivity revenue grew 194% over the two years to 2025. Beyond subscribers, the offset is enterprise and government revenue, which the subscriber metric excludes and which grew 51% in 2025. Our DCF holds consumer ARPU at $80-83 through FY2028 and lets enterprise mix lift the blended figure. The consumer path already sits above Q1 2026's $66, and it is the most exposed assumption in our Starlink valuation.

6. Falcon 9 Reliability Track Record: A Strategic Asset

The S-1 cites an over-99% mission success rate across Falcon rockets, about 650 orbital launches through March 31, 2026 (over 540 of them on flight-proven rockets), and a first-stage booster reflown 34 times. Since 2023, SpaceX has launched more than 80% of the world's mass to orbit each year. In 2025 it flew 11 of the 12 National Security Space Launch medium and heavy lift missions.

Assessment: Reliability is real and durable. Operational data, refined procedures and proven reuse create a flywheel that competitors (Blue Origin, Rocket Lab, ULA) cannot easily copy, and it underpins SpaceX's role as the U.S. government's primary launch provider. The S-1's own framing puts the core advantage in "launch costs and cadence" rather than reliability alone. We think investors focus on Starlink growth and Starship optionality and underrate how much of the business rests on Falcon's record.

7. In-Orbit Refueling: Starship's Binding Constraint

The S-1 calls in-orbit refueling essential to its lunar, Mars and deep-space ambitions, and states plainly that the company has not yet demonstrated or attempted it. Refueling is not required for Starship's low-Earth-orbit programs, such as V3 Starlink deployment. It is required for the lunar work, including Starship's role as NASA's Human Landing System for Artemis, a program that aims to land humans on the Moon by the late 2020s. The S-1 gives no lunar landing date and no count of tanker flights per mission.

Assessment: Refueling is the most concrete execution risk in the S-1. If it proves harder than planned, the lunar timeline and any deep-space revenue shift right by years. The mitigating factor is that Starship has revenue paths that do not need refueling, starting with V3 Starlink deployment and commercial heavy lift, so a refueling delay does not by itself end the Starship option.

8. Starlink Mobile: The Next Monetization Wedge

Starlink Mobile provides text, light data and over-the-top voice to ordinary phones through about 650 V1 Mobile satellites, reaching about 7.4 million monthly unique devices in about 30 countries. SpaceX partners with about 30 carriers on six continents, including T-Mobile in the United States and One NZ, Optus, Telstra, Rogers, KDDI, Salt, Entel, Kyivstar and VMO2 abroad. It charges carriers a fixed fee or a per-user fee, and also earns revenue through revenue-sharing arrangements. SpaceX has agreed to buy EchoStar spectrum for about $19.6B: about $11.1B in Class A stock plus up to $8.5B of EchoStar debt payoff. The FCC approved the deal on May 12, 2026, and it is expected to close around November 30, 2027.

Assessment: Starlink Mobile is the largest under-discussed growth wedge in the S-1. The S-1 sizes the market at $740B, carrier partnerships provide distribution, and the EchoStar spectrum will give SpaceX its own licenses. Revenue is still small and is not broken out: it sits inside Enterprise & Government, and the S-1 says only that mobile contributed $632M of growth in 2025 and $85M in Q1 2026. It could reach $5-10B+ a year by FY2030 in bull scenarios. Competition is real, and the S-1 names AST SpaceMobile, Lynk, Globalstar and Skylo. Country-by-country licensing is the main friction.

9. AI Compute Capex: The Burn Engine

AI capex was $12,727M in 2025, 61.4% of the consolidated $20,737M, and $7,723M in Q1 2026 alone, 76.4% of that quarter's total. That is hyperscaler-scale spending inside a company with $18.7B of annual revenue. The S-1's use-of-proceeds statement lists "the expansion of our AI compute infrastructure" first. The company also says it could cut data-center spending and reallocate it if near-term needs shrink or ramp more slowly. It already sells spare capacity: the Anthropic agreements pay $1.25B a month through May 2029.

Assessment: AI capex is the largest single use of cash in the S-1 and the main source of the cash burn the IPO proceeds will fund. Unlike Starship, where pre-revenue investment has an operational end point, AI capex is open-ended: competing in frontier models requires continued compute scaling. The bull case treats this as the early AWS build-out (heavy capex now, dominant returns later); the bear case treats it as a structural capex sinkhole. Our base case assumes AI capex of $10-15B a year for the next 3-4 years. Q1 2026's $7.7B in a single quarter already runs well above that pace, which puts this assumption at risk.

