SpaceX (SPCX) files to go public at a preliminary price range; offering size and terms TBD

SPCX has a dual-class structure where Class B shares carry 10 votes per share versus 1 vote for Class A shares. Class B shares convert to Class A upon transfer except for permitted transfers, and Class B holders elect a majority of the board. This structure concentrates voting control with Class B holders (primarily Mr. Musk) while public Class A holders have limited governance influence.

Proceeds will fund AI compute infrastructure expansion, launch infrastructure and vehicle enhancements, and satellite constellation scaling. The AI segment had capital expenditures of $7,723M in Q1 2026 (76% of total capex of $10,107M), indicating substantial ongoing AI infrastructure investment needs.

SPCX reported a net loss of $4,276M for Q1 2026, compared to a net loss of $528M in Q1 2025 and a net loss of $4,937M for full-year 2025. The company was profitable in 2024 with net income of $791M but returned to losses in 2025 and Q1 2026, driven primarily by AI segment losses of $2,469M in Q1 2026.

The AI segment reported an operating loss of $2,469M in Q1 2026, up from $936M in Q1 2025, and a full-year 2025 operating loss of $6,355M. This segment has been loss-making across all periods shown, with losses accelerating as the company scales AI compute infrastructure (nameplate compute draw increased from 0.3 gigawatts in Q1 2025 to 1 gigawatt in Q1 2026).

SPCX does not plan to pay dividends in the foreseeable future and will retain earnings to finance growth. Credit agreement covenants also restrict dividend payments, and future borrowing arrangements may impose additional restrictions.

SpaceX reported Q1 2026 consolidated revenue of $4,694 million, an operating loss of $(1,943) million, and Adjusted EBITDA of $1,127 million. The operating loss reflects heavy investment across segments, particularly in AI, while Adjusted EBITDA remains positive, indicating cash generation before non-cash charges.

For full-year 2025, SpaceX reported consolidated revenue of $18,674 million, an operating loss of $(2,589) million, and Adjusted EBITDA of $6,584 million. The company is unprofitable on a GAAP operating basis despite generating substantial positive Adjusted EBITDA, reflecting significant R&D and capital investment.

AI capital expenditures were $7,723 million in Q1 2026 (76% of total capex of $10,107 million) and $12,727 million in 2025. The AI segment, acquired in early 2026 via the xAI acquisition, is consuming the majority of SpaceX's capital as the company builds gigawatt-scale compute infrastructure (COLOSSUS, COLOSSUS II) and plans orbital AI compute satellites.

The AI segment reported Q1 2026 revenue of $818 million with an operating loss of $(2,469) million and negative Adjusted EBITDA of $(609) million. For 2025, AI revenue was $3,201 million with an operating loss of $(6,355) million and negative Adjusted EBITDA of $(1,237) million. The segment is in heavy investment mode and unprofitable, reflecting its early stage and the company's stated priority to capture AI opportunities.

SpaceX acquired xAI in early 2026, adding AI compute infrastructure and the Grok frontier model to its space and connectivity businesses. The acquisition is framed as strategic integration to enable orbital AI compute satellites and leverage SpaceX's launch and satellite capabilities for AI workloads, but it also brought substantial operating losses and capital intensity.

SpaceX has an option to acquire Cursor at a $60.0 billion implied equity value, payable in Class A common stock based on a seven-day VWAP. If SpaceX terminates the option or Cursor terminates due to SpaceX's material breach, SpaceX owes Cursor a $1.5 billion termination fee plus an $8.5 billion deferred services fee, totaling $10 billion in potential cash or stock obligations. This represents significant contingent liability and dilution risk.

SpaceX has secured a $1.25 billion per month compute services contract with Anthropic running through May 2029, representing approximately $45 billion in contracted revenue over three years. However, either party can terminate with 90 days' notice, so the revenue is not fully locked in. The customer retains ownership of its AI models and data.

Musk will control the majority of voting power through Class B shares, which elect a majority of the board. As holder of a majority of Class B shares, Musk can elect, remove, or fill vacancies among Class B Directors and control shareholder votes. The specific percentages are not disclosed in this preliminary filing.

This is a preliminary S-1 filing with placeholder dollar amounts not yet filled in. The actual net proceeds to the company (after underwriting discounts and estimated expenses) will be disclosed when the price range is set in an amendment or final prospectus. The filing shows two scenarios: base offering and with full greenshoe exercise.

The company plans to allocate proceeds across three specific capital-intensive areas: AI compute infrastructure (which represented 76% of total capex in the latest period), launch infrastructure and vehicles, and satellite constellation expansion, with remainder for general corporate use. No specific dollar amounts or percentages are allocated to each category, giving management broad discretion.

The company will not pay dividends and will retain all earnings to fund growth. Returns to shareholders will come only from share price appreciation. Credit agreement covenants also restrict dividend payments.

The section discloses that new investors will experience immediate and substantial dilution, but all specific figures (net tangible book value per share, offering price, dilution amount, share counts, consideration amounts, ownership percentages) are blank placeholders. The framework shows dilution will be calculated as the difference between the offering price and the adjusted pro forma net tangible book value per share after the offering, with separate impacts disclosed for existing investors (a decrease in net tangible book value per share) and new investors (dilution). Actual dilution metrics will be filled in the final pricing prospectus.

