Now you might wonder what a rocket company has to do with AI. Well, Elon Musk, who controls SpaceX, also controlled xAI, an AI company that owns Grok and the social media network Twitter X. And in February of this year, he sold xAI to himself SpaceX for $250 billion in shares. SpaceX now has three businesses, a very profitable satellite internet part called Starlink, an unprofitable rocket part, and a money-burning AI part. But in the documents filed for the SpaceX IPO it is claimed that it is the AI part that will address $26.5 trillion of "total addressable market" for all three parts of $28.5 trillion. The AI part of SpaceX is what is used to justify a total valuation for the whole SpaceX of $1.75 to $2 trillion. That makes the SpaceX IPO the biggest IPO in the history of the world, mostly based on a business that is losing billions of dollars of money every year, and doesn't have a very clear path towards a profitable future.

The SpaceX IPO will also be one of the first to profit from new rules for the Nasdaq index. Under rules valid until May 1st of this year, SpaceX wouldn't even have made it into the Nasdaq index, as for that at least 10% of a company's shares needed to be on the market. SpaceX will only have 4% to 5% of its shares on the market, leading to higher volatility, but this is now allowed since this month. Under another new rule, SpaceX will be added to the index after 15 days of trading, previously at least 3 months.

So there is a definitive possibility that the SpaceX IPO will, at least initially, do very well. With so little of the company shares being actually sold, the rarity of those shares could very possibly keep prices up initially. And 15 trading days later, every Nasdaq ETF would then have to buy SpaceX shares when the company is added to the index, further increasing demand for these rare shares. Meanwhile the people who hold the other 95% of those shares are initially locked out from selling them, with a staggered lockout period lasting a total of 180 days. As a speculative share for day traders and later short-sellers, SpaceX might be quite a hot ticket. For long-term investors, it is probably a very bad deal, as SpaceX is even more likely to follow a typical share price evolution after IPO, peaking shortly after the IPO, and then falling in the following 6 months. If SpaceX as expected is initially high on the Nasdaq index, any fall would be amplified by those ETFs shedding the share automatically.

Unusually, 30% of the SpaceX IPO is reserved for retail investors, which usually get nothing of hot IPOs. But of course there is a large Elon Musk fan club, who are likely to be interested. Despite a long history of overpromising and underdelivering, a lot of people still believe that any venture of Elon Musk can only make them rich. Even an IPO that list under its business goals to build a colony of 1 million people on Mars, and whose AI business prospects are also very much science fiction.

Where it gets interesting is the possibility of the SpaceX IPO either not doing very well, or crashing even harder than expected after the initial hype. If you believe in an old school idea of the value of a company being the sum of future profits, the numbers don't really add up. SpaceX is losing $5 billion on $18 billion of revenue. You can't even calculate a price to earnings ratio, as there aren't any earnings. If you calculate a price to revenue ratio, you get a ratio of 100. If you consider that Nvidia has a price to revenue ratio of 20, you begin to understand that it would take a major miracle for SpaceX to ever return the money to investors. Not only would AI have to become fabulously profitable, SpaceX would also have to beat all the other AI companies like OpenAI, Anthropic, Google, and Microsoft to the biggest slice of the cake. Seeing how Grok is currently most famous for being able to undress women against their will on social media, that is quite a stretch.

So the SpaceX IPO is a first test case, testing the markets true feelings on the future of AI. And if that goes well, the 6 months after the IPO will be another, longer test case. If the market holds up, later this year the IPOs of OpenAI and Anthropic are expected. Overall, AI companies will have IPOs valued between $3 and $4 trillion this year. This could end in an US stock market that is even more concentrated on AI by the end of this year. Or somewhere before that end of the year investors will get cold feet, reject a hot ticket IPO, and cause a cascade of rethinking of valuations for AI companies, pricking the AI bubble. 2026 is going to be an interesting year for AI valuations, and thus for the stock market as a whole.