10. The Accumulated Deficit: $41.3B

The accumulated deficit was $37,035M at December 31, 2025 and $41,311M at March 31, 2026. Because the S-1 combines SpaceX, xAI and X under common-control accounting, this deficit includes the historical losses of X and xAI as well as SpaceX. It is not a measure of SpaceX's own losses since its founding in 2002. Recent periods account for a large part of it: net losses were $(4,628)M in 2023, $(4,937)M in 2025 and $(4,276)M in Q1 2026, against net income of $791M in 2024.

Assessment: The deficit says less about SpaceX's history than the headline suggests, because it mixes three companies. The recent pattern matters more: the combined company was profitable in 2024 and returned to losses as AI spending ramped. Investors must underwrite a multi-year path back to GAAP profitability, which runs through Starlink scaling, AI revenue catching up with its cost base, and Starship moving to paying launches.

11. Capital Intensity

Capex was 111% of revenue in 2025 ($20,737M on $18,674M) and 215% in Q1 2026 ($10,107M on $4,694M). Operating cash flow covered about a third of 2025 capex. The indicated raise net of the bridge repayment, plus $23,675M of cash and short-term securities, gives roughly $79B of liquidity. That lasts about five and a half years at 2025's cash burn, or a little over two years at Q1 2026's pace, after which the company would need much higher operating cash flow or more capital.

Assessment: Capital intensity is the structural constraint on SPCX's valuation. Until operating cash flow can fund capex, the company is implicitly committed to raising more capital over time. The IPO proceeds solve the near-term funding need but do not change the long-term capital intensity. We model free cash flow turning positive by FY2028-2029 in our base case (bull FY2027; bear FY2030+), but that path assumed capex near 2025's level.

12. International Expansion and Geopolitical Exposure

Starlink serves 164 countries, territories and other markets, and the S-1 treats foreign regulation and politics as a core risk. Its examples are specific. In August 2024 Brazil's Supreme Court froze Starlink's Brazilian financial assets over purported violations of Brazilian law by X, which SpaceX did not then own. Certain foreign governments have publicly discussed using anti-satellite weapons against the Starlink constellation. The S-1 also cites licensing processes that may be influenced by protectionist policies, challenges to ITU spectrum priority, and export controls, and it excludes China and Russia from its market-size estimates.

Assessment: Geopolitical exposure is multi-faceted. The Brazil episode shows that actions aimed at X or at Musk personally can reach Starlink's assets, and that link is tighter now that X sits inside the same company. Licensing and spectrum-priority disputes can fragment the addressable market country by country. We assess geopolitical risk as a 5-10% drag on our long-term market assumption.

Use of Proceeds & Offering Structure

The S-1 states its intended use of proceeds in general terms: "We intend to use the net proceeds from this offering to fund our growth strategy, including the expansion of our AI compute infrastructure, enhancements to our launch infrastructure and launch vehicles, increases in the scale and capacity of our satellite constellations, and any remaining amounts for general corporate purposes." It gives no dollar allocation. It also does not mention the bridge loan in that section, although the loan's terms require an amount equal to the net proceeds to be applied to it within six months. Based on those terms and on operating cash needs, we estimate the indicated $75 billion of gross proceeds will be allocated approximately as follows:

Offering Mechanics

Share Structure

The S-1 counts 6,932,508,000 Class A shares and 5,602,790,410 Class B shares outstanding as of May 1, 2026, after the five-for-one stock split, the Class C reclassification and the preferred-stock conversion: 12,535M shares in all. At the indicated $1.75T, that is about $140 per share before any new shares from the offering. A $75B raise at that price would add about 537M Class A shares, roughly 4.1% of the enlarged share count. Musk's 849M Class A and 5,569M Class B shares, which include restricted stock and options, carry 85.1% of the vote.

xAI Risk Subsection

The xAI combination introduces risk categories that pre-merger SpaceX did not carry. They deserve separate treatment because they materially change the SPCX investment profile.