The section includes a table framework showing how shares and total consideration will be split between Elon Musk and other existing investors versus new investors, with average price per share for each group. All specific numbers (share counts, percentages, dollar amounts) are blank. The disclosure separately identifies Elon Musk from other existing investors in the ownership structure, indicating his significant individual stake will be broken out in the final prospectus.

The company's entire growth strategy—next-generation V3 satellites, V2 satellite-to-mobile connectivity, and orbital AI compute infrastructure—depends on successfully developing and deploying Starship at scale. Current operational rockets (Falcon 9 and Falcon Heavy) cannot deploy V3 or V2 Mobile satellites. Delays in Starship development, achieving launch cadence, or full reusability would slow deployments, increase costs, and defer revenue. Without full reusability and rapid turnaround, AI compute satellites at scale would not be economically compelling.

The company requires FAA launch licenses for each launch operation. Current FAA regulations do not permit return-to-launch-site reentries for Starship, requiring a waiver that is not guaranteed. Following anomalies or mishaps, the FAA may impose investigations, corrective actions, or restrict launch operations. As launch frequency grows, FAA resource constraints could lengthen application processing times, delaying the high launch cadence required for the company's growth strategy.

The company's AI products (Grok) include less-constrained modes ("Spicy" Imagine Mode, "Unhinged" Voice Mode) that carry heightened risks of generating explicit content, misinformation, nonconsensual imagery, or discriminatory outputs. The company is subject to investigations from regulators and law enforcement in the U.S. and internationally concerning allegations that its AI products were used to create nonconsensual explicit images or content representing children in sexualized contexts, and faces ongoing putative class action lawsuits on these allegations. In February 2026, the Irish Data Protection Commission launched a large-scale inquiry into whether the company's AI segment complied with GDPR obligations, including processing personal data of EU children using Grok within the X platform.

The company's satellite-to-mobile connectivity services depend on spectrum access. The EchoStar spectrum transaction (AWS-4 and H-block licenses) received FCC approval on May 12, 2026 but remains subject to other closing conditions and is not expected to close until November 2027. Even if completed, there is no assurance the acquired spectrum will be sufficient for growing needs. The company must also secure the global right to use this spectrum from numerous international telecommunications regulators to make V2 satellite-to-mobile services usable worldwide, with no assurance such authorizations will be granted on acceptable terms or at all.

The company plans to deploy orbital AI compute infrastructure, a technology that has never been operated in space by anyone. The effects of the space environment (radiation, thermal cycles, micrometeoroids, debris) on AI infrastructure are untested. Once deployed, the infrastructure cannot be easily repaired or upgraded, so component failures would result in permanent capacity loss, accelerated depreciation, or need for replacement. AI compute satellites at scale require full Starship reusability to be economically compelling.

The company disclosed that AI capital expenditures represented 76% of total capital expenditures in the latest period, totaling $7,723 million out of $10,107 million. This concentration indicates the company is prioritizing AI infrastructure investment over its Space and Connectivity segments, which may constrain capital available for satellite deployment, launch facilities, or Starlink expansion.

The company disclosed total principal indebtedness of $29,132 million as of March 31, 2026. This substantial debt level increases vulnerability to economic downturns, requires dedicating significant cash flow to debt service, and limits operational flexibility. The debt is at variable interest rates, exposing the company to interest-rate risk.

The company disclosed that achieving orbital AI at scale requires significantly more AI chips than currently available. Terafab, a planned chip-manufacturing facility, is intended to address this shortage, but the company states it may not be successful. The company also disclosed it has no long-term contractual arrangements with direct chip suppliers and procures all GPUs on a purchase-order basis, leaving it exposed to supply disruptions and prioritization of other customers.

The company reports Q1 2026 consolidated revenue of $4,694 million, an operating loss of $(1,943) million, and Adjusted EBITDA (a non-GAAP measure) of $1,127 million. The operating loss indicates the company is not yet profitable on a GAAP basis despite generating revenue across its three segments.

For full-year 2025, the company reports consolidated revenue of $18,674 million, an operating loss of $(2,589) million, and Adjusted EBITDA of $6,584 million. The company remains unprofitable on a GAAP operating basis despite substantial revenue scale.

AI capital expenditures were $7,723 million in Q1 2026 (76% of total capex of $10,107 million) and $12,727 million in 2025 (62% of total capex of $20,737 million). The AI segment is consuming the majority of the company's capital investment, reflecting the company's stated priority to scale compute infrastructure despite the segment's operating losses.

The AI segment generated $818 million revenue in Q1 2026 with an operating loss of $(2,469) million and negative Segment Adjusted EBITDA of $(609) million. For 2025, AI segment revenue was $3,201 million with an operating loss of $(6,355) million and negative Segment Adjusted EBITDA of $(1,237) million. The segment is in heavy investment mode with substantial losses relative to revenue.

