We Are All About to Become Bagholders in the SpaceX IPO
And that is without buying a single share.
This looks like the largest transfer of wealth from ordinary people to institutions in recent memory. Here is how it works.
Nasdaq Changed the Rules for SpaceX
Nasdaq quietly changed its index inclusion rules ahead of the SpaceX IPO. Companies can now be included in the Nasdaq 100 just 15 days after their IPO. The historic seasoning period was three to twelve months. They also removed the 10% free float requirement — SpaceX is only listing 5% of its shares.
These rule changes appear to have been made specifically for this IPO.
Here is why that matters. Retirement funds and retail investors holding NASDAQ 100 ETFs like QQQ will be forced to buy SpaceX stock. Not because they chose to. Because it is in the index. Every passive investor automatically becomes a SpaceX shareholder at whatever valuation it lists at.
That is how you and I become bagholders whether we want to or not.
SpaceX valuation is based entirely on AI
SpaceX is a genuinely great business. It has a near-monopoly in rocket launches. It owns Starlink, X, and xAI. Goldman Sachs is projecting $474 billion in revenue by 2030, and the breakdown is where things get interesting.
Of that $474 billion:
- $322 billion from AI
- $144 billion from Starlink
- $8.3 billion from the actual space business
Less than 2% from rocket launches — which is supposed to be the core business. So what is the $322 billion AI revenue built on?
Space Data Centers: The 100 Gigawatt Fantasy
SpaceX currently has two operational data centers. Colossus 1 is fully sold out to Anthropic. Part of Colossus 2 is leased to Google and the combined annualized value of both deals is around $26 billion. $1.25b per month from Anthropic and $920m per month from Google.
You will ask why a major LLM developer is leasing their capacity to competitors - they faced severe technical issues with Colossus 1 when trying to train Grok so they decided to lease to Anthropic. They have excess capacity for Colossus 2 which they decided to lease to Alphabet. Don't forget that this way they will show solid revenue growth in their next reports.
The $322 billion AI revenue projection is built on plan to launch data centers into space to capture solar energy and beam it back as compute power. The target is 100 gigawatts of compute power deployed in space per year.
Their S-1 filing describes a deployment plan requiring roughly one million metric tons transported to orbit annually, using satellites carrying over 100 kilowatts of compute power each.
Let me do the math. 100 gigawatts divided by 100 kilowatts per satellite is 1 million satellites per year. If a Starship carries around 100 satellites per launch, that requires 10,000 launches per year — one launch every 52 minutes, every day, for the entire year.
SpaceX posted on their official website that they did approximately 650 launches in total.
These are rough numbers, but they show the scale of what is being promised. Elon Musk has a long track record of overpromising to investors. Tesla shareholders were told autonomous vehicles and humanoid robots were coming. They are still waiting. The 100 gigawatt space compute plan reads like a number designed to impress IPO investors, not a plan with a credible path to execution.
Starlink Is the Real Business with Operating margin of 39%
Starlink is the only profitable segment SpaceX currently has. It accounts for 61% of their 2025 revenue and generated $4.4 billion in operating profit at a 39% operating margin.
Goldman Sachs projects Starlink reaches $144 billion in revenue by 2030. That is a 12x increase. If operating margins hold at 39%, that implies $56 billion in operating income. Compared to the space compute fantasy, this at least looks like a number rooted in reality — though it still requires perfect execution and no meaningful competition.
That last part is the risk. Amazon LEO is an emerging competitor, and it should not be underestimated. Amazon has a first-mover disadvantage in satellites but a massive distribution advantage through AWS. They can call their existing cloud customers and bundle LEO connectivity directly into the offering. That is a serious threat to Starlink's pricing power and margins. Other smaller players are also entering the market.
Starlink's moat is real. But it is not unassailable.
The Space Business Is Not the Growth Story
Goldman Sachs projects only $8.3 billion from the actual space launch business by 2030. That sounds underwhelming next to the AI and Starlink numbers, but I think this is the most interesting part of the company.
Rocket launches are a business with enormous barriers to entry. SpaceX holds a near-monopoly. Their launch capability is what makes everything else possible — it is the lever they use to build and defend the other two businesses.
The space business is not the growth story. It is the moat that protects the growth story.
SpaceX fundamentals an AI analysis
The Summary
SpaceX is a real business with a genuine moat in rocket launches. Starlink is profitable and growing. The management team has executed on things most people said were impossible.
But the valuation at $2.1 trillion is pricing in $322 billion in AI revenue that depends on launching rockets every 52 minutes, plus a 12x increase in Starlink revenue, plus perfect execution with no competitive disruption.
The rule changes Nasdaq made — 15 days to inclusion, no free float requirement — mean passive investors will fund this valuation automatically.
For fundamental charts: stockpicker.tech