SpaceX IPO: Here's Why — and What Every Investor Needs to Know Before the Biggest IPO in History
$11.4B Starlink 2025 Revenue
63% EBITDA Margin
Is SpaceX a broadband internet provider? An energy company? An AI infrastructure play?
The answer is: all three. And that's exactly what makes it impossible to value — and impossible to ignore.
And, Does Blue Origin CEO Jeff Bezo agree with the direction? lets find out.
01 · THE HOOK: The Elephant That Learned to Sprint
Opening frame: set up the paradox
SpaceX is about to file what could be the largest IPO in U.S. history — targeting $2 trillion. But here's the thing: most investors still think it's a rocket company.
The real story? Starlink is the whole game. Everything else — rockets, Starship, launch services — is currently running at a loss.
A company with $11.4B in satellite internet revenue growing at 50% YoY is not a startup. It's a category-defining infrastructure giant — mid-flight.
📡 Identity Crisis — Or Master Strategy?
Ask ten analysts what SpaceX is and you'll get ten different answers. A satellite ISP? A defense contractor? An AI compute play? A clean energy company in orbit? The answer is all of the above — and that ambiguity is either its greatest risk or its greatest moat. We argue: it's the moat.
02 · THE NOW
Starlink: A Cash Machine at Altitude
The financial reality behind the IPO hype
61% of total revenue in 2025 came from Starlink's Connectivity segment — and it generated virtually all operating profit ($4.42B positive). Every other division is in the red.
Revenue grew +49.8% YoY. Operating profit grew +120.4%. EBITDA grew +86.2% — at a base of $11.4 billion. This is a giant elephant moving at cheetah speed.
Subscribers: 2.3M → 4.4M → 8.9M → 10.3M (Mar 2026). Now in 164 countries. Potential addressable population: 3.3 billion people.
EBITDA margin: 63% — higher than most software companies. SpaceX builds the rockets, launches the satellites, and owns the customer relationship end-to-end.
Bear case to address: Average Revenue Per User (ARPU) fell from $99 to ~$81/month — international expansion is diluting unit economics. Volume must compensate.
03 · THE FUTURE
Space Data Centers: The 100× Bet, From selling internet to selling compute — in orbit
💡 The Hidden Revenue Line
Anthropic — the AI company behind Claude — pays approximately $1.25 billion per month to SpaceX for terrestrial data center services. That's ~$15B/year from a single customer. If SpaceX builds orbital data centers and captures even a fraction of the exploding AI compute market, the revenue potential is staggering. Some projections suggest a 100× revenue multiplier over current data center contracts.
SpaceX has filed FCC plans for an orbital data center network of up to 1 million satellites — dwarfing Blue Origin's 51,600 Project Sunrise filing.
Why orbit? 24/7 uninterrupted solar power — no clouds, no weather. AI training clusters need constant, massive power. Earth-based grids are at capacity. Orbit is the release valve.
SpaceX already has the launch infrastructure, the satellite network, and the ground station ecosystem. The marginal cost to add compute payloads to next-gen Starlink satellites is dramatically lower than any competitor.
The xAI merger (Feb 2026) merged Grok, X (formerly Twitter), and AI infrastructure under one roof. The flywheel: Starlink data feeds Grok, Grok adds value to Starlink enterprise, orbital data centers power both.
04 · THE RIVAL
Blue Origin's Project Sunrise — and a notable reality check from the competition, CNBC · May 20, 2026
"Space is going to be a gigantic industry. I would caution people who think it's all science fiction to be a little cautious with their judgment, because it is real, it is happening."
On SpaceX's 2–3 Year Timeline
"Some of the timelines we hear are very short. People would talk about two or three years. That's probably a little ambitious."
The Long View
"We will be able to beat the cost of terrestrial data centers in space in the next couple of decades. These giant training clusters will be better built in space, because we have solar power there, 24/7. There are no clouds and no rain, no weather."
Jeff Bezos— Fall 2025 Remarks (previously on record)
Bezos agrees on the destination — space data centers will happen. The dispute is purely on timing: SpaceX says years, Bezos says decades.
Blue Origin's Project Sunrise (FCC filing): 51,600 sun-synchronous satellites, data speeds up to 6 Tbps, targeting 100,000 enterprise customers. Launch starts: Q4 2027.
Key barriers Bezos cited: energy management, chip costs, and launch economics — all areas where SpaceX has a significant structural head start via Falcon 9 reusability and vertical integration.
SpaceX / Starlink
1M satellite filing (FCC)
10,000+ active satellites today
10.3M subscribers, 164 countries
Falcon 9: 29× booster reuse
Already profitable at scale
$1.25B/mo AI customer (Anthropic)
Blue Origin
51,600 Project Sunrise satellites
TeraWave: 5,408 constellation
Deploy start: Q4 2027
New Glenn: recent orbit failure
Bezos-funded; seeking outside capital
Calls 2–3 yr timeline "ambitious"
05 · THE TENSION
The $5 Billion Loss Nobody Is Talking About
Behind the IPO hype — the risks that matter
⚠ Bear Case / Analyst Friction
Reuters initially reported ~$8B in profit on $15–16B revenue. Then The Information revealed a $5B GAAP net loss. The gap? SpaceX's $20.7B in capex makes depreciation enormous. EBITDA looks great; net income does not. Investors need to decide which number to trust.
xAI consumed 61% of SpaceX's $20.7B in 2025 capex — the AI bet is burning enormous cash, with customer contracts cancellable in 90 days. This is a venture inside a public company.
Valuation at $1.75–2T implies a ~94× price-to-sales ratio. Aswath Damodaran's DCF model: ~$1.22T — 30% below IPO price.
Starlink ARPU is falling: $99 → $81/month. International growth is subscriber-rich but margin-dilutive. The subscriber growth story must outpace the ARPU erosion story.
Elon Musk himself warned of "genuine risk of bankruptcy" if Starship fails to reach bi-weekly launch cadence. That's the man who knows the company best — not a short seller.
06 · THE THESIS
So What Is SpaceX, Really?
"Starlink is not a telecom company. It's the first global energy-internet-compute utility — the infrastructure layer for the next century of civilization. Whether the valuation is right or wrong, the category is real."
As a broadband provider: Already the largest satellite ISP on Earth. 3.3B addressable users. If it captures 5% of global broadband market, that alone justifies hundreds of billions in enterprise value.
As an energy company: Orbital data centers powered by 24/7 solar could make SpaceX one of the largest clean energy compute providers by 2040. Bezos agrees — he just disagrees on timing.
As an AI company: The xAI merger means Grok, user data from X, and Starlink's ground-level connectivity form a closed AI flywheel. Add orbital compute and it's a vertically integrated AI infrastructure stack that no terrestrial cloud provider can replicate.
Questions for IPO Filing
Starlink ARPU trajectory through 2027
Timeline for first orbital compute payload
xAI contract terms and cancellation risk
Starship launch cadence targets
Anthropic contract details (duration, renewal)
Watch These Signals
Blue Origin TeraWave launch Q4 2027
SpaceX S-1 public filing date
Starlink Q2 2026 subscriber count
AI capex as % of total SpaceX spend
FCC approval: orbital data center spectrum