The Connectivity segment is the company's most profitable segment, generating $3,257 million revenue and $1,188 million operating income in Q1 2026. For 2025, Connectivity revenue was $11,387 million with operating income of $4,423 million and Segment Adjusted EBITDA of $7,168 million, showing strong year-over-year growth (49.8% revenue, 120.4% operating income, 86.2% Adjusted EBITDA). This segment's profitability partially offsets losses in Space and AI.

AI segment capex reached $7,723 million in Q1 2026, up from $2,567 million in Q1 2025, representing 76% of the company's total $10,107 million capex. The increase was driven by rapid expansion of terrestrial data centers (COLOSSUS and COLOSSUS II). For full-year 2025, AI capex was $12,727 million versus $5,633 million in 2024. This reflects the company's strategy to scale compute infrastructure at unprecedented speed, though the AI segment remains deeply unprofitable.

The AI segment posted a $2,469 million operating loss in Q1 2026, up from a $936 million loss in Q1 2025. For full-year 2025, the AI operating loss was $6,355 million versus $1,561 million in 2024. Losses are driven by cloud computing costs, GPU depreciation, data center infrastructure, and employee expenses, partially offset by higher revenue. The company states it expects a multi-year investment horizon before sustained positive Adjusted EBITDA.

The Space segment posted a $662 million operating loss in Q1 2026 versus a $70 million loss in Q1 2025. For full-year 2025, Space had a $657 million operating loss versus $21 million operating income in 2024. The deterioration is attributed to accelerated Starship vehicle development and launch facility investments, plus a decrease in customer launch revenue. The company is prioritizing Starship development speed over near-term profitability.

The company disclosed that AI capital expenditures represented 76% of total capital expenditures in the latest period, totaling $7,723 million out of $10,107 million. This indicates a heavy investment focus on AI infrastructure (data centers, compute clusters) relative to other segments, reflecting the company's strategic prioritization of AI growth and its belief that AI will be a primary near-term revenue driver.

The company reported 10.3 million Starlink Subscribers as of March 31, 2026, across 164 markets, representing a small fraction of an estimated 3.3 billion potential end users in served markets. This disclosure establishes the current subscriber base and highlights significant headroom for growth, positioning Starlink as an early-stage penetration story in a massive addressable market.

The company disclosed that Starlink Mobile serves approximately 7.4 million monthly unique devices across approximately 30 countries as of March 31, 2026, and that it entered into agreements in 2025 to acquire 65 MHz of U.S. spectrum and global Mobile Satellite Service spectrum from EchoStar. The spectrum acquisition is described as enabling a step-change in service possibilities, and the company expects to deploy next-generation V2 Mobile satellites in 2027 to increase capacity by orders of magnitude. Regulatory approvals are pending in most international jurisdictions, and the U.S. FCC approved the EchoStar license transfer in May 2026, with remaining U.S. authorizations expected in Q2 or Q3 2026. This establishes the current scale of the mobile offering and the regulatory and technical milestones required for the planned Gen2 service expansion.

The company claims a $740 billion TAM for Starlink Mobile based on 8 billion mobile devices globally and a weighted average monthly ARPU of $8 per user. This is a company-provided estimate, not an independent market figure. The TAM also includes a $660 billion consumer broadband market, $200 billion enterprise broadband market, $5 billion government satellite communications market, and a $26.5 trillion AI market opportunity (comprising $2.4 trillion AI infrastructure, $760 billion consumer AI subscriptions, $600 billion digital advertising, and $22.7 trillion digital economy for enterprise applications).

The company states Starship will begin payload delivery in H2 2026 after 11 flight tests to date. The filing discloses that achieving targeted launch cadence requires significant capital investment in land, launch sites, vehicle production scaling, propellant facilities, power supply, and FAA regulatory approvals. However, the company notes that full upper-stage reusability and in-orbit refueling are not required for deploying V3 satellites or V2 Mobile satellites in LEO, only for missions beyond LEO.

AI capital expenditures represented 76% of total capex in the latest period, totaling $7,723 million out of $10,107 million in total capital expenditures. This indicates the company is making substantial infrastructure investments in AI compute capacity, which aligns with its stated $2.4 trillion AI infrastructure market opportunity based on 104 million GPUs required globally by 2030.

The company discloses a subscriber base of 10.3 million for its Starlink broadband service as of March 31, 2026, across 164 countries and territories. This is the first quantitative disclosure of the Starlink subscriber count in this filing and provides a baseline for measuring the scale of the Connectivity segment's consumer broadband business.

The company operates approximately 9,000 broadband satellites in low Earth orbit as of March 31, 2026, which it describes as the largest satellite constellation in human history. The filing also states the company launched approximately 3,100 Starlink satellites in 2025, roughly five times more than the entire second-largest LEO constellation. This scale is foundational to the Connectivity segment's competitive position and capacity.

PricewaterhouseCoopers LLP included an explanatory paragraph in their audit report specifically calling attention to the Company's significant related-party transactions detailed in Note 18. An explanatory paragraph is unusual and signals the auditor deemed these transactions material enough to warrant special emphasis beyond a clean opinion.