Regulatory and Legal Exposure

Beyond the matters in Key Topic 3, the S-1 discloses these AI and platform exposures:

• EU privacy regulation: the Irish Data Protection Commission's February 2026 inquiry covers the processing of EU users' personal data, including children's, through Grok's generative AI functionality on X.
• Child safety: the FTC inquiry concerns how chatbots are evaluated when they act as companions to children and teens.
• Copyright: the company is a defendant in litigation alleging copyright infringement from the use of copyrighted works to train its AI models.
• Grok image generation: putative class actions filed in January and March 2026, and the City of Baltimore's suit, allege that Grok's image-generation and editing features enabled nonconsensual explicit images.
• New AI and platform laws: the S-1 lists the EU AI Act, California and New York AI laws, and the UK Online Safety Act among the legal frameworks that could apply to its AI and platform businesses. It names no specific enforcement action under them, although it says such laws and related inquiries have already led to unfavorable outcomes such as compliance costs and product changes.

Strategic Fit Risk

The xAI and SpaceX combination is structurally unusual: it joins frontier AI and a social platform with space transportation in a single public company. The strategic rationale (vertical AI integration, talent retention, distribution synergies) is debatable. Public market shareholders may price a "conglomerate discount" of 5-15% for the lack of pure-play exposure.

Capability Lag Risk

In our view, xAI's Grok models have lagged OpenAI's GPT and Anthropic's Claude through 2025-2026, although the S-1 notes that Grok reached frontier-level scientific reasoning on the GPQA Diamond benchmark within two years of its first release. Grok 5, which the S-1 says is being trained at COLOSSUS II, is the critical milestone: closing the capability gap is essential to xAI's $250B+ standalone valuation. A failure to reach parity would compress xAI's implied valuation toward $150-200B and put downward pressure on the consolidated valuation.

Talent Concentration

We believe xAI's research talent base is concentrated among a relatively small number of senior researchers. Compensation packages, retention bonuses and equity structures are critical to retention and represent a meaningful ongoing operating cost. Loss of key research talent could materially affect Grok roadmap execution.

Starlink TAM-Build DCF: The Analytical Centerpiece

Starlink is the economic engine of SPCX: Connectivity was 61% of FY2025 revenue, and its $7,168M of Segment Adjusted EBITDA exceeded the consolidated $6,584M. The Starlink standalone valuation is therefore the most important single component of any SPCX valuation framework. We construct a multi-year TAM-build DCF with explicit assumptions on subscribers, ARPU, capex and operating margin.

Subscriber Build (FY2025-FY2035)

The S-1's Starlink Subscriber count covers Personal and Business plans and excludes managed enterprise and government contracts. It gives no consumer and enterprise split, so the FY2025 split is not shown. The forecast splits are our estimates.

Blended ARPU Build (FY2025-FY2035)

Starlink Revenue + Operating Margin Build

FY2025A is the Connectivity segment as reported: $11,387M of revenue, $4,423M of operating income and $4,178M of capex. The S-1 does not report segment free cash flow.

DCF Mechanics

• WACC: 10.0% (reflects: risk-free 4.5% + equity risk premium 5.0% + business-specific premium 0.5% for execution/regulatory)
• Terminal growth rate: 3.0% post-FY2035 (mature subscription business, growing modestly above inflation)
• Tax rate: 21% (assumed federal statutory)
• Discount period: FY2026-FY2035 explicit + terminal value
• FY2025 base: $11.4B revenue, $4.42B op income

Starlink DCF Output

Starlink standalone fair value range: $760B to $1.30T, with base case $1.05T. This is the analytical anchor of our SPCX valuation framework. At the base case, Starlink alone is worth more than the median S&P 500 component and exceeds the market caps of Verizon, AT&T, T-Mobile and Comcast combined.

Comp-Based Valuation: Launch / xAI / Starship

Launch Services Standalone Valuation: $40-50B

SpaceX's launch business has no direct public-market comparable at scale. Our valuation combines three pieces. (a) The present value of contracted launch work: ~$10B in the base case. The S-1 prints only a consolidated backlog of $28,377M across all three segments, so this is our estimate of the Space share. (b) The ongoing launch business at roughly 4-8.5x its 2025 Launch Services revenue of $2,576M, or $10-22B with $15B in the base case. (c) The option value of SpaceX's national-security launch position: $10-30B.

xAI Standalone Valuation: $200-300B

Reference points: the reported $250B merger mark, and competitor marks (OpenAI $500B, Anthropic $350B). On $3,201M of 2025 revenue, most of it X advertising, and a $(6,355)M operating loss, the AI segment is a venture-stage business at a hyperscaler valuation. The range reflects (a) downside from Grok regulatory and capability-lag risk vs. (b) upside from potential parity with OpenAI and Anthropic over 18-24 months. These scenarios were set without the Anthropic compute contract: at full rate it alone would exceed the $5-8B of FY2028 revenue in our base case, but either party can end it on 90 days' notice.

