Why SpaceX Bought xAI Before the $1.75T IPO

On January 6, 2026, xAI, the Elon Musk-founded artificial intelligence company that makes the Grok chatbot, closed a $20 billion funding round at a reported valuation of roughly $230 billion.^[1]Nvidia was in. Cisco was in. Qatar Investment Authority, Abu Dhabi's MGX, and Fidelity were in. The round was oversubscribed, more than $5 billion above the original $15 billion target.^[1]

Twenty-four days later, on January 30, xAI's board concluded that the company's fair value had “doubled” to $250 billion, and the company merged into SpaceX at a combined $1.25 trillion valuation. Musk announced the deal in a blog post on February 2.^[2]

Fifty-eight days after that, on April 1, SpaceX filed confidentially with the Securities and Exchange Commission (SEC) for what would be the largest initial public offering (IPO, the event where a private company sells stock to public investors for the first time) in history. The target: $1.75 trillion valuation, up to $80 billion raised, on a float (the share of a company's stock that actually trades publicly) of roughly 3% to 5%.^[3]

It landed almost exactly there. SpaceX priced on June 12, 2026, selling 555.6 million Class A shares at $135 on the Nasdaq under the ticker SPCX, a $1.77 trillion valuation. The stock closed its first day at $161, up 19%, a $2.1 trillion market cap.^[12]The plan worked. That doesn't make the plan what it was advertised to be.

The official story is that Musk is building “the most ambitious, vertically-integrated innovation engine on (and off) Earth”, and that combining rockets, satellite internet, a large language model (LLM, the kind of AI system that powers chatbots like Grok and ChatGPT), and a social platform gives SpaceX a moat in the coming era of solar-powered orbital data centers. That's the story Bloomberg, CNBC, and Reuters mostly went with. It's the story the underwriters sold in June.

I don't think it's the real story.

The way I read the sequence, $80 billion to $113 billion to $230 billion to $250 billion to $1.75 trillion in twelve months, with no corresponding change in product, revenue, or addressable market, is that the merger was initial public offering plumbing, not corporate strategy. It locked xAI's valuation at $250 billion, 9% above ($250B / $230B = 1.087) the just-closed private round. It gave xAI's backers a liquid path out through SpaceX's tape. And it built one vertically-integrated narrative big enough to carry a $1.75 trillion listing on a float thinner than Aramco's.

The orbital data centers are real ambition. They're not a 2026 product. They're the story.

The merger was IPO plumbing, not a space strategy.

Under the Jumpstart Our Business Startups Act (the 2012 law that created the “emerging growth company” path to market), confidentially-filed registration documents aren't public on EDGAR (the SEC's online filing database, short for Electronic Data Gathering, Analysis, and Retrieval) until at least fifteen days before the roadshow, the pre-IPO tour where management pitches the deal to big investors.^[4]That's why nobody saw a line of SpaceX's financials until May, seven weeks after the confidential filing went in. Musk's February announcement had already made clear the IPO was the downstream event.^[2]

Look at how tight the sequence is. The merger closed January 30, 2026. Announced February 2. The confidential S-1 (the registration document a company files with the SEC before going public) went in April 1. The public S-1 landed May 20. Pricing June 12.^[12] Every one of those dates was picked to land SpaceX on Nasdaq at maximum valuation, with a unified corporate story, before AI-cycle sentiment had a chance to reset.

xAI was burning a billion a month and the clock was running out.

xAI's spending run through 2025 was extraordinary. The company burned through roughly $7.8 billion in operating expenses over the first nine months of 2025, about $870 million a month, and Bloomberg reporting put the run-rate closer to $1 billion a month by year-end as compute commitments escalated.^[7] Almost all of it went to compute: training runs on Nvidia graphics processing units (GPUs, the AI training chips that cost $25k to $40k each at scale), power contracts for the Colossus data center in Memphis, and a race against OpenAI, Anthropic, and Google to ship the next version of Grok.

The January $20 billion round was supposed to give xAI twelve to eighteen months of cushion at current burn rates. It was oversubscribed. The board didn't have a liquidity problem in the ordinary sense.

But every month xAI stayed private was a month where the AI-valuation regime could shift. Anthropic, OpenAI, and Perplexity were all telegraphing 2026 or 2027 IPO paths. If any of them printed a lower-multiple tape, or if the broader mood turned, xAI's next round would be pricing against that print. A down round at xAI's scale, with Qatar Investment Authority, Abu Dhabi's MGX, and Sequoia on the cap table, is a headache nobody wanted.

Merging into SpaceX solved that problem by ending the problem. No more rounds. One liquid exit on someone else's tape.

Where did the other $500 billion come from?

Start from the arithmetic. On January 30, 2026, the combined SpaceX entity was marked at $1.25 trillion by its own board. On June 12 it priced at $1.77 trillion.^[12]That's a half-trillion-dollar uplift in about four months, on the same business that existed the day the merger closed.

I spent a while trying to reconcile that half-trillion to a specific operational event, and I can't. No major contract landed. No regulatory risk was removed. Starship was still working through its test campaign. The only real fundamental catalyst pointing upward is Starlink: $11.4 billion of 2025 revenue, roughly 61% of the whole company, and about $7 billion of segment EBITDA at 63% margins.^[12]That's a real tailwind. It's the only real tailwind, and it was already visible in January.

