The world's biggest IPO is launching a satellite company that wants to train AI in space, but protesters want you to know about the chatbot it already owns.

The Summary

The Signal

SpaceX acquired xAI and its Grok chatbot earlier this year, a detail that matters more now that the rocket company is going public. The S-1 filing explicitly calls out Grok's not-safe-for-work capabilities as a risk factor, unusual language for a company primarily known for launching satellites and ferrying astronauts to the International Space Station. This isn't a footnote. It's a warning that SpaceX's business model now includes AI infrastructure it doesn't fully control.

The orbital data center plan is ambitious in scale and relatively simple in execution. Each satellite will pack AI chips, solar panels, and liquid radiators into a 20-meter frame, using mostly existing Starlink technology. Musk called the design "much simpler" than current Starlink satellites, which suggests SpaceX sees this as a manufacturing problem, not an R&D moonshot. The solar panels will come from a new factory in Bastrop, Texas, built specifically for this project.

"We don't think this is a super hard problem, compared to things we already do."

But here's the tension: SpaceX wants to train AI models in space while the AI product it already owns generates content toxic enough to warrant protest and S-1 disclosure. The timing matters. The inflatable appeared the day before the biggest IPO in history, tattooed with "SpaceX's Grok makes AI child porn" across its abs and back. The protesters aren't confused about corporate structure. They know SpaceX owns Grok now, and they're telling potential investors to connect those dots.

The scale of the orbital plan is staggering:

  • Up to 1 million satellites planned for launch
  • 70-meter wingspan makes them the largest satellites SpaceX has ever built
  • First detailed specs released during IPO week, not buried in technical papers

Musk responded to earlier complaints by saying anyone using Grok to make illegal content "will suffer the same consequences as if they upload illegal content." X's official policy claims zero tolerance for child sexual exploitation. But policy statements don't change what the model can do, and SpaceX's own filing acknowledges the gap between what should happen and what could happen.

The orbital data centers will train models on hardware SpaceX controls in an environment free from terrestrial constraints. No power grid issues. No cooling limitations beyond what radiators can handle. No regulatory ambiguity about who owns the compute happening 340 miles above Earth. This is about control, not just capability. The $1.75 trillion valuation assumes investors believe SpaceX can build the rails for the next generation of AI training infrastructure.

The Implication

Watch how SpaceX separates its orbital ambitions from its terrestrial AI baggage. The company needs to convince investors it can build trillion-dollar space infrastructure while managing an AI product that already made it into the risk factors section. If protests show up at the IPO, or if early investors start asking hard questions about Grok's content moderation, SpaceX may need to spin off or sunset the chatbot entirely. The orbital data center plan doesn't need Grok to succeed, but Grok's reputation could absolutely drag down SpaceX's AI credibility.

For anyone building in the agent economy, this is what regulatory capture looks like from orbit. SpaceX is racing to build AI training infrastructure where Earth-based rules don't easily apply. The first company to crack orbital compute at scale gets to write the playbook. Just don't expect protesters to stop pointing out what the company already built down here.

Sources

Business Insider Tech | Bloomberg Tech | Business Insider Tech