Starship Optionality: $50-100B

Starship is pre-revenue, with over $15B invested to date. We treat it as a long-dated call option on heavy-lift launch and Mars logistics. Our approach is real-options-based: probability-weighted scenarios on (a) successful operational deployment (probability 60-80%), (b) HLS delivery (probability 40-60%), (c) commercial heavy-lift market capture (probability 70-85% if Starship works), and (d) Mars revenue (essentially zero in a 10-year window).

Sum-of-the-Parts Bridge

SOTP base case: $1.27 trillion. Range: $858B to $1.85T. Before the governance discount, the base case is ~$1.39T. We set our fair value band at $1.35-1.55T, above the discounted base case. The $1.75T indication sits inside our bull case but well above our base case.

Street Perspective

Debate: Is the $1.75T IPO Target Justified by Starlink Alone?

Bull view: Starlink's standalone fair value approaches $1.0-1.3T in the base and bull cases. With Launch Services contributing $40-60B, the SpaceX core business alone supports $1.0-1.4T, and the xAI and Starship layers are the upside option. The IPO is therefore reasonably priced if you believe in the Starlink trajectory.

Bear view: Even on bullish assumptions, Starlink standalone tops out around $1.3T, and that requires aggressive multi-year compounding. The $1.75T indication implies Starlink at the high end of the bull case plus substantial value for xAI and Starship plus governance neutrality. The probability that all three upside components arrive together is meaningfully below 50%.

Our take: The bull view requires Starlink to deliver flawless execution against an optimistic forward curve; the bear view is too conservative on operating momentum but right that the IPO embeds several speculative assumptions. We lean modestly bear. Starlink alone supports a $1.0-1.1T base case, and the remaining $650-750B of the indicated valuation needs xAI, Starship and premium pricing to come through together. Possible, but not the most probable outcome.

Debate: Does the xAI Consolidation Enhance or Detract from the Valuation?

Bull view: xAI brings $250-300B of standalone AI exposure to SPCX, a mega-cap AI position investors otherwise cannot buy (OpenAI and Anthropic are private). Integration with Starlink (an AI distribution layer) and with X (real-time data and distribution) creates ecosystem effects, and the Anthropic contract shows the compute can also be sold to third parties. The xAI valuation will move toward OpenAI and Anthropic as Grok improves.

Bear view: The xAI integration is structurally problematic. The conglomerate (space transportation plus frontier AI plus a social platform) lacks operating synergy, dilutes pure-play exposure, and exposes SPCX to AI and platform regulatory risk it would not otherwise face. The AI segment spent $12.7B on capex in 2025 and $7.7B in Q1 2026 alone, and lost $(6,355)M at the operating line in 2025. Public market investors will price a 5-15% conglomerate discount versus a hypothetical Starlink pure-play.

Our take: The xAI consolidation is net-negative for the SPCX equity story. xAI carries optionality, but the operating complexity, regulatory overhang and capex burden outweigh the strategic benefits. A pure-play SpaceX (Starlink + Launch + Starship) at $1.2-1.4T would be a cleaner investment than the combined entity at $1.75T. We think the consolidation reflects Musk's broader AI strategy more than a strict optimization for SPCX shareholders.

Debate: Starship, Generational Optionality or Capital Sinkhole?

Bull view: Starship is the next-generation launch architecture that will cut per-kg launch costs by another order of magnitude, unlock heavy commercial payloads, enable mass Starlink V3 deployment, and ultimately support Mars settlement. The $15B+ invested is a fraction of the potential payoff. Even at a 50% probability of operational success, the expected value is $75-150B, and full-success scenarios value Starship at $300B+.

Bear view: Starship has consumed over $15B, and fundamental architecture questions remain open, above all in-orbit refueling, which the company has not yet attempted. Space R&D of $3.0B in 2025 is a capital drain that lengthens the path to consolidated profitability. Even successful Starship operation does not guarantee returns at bull-case valuations.