The only way the number works is if the reader accepts that the uplift is a narrative rerating, not a business event. The pitch says the combined SpaceX, xAI, X, Starlink, and Starshield (SpaceX's military and intelligence satellite business, which holds classified contracts with the US Department of Defense) entity is worth more than the sum of its pre-merger parts because the parts now reinforce each other: Grok runs on Starlink backbone, xAI provides autonomy software for Starshield, Starship enables orbital compute, X is the distribution layer. The parts are technically real. None of it shows up in the numbers yet.

Why a full merger, not a secondary or a tender?

This is the question the merger was designed to prevent getting asked. Whether $250 billion is really what xAI is worth on its own, priced as a standalone AI company against OpenAI and Anthropic comparables, is the one question the combined prospectus never has to answer. Folded into SpaceX, it's one segment inside a $1.75 trillion headline. Separated, it's a contested private mark that would have to survive the actual market.

This is also the question that separates “Musk is optimizing for the IPO” from “Musk is making a genuine corporate-strategy call.” He controlled both cap tables. He could have:

1. Run a secondary tender at SpaceX to buy out some xAI investors. No merger, no integration risk, no combined-entity consent needed.
2. Created a holding vehicle that owned stakes in both. Cleaner corporate separation, with xAI left to IPO on its own timeline.
3. Done nothing. Let xAI raise again at whatever private mark it could get in 2026 or 2027.

He did none of those. He merged the two companies entirely, in an all-stock deal, on a tight timetable, a few weeks after a $20 billion round had just priced xAI at a specific market-tested number.

The most coherent reading is that a full merger did three things a secondary or a holding vehicle could not:

• It locked xAI's valuation permanently at $250 billion, 9% above the market-tested $230 billion, before any subsequent repricing could happen.
• It gave every xAI private investor, including the ones who just bought in at $230 billion, a path to liquidity through SpaceX's IPO prospectus. A secondary cashes out a subset. A merger converts everyone into the same liquidity event.
• It lets underwriters write a single S-1 against a combined set of financials, not a SpaceX S-1 with an xAI equity-method investment on the side. Combined financials justify a combined headline valuation. Separate financials invite the question of whether the $250 billion xAI mark can be defended on its own.

The tight float is the whole playbook.

Most retail investors reading about the SpaceX IPO focus on the valuation. The more interesting number is the float.

SpaceX sold 555.6 million Class A shares at $135, roughly $75 billion of stock, on a float of about 5%.^[12] That is a colossal raise sitting on a sliver of the company. In dollar terms it is the biggest offering ever. In percentage terms it is one of the thinnest.

The template is Saudi Aramco's December 2019 IPO. Aramco listed at a $1.7 trillion valuation with a 1.5% float on the Saudi Tadawul exchange. It raised $25.6 billion, eventually $29.4 billion after the greenshoe (the underwriter option to sell up to 15% more shares than the base deal if demand is strong).^[9] The offering fell short of original ambitions, 5% on an international exchange at a $2 trillion mark, and settled for a scarcity-driven local listing.^[10]Aramco held the world's largest headline market cap for years, but institutional investors largely sat out, and the stock has mostly traded in a range rather than rewarding the early retail buyer who chased scarcity at IPO.

That's the playbook SpaceX ran, at bigger scale. Tight float plus a 30% retail allocation, roughly triple the Wall Street standard of about 10%,^[12] means a lot of the supply-demand balance sits with retail buying interest rather than institutional due diligence. Retail buying interest follows the story. The story, again, is orbital data centers.

The difference from Aramco is that Aramco had $88 billion in net incomethe year it went public. SpaceX had a $4.9 billion net loss in 2025.^[12]The scarcity mechanic is the same. The underlying cash generation isn't.

What the S-1 actually showed.

The public S-1 landed on May 20. It answered the questions the confidential filing let everyone speculate about, and the answers are not flattering to the orbital-data-center pitch.

• Starlink is the company. $11.4 billion of 2025 revenue, roughly 61% of the $18.7 billion total, throwing off around $7 billion of segment EBITDA at 63% margins.^[12] If you bought SPCX, you mostly bought a satellite broadband business with a launch business attached.
• xAI is the hole in the boat. The AI segment posted about $3.2 billion of revenue against roughly $6.4 billion of operating loss.^[12]That's the segment carrying a $250 billion mark inside a $1.77 trillion headline. The prospectus never has to defend that mark on its own, which was the point.
• The loss is real. $4.9 billion net loss for 2025 on the combined entity.^[12]Starlink's profit is funding xAI's burn. That's the actual capital structure under the narrative.
• Musk owns the votes. He holds roughly 85.1% of the voting power post-offering through Class B stock, with the Class B holders electing a majority of the board as a separate class.^[12] Public shareholders got economic exposure and essentially no governance.
• S&P 500 inclusion is off the table.S&P Dow Jones Indices restricts multiple-share-class companies.^[11]At an 85% voting split, SPCX doesn't clear it. Russell inclusion runs on its own schedule and is the realistic near-term index path.

None of this is a claim that the IPO flopped. It obviously didn't. The stock is up on scarcity and story, exactly the way tight-float mega-offerings do for a while. Aramco did. What I'm saying is narrower: the merger is a better lens on what the $1.77 trillion actually represents than the orbital-data-center pitch is. The business is the business it was in January, plus Starlink growth that was already visible, and minus an xAI that was going to need capital one way or another.

The question the merger was designed to prevent getting asked, whether $250 billion is really what xAI is worth standing on its own with a $6.4 billion annual operating loss, still hasn't been asked by the market. It has just been rolled into a bigger number. That's the one a real investor has to answer first, and the lockup expiries later this year are when it starts getting asked out loud.