An IPO that goes far beyond space

SpaceX’s IPO document released this week tells a much broader story of an atypical space company. Behind the most anticipated IPO of the technological decade lies an industrial infrastructure combining space, telecommunications, artificial intelligence, energy and cloud computing.

Above all, the S-1 presentation document shows that Elon Musk is no longer just preparing a listed space company. It attempts to structure a group capable of controlling several critical layers of the global digital economy: orbital launch, global connectivity, high-performance computing, energy infrastructure and artificial intelligence.

The figures also give an idea of ​​the scale of the project. In 2025, SpaceX generated $18.674 billion in revenue for $6.584 billion in Adjusted EBITDA, while remaining operationally loss-making at $2.589 billion. The group’s logic is to massively generate cash on certain activities to finance industrial investments rarely observed in modern tech.

Starlink has become the real cash machine

Reading the SpaceX IPO document reveals one of the blind spots of the empire built by Elon Musk: behind the rockets, Starlink has become the real economic engine of the group, and undoubtedly the most strategic asset of its entire industrial architecture. The satellite network no longer only finances SpaceX’s space ambitions, and now fuels the group’s expansion in artificial intelligence, giant data centers and, tomorrow, orbital computing.

In 2025, the Connectivity segment, mainly driven by Starlink, generated 11.387 billion dollars in revenue, compared to 4.086 billion for the historical space activity. Above all, Starlink posts $4.423 billion in operating income and $7.168 billion in Adjusted EBITDA.

The network now claims 10.3 million subscribers across 164 countries, territories and markets, powered by approximately 9,600 low-orbit satellites. At this point, SpaceX already operates one of the largest private physical infrastructures in the world.

SpaceX’s real bet is called Starship

All this architecture, however, relies on one critical element: Starship. The S-1 explicitly recognizes that the success of the group depends largely on the industrial success of the program. Without fully reusable Starship, no very high capacity V3 satellites, no massive direct-to-cell, no profitable orbital computing and probably no lunar economy.

SpaceX presents Starship as the future space equivalent of a commercial aircraft, a vehicle capable of rapid rotations with extremely low marginal costs. The group claims to be aiming for a 99% reduction in the historical cost of access to orbit. If this hypothesis materializes, SpaceX would have an industrial advantage that is practically impossible to reproduce in the short term. The document clearly shows that Starship is not only a space project but represents the logistical infrastructure essential to the rest of the group’s economic model.

AI becomes SpaceX’s new industrial front

This connectivity income now finances SpaceX’s other strategic priority: artificial intelligence. The S-1 details at length the integration of xAI and Grok within the group. SpaceX explicitly describes AI as a logical extension of its physical infrastructure. The document even states that “The future of artificial intelligence will be determined by control of the physical layer.

Behind this formula lies an industrial thesis where the future of AI will depend less on the models themselves than on the control of energy, chips, data centers, networks and orbital transport.

The group is already investing massively in computing infrastructures. In the first quarter of 2026 alone, investment spending in the AI ​​segment reached $7.723 billion. The COLOSSUS and COLOSSUS II data centers constitute the heart of this strategy. SpaceX claims to have built one of the largest coherent AI clusters in the world, with around a gigawatt of computing power available.

However, this activity remains extremely loss-making. In 2025, the AI ​​segment posts $3.201 billion in revenue but $6.355 billion in operational losses. The group openly accepts this logic of destroying cash in the short term in order to accelerate the deployment of computing capacities.

The most radical project of S-1: moving computing into space

The most spectacular point of the document, however, concerns orbital computing. SpaceX openly explains that it wants to move part of the computing into space. The group believes that land-based energy constraints will ultimately limit the expansion of traditional data centers. Space would then offer several structural advantages: quasi-continuous solar energy, passive cooling and massive deployment capacity thanks to Starship.

The document even mentions the potential deployment of constellations with “millions of satellites” dedicated to orbital computing. Few listed companies have already presented at this level of detail a project combining cloud, energy and space infrastructure.

This vision also explains the logic of the Terafab project, carried out with Tesla and Intel, intended to produce chips on a very large scale. SpaceX is gradually trying to integrate the entire physical chain of AI:

• energy,
• semiconductors,
• data centers,
• connectivity,
• and orbital infrastructure.

A very atypical IPO on a financial and political level

The SpaceX IPO also presents singularities, thus the group will adopt a dual-class structure allowing Elon Musk to retain almost absolute control of the company after the IPO. SpaceX will be considered a “controlled company”, which will considerably limit the influence of minority shareholders on governance.

The document also reveals a considerable level of capital intensity. Between launchers, satellites, data centers, energy infrastructure and future industrial projects linked to semiconductors, SpaceX looks more like a mix between a cloud hyperscaler, a telecom operator and a heavy industrial group than a classic software company.

The group already carries more than $29 billion in debt. The risks identified in the S-1 cover:

• Starship addiction,
• regulatory constraints,
• orbital risks,
• energy availability,
• export controls,
• cyberattacks,
• and the potential failure of technologies that do not yet exist on a large scale.