Our take: Starship has genuine optionality but is unlikely to deliver returns commensurate with its bull-case framing in the next 5-10 years. Our base case ascribes $75B in option value, and we view bull cases above $150B as speculative. Unproven in-orbit refueling is the concrete near-term risk; the multi-decade upside is real but not part of the base-case investment thesis.

Debate: Musk Governance, Founder Premium or Discount?

Bull view: Musk's track record (Tesla, SpaceX's private-market compounding, multiple successful exits) justifies a founder premium. Concentrated voting allows long-term decision making free from quarterly capital-markets pressure. Other Musk-led businesses have generated meaningful long-term shareholder returns despite governance critiques.

Bear view: Musk's 85.1% of voting power before the offering is among the most concentrated structures brought to US markets at this scale, and the controlled-company exemptions reduce independent oversight. Musk's other roles (Tesla, Neuralink, The Boring Company) raise key-person and conflict-of-interest concerns, and the S-1 notes he does not devote his full time to SpaceX. Public shareholders bear the economic risk with little governance recourse.

Our take: We apply a governance discount of roughly 9-10% of pre-discount value, reflecting the concentration of control. The discount is consistent with historical dual-class trading patterns (Meta, Alphabet and Snap have traded at 5-15% discounts to comparable single-class peers). The debate partly resolves itself: in periods of strong execution the market underweights governance; under stress it overweights it. We assume cyclical re-pricing of the governance factor over time.

What They're NOT Saying

1. Offering Terms and Post-Offering Ownership: The price range, shares offered, size of the option to purchase additional shares, size of the directed share program, and every post-offering ownership and voting percentage are blank. Investors are asked to evaluate governance before seeing the post-IPO voting table.
2. Starlink ARPU Breakdown: Subscriber ARPU is given in total only, with no split by region, plan tier or customer type, and there is no enterprise ARPU at all. The S-1 does split Connectivity revenue into Consumer and Enterprise & Government, and gives total revenue by country (USA, Ireland, Canada, all other), but that does not show how pricing differs across markets.
3. Starlink Subscriber Churn: No churn rate is disclosed. The only retention data point is that no Starlink Enterprise customer contributing more than $750,000 of annual revenue has voluntarily discontinued service since 2023. At 10.3M subscribers, even modest churn implies meaningful spending to replace lost lines.
4. Backlog by Segment and Customer: Backlog of $28,377M is disclosed in aggregate across all three segments. The largest customer (20.9% of 2025 revenue, across all segments) is not named, and single-customer and re-bidding risks are obscured.
5. Starlink Mobile Revenue: Mobile revenue sits inside Enterprise & Government and is not broken out. The S-1 gives the device count and says mobile added $632M of growth in 2025 and $85M in Q1 2026, but not the revenue level or carrier economics.
6. Insider Sale Plans: No 10b5-1 plans or intended post-lock-up selling are disclosed. The lock-up release schedule is disclosed, and the founder's shares are locked for 366 days.
7. Network Capacity Utilization: The S-1 gives capacity (over 700 Tbps of cumulative downlink) but not utilization, and it quantifies no spectrum or orbital constraints.
8. Mars Program Capital Allocation: Mars is outside the quantified market estimate and absent from the use of proceeds, and its spending is not separated from Space R&D. Implicit Mars-program capital could be $1-5B a year within Starship development (our estimate).
9. Launch Pricing and Unit Cost: The S-1 gives no current price per launch or per kilogram and no cost per reused Falcon 9, only that customer pricing was roughly flat in 2025. Competition from Blue Origin, Rocket Lab Neutron and Chinese launchers could pressure pricing.
10. AI Customer Pipeline: Beyond the Anthropic compute agreements, a product list (Grok Business, Grok Enterprise, Grok API, xAI Gov) and the Cursor agreement, the S-1 gives no AI customer counts or pipeline. The Anthropic agreements can be ended by either party on 90 days' notice, so a contract worth $1.25B a month is less durable than its May 2029 end date suggests. The competitive position against OpenAI (Microsoft channel) and Anthropic (Amazon channel) in enterprise deployments matters substantially.
11. Cybersecurity Incident History: The S-1 acknowledges that security incidents have disrupted operations in the past but gives no details of specific incidents or breaches.
12. International Spectrum Disputes: The S-1 describes spectrum and ITU-priority risks in general terms but does not enumerate specific pending disputes or at-risk markets.

