The IPO Race That Will Determine Who Owns AI

SpaceX lost $2.47 billion on artificial intelligence in the first quarter of 2026. Its AI division generated $818 million in revenue. The company is targeting a $1.75 trillion valuation in an upcoming IPO. These three facts belong in the same sentence because they are the same story.

The jury verdict in the Musk v Altman case — a 9-0 win for OpenAI on a statute-of-limitations technicality — told the courtroom who won the day. The SpaceX filing tells you who is winning the decade. Elon Musk's rocket company merged with his AI startup xAI at a $1.25 trillion valuation in February, and xAI immediately became the load-bearing wall of the combined entity's balance sheet. According to Reuters, xAI accounted for 76 percent of SpaceX's $10.1 billion in capital spending during the first quarter of 2026. The AI division's losses exceeded the revenue it generated by roughly three to one.

That is not a startup problem. It is an infrastructure bet. SpaceX has built or leased data center clusters in Memphis, Tennessee — branded the Colossus and Colossus II clusters — and sold access to them at prices that suggest the compute is the asset, not the rockets. Anthropic is paying SpaceX $1.25 billion per month to use those clusters through May 2029, according to a term sheet summarized in SpaceX's IPO filing. The rocket company is not just a rocket company anymore. It is an AI compute wholesaler with a very expensive customer inside its own corporate family.

OpenAI is betting the same way, from the other direction. OpenAI burned cash at historic rates while posting a March 2026 funding round valuation of $852 billion, and filed its own IPO paperwork within 48 hours of SpaceX's filing becoming public — moving the rivalry from the courtroom to Wall Street in the span of a single news cycle. The two companies are racing each other to Wall Street at the same moment they are racing each other to build out AI data centers that nobody has yet proven can be funded by AI revenue alone. As Reuters reported, SpaceX posted $4.69 billion in first quarter revenue but a $1.94 billion operating loss, with Starlink generating a $1.19 billion operating profit that was consumed entirely by the AI division's losses.

The courtroom was always secondary to this. Elon Musk's lawsuit sought $134 billion to be redistributed from OpenAI's for-profit arm to its nonprofit, and removal of Sam Altman and Greg Brockman. The jury deliberated for hours and returned a 9-0 verdict that Musk filed outside the three-year statute of limitations. OpenAI's attorney called it "not a technical decision; it's a substantive one" — a framing the jury rejected. Musk donated roughly $38 million to OpenAI between 2015 and 2018 and spent the intervening years building a competing company, and the legal theory that he was owed a stake in the nonprofit's converted for-profit entity never reached a jury's examination of the merits.

What the courtroom produced was a procedural win for Altman and an IPO timeline. Both companies now need to demonstrate that the infrastructure they are building will eventually be funded by the customers using it — not by the investors buying the IPOs. OpenAI converted to a for-profit Public Benefit Corporation structure in late 2025, a move that let its nonprofit foundation retain a governance role while releasing equity to investors — and that foundation is now one of the largest nonprofit endowments in history. OpenAI's converted for-profit structure left a nonprofit foundation holding a stake valued at approximately $130 billion after the October 2025 conversion. That foundation is simultaneously OpenAI's governance relic and one of the largest nonprofit endowments in history. It is also one of the entities most exposed to whether the IPO pricing holds.

The market that will judge both companies is not a courtroom. It is the group of institutional investors who decide what a $1.75 trillion SpaceX and an $852 billion OpenAI are actually worth when the private-market leverage of a merged xAI-SpaceX and the cash-burn trajectory of OpenAI become public documents subject to quarterly scrutiny. The trial was over in a day. The IPO race has years to run.

What to watch: whether the first post-IPO earnings reports from either company show AI compute revenue growing fast enough to justify the capital spending already committed. If the numbers do not justify the valuations within 18 months of listing, the infrastructure story becomes a refinancing story — and refinancing frontier AI companies at scale is a problem nobody has solved in public markets yet.