Market Context

• Pre-IPO secondary market price (Forge Global): $650.66 per share as of May 20, 2026, quoted as an implied ~$1.45T. The price is on a pre-split basis; on the S-1's 12,535M post-split shares (about 2,507M pre-split), it implies about $1.63T
• Most recent tender offer (Dec 2025): $421/share (pre-split) at an $800B valuation of SpaceX before the xAI merger
• IPO target valuation: $1.75T, about $140 per share on the S-1 count: +21% over the $1.45T secondary quote (+7% over the ~$1.63T that price implies on the S-1 count), and about +66% per share from the Dec 2025 tender ($421 pre-split, about $84 after the split)
• Implied public float: ~4.3% at a $75B raise on a $1.75T valuation, among the lowest IPO floats at this scale
• Lock-up: the founder and certain significant investors hold a 366-day block (all of the founder's shares); the remaining shares have staggered releases, tied to the Q2 and Q3 2026 results and to days 70 through 135, that finish at day 180, about six months after pricing
• Expected Russell 1000 inclusion: June 2027 reconstitution
• Expected Nasdaq 100 inclusion: December 2026 or March 2027 quarterly rebalance
• S&P 500 inclusion: blocked by negative GAAP earnings until consolidated profitability is reached (likely 2-3 years post-IPO)

The pre-IPO secondary quote ($650.66, quoted at ~$1.45T) sits inside our $1.35-1.55T fair value band, but on the S-1's share count the same price implies ~$1.63T, above it. Either way, the $1.75T indication sits above both the secondary market and our band (by 13-30% for the band). Two interpretations: (a) the underwriters are seeking maximum primary capital and have priced the offering aggressively, expecting strong index and retail demand to support trading at or near the IPO mark; or (b) the secondary market is under-pricing SPCX because it lacks the visibility into the prospectus that institutional buyers will have during the roadshow.

Historical mega-IPO patterns suggest first-year returns are typically negative: Saudi Aramco -3%, Alibaba -22%, Meta -30% in Year 1, SoftBank Mobile -28%. The $75B raise is unprecedented, which makes historical comparisons imperfect, but the structural overhang (staggered lock-up releases, a low float and heavy retail participation) is real. Our base-case expectation: SPCX trades in a $1.4-1.7T range through its first 12 months, with a wider $1.2-1.9T range possible depending on lock-up dynamics and broader risk-asset performance.

Thesis Scorecard: Initiating Coverage

Aardvark Labs Preliminary PT Framework

Per-share values use the S-1's 12,535M shares outstanding as of May 1, 2026, before any shares issued in the offering. The $1.75T indication equals about $140 per share on the same count.

Aardvark Labs Rating: Initiating at Hold at the indicated $1.75T valuation.

The operating quality of SpaceX is real. Starlink at scale is one of the best businesses in technology: high growth, expanding margins, a durable moat and a large addressable market. Falcon reliability is unmatched. Starship optionality is genuine. The xAI consolidation is debatable but provides AI optionality at the consolidated level.

But the $1.75 trillion indication pushes against the upper bound of what our sum-of-the-parts framework supports. Our SOTP base case of $1.27T sits 27% below the indication, and even our bull case at $1.85T only slightly exceeds it. The indication also sits 13-30% above our $1.35-1.55T fair value band. The path to upside from this entry requires either (a) IPO pricing 15%+ below the indicated $1.75T, (b) post-IPO multiple expansion driven by Starship operating milestones or AI revenue gains, or (c) sustained index-inclusion demand absorbing the lock-up releases.

For long-term investors with multi-year horizons and tolerance for the governance structure, SPCX is a reasonable hold at the indicated price, with re-rating optionality. For trading-horizon investors, the post-IPO dynamics (low float, heavy retail participation, staggered lock-up releases) suggest waiting for either a price reset toward the ~$1.45T pre-IPO secondary quote or the final 180-day lock-up release, about six months after pricing.

We will revisit the rating after (a) final IPO pricing, (b) the first post-IPO earnings release (anticipated late Aug or mid-Nov 2026), and (c) the lock-up releases through day 180. The next 6-9 months will clarify whether $1.75T is the right anchor or whether the market settles at a different equilibrium.