SpaceX
IPO Briefing
The business, the moat, and the key debates ahead of the SPCX listing.
The core flywheel
SpaceX lowered the cost of getting to orbit. Then it used that edge to build Starlink. Starlink then underpins demand for more launches, enabling SpaceX to amortize its infrastructure and drive down costs further.
1. Infrastructure: Launch
Reusable rockets make launch cheaper and more repeatable. This is an infrastructure layer that serves all launch use cases.
2. Application: Starlink
SpaceX uses launch cost advantages to build a global internet network, one of the highest-value applications of its launch infrastructure.
3. Flywheel: Launch more
Starlink needs more satellites, which gives SpaceX more internal demand, which allows it to scale up launch to drop costs further.
What SpaceX sells today
SpaceX sells launch services, plus internet, enterprise, and AI services that benefit from access to space.
Launch
- Falcon 9 and Falcon Heavy carry satellites, cargo, and crew.
- SpaceX is the primary launch provider for the U.S. government.
Starlink broadband
- Consumer internet across 164 countries, territories, and markets.
- 10.3M direct Starlink service lines at March 31, 2026.
Enterprise and government
- Connectivity for remote sites, aviation, maritime, retail, disaster response, and public services.
- Starshield is a dedicated network for U.S. government customers.
xAI
- Frontier lab, Grok, X, and large amounts of compute spending.
- A long-term dream is AI compute in orbit.
The moat: Learning curve + scale economies, shared
SpaceX spent more than two decades learning how to build, launch, land, reuse, and relaunch rockets. SpaceX shares its learning curve and economies of scale with customers by providing the lowest launch costs, who then give SpaceX even more volume, dropping costs further.
Cost advantage
Falcon 9 cut launch cost per kilogram by about 85% versus the historical average.
Starship is designed for full reuse and roughly 100 metric tons to orbit; if that works at cadence, cost per kilogram could fall again.
Process power
Decades of learning across manufacturing, software, launch ops, recovery, and refurbishment, tightly integrated together.
Continuing down the cost curve
Industry-leading launch cadence and volumes give SpaceX engineers more data, spreads fixed costs, and keeps the learning curve moving.
Starlink gives SpaceX internal demand
SpaceX built a customer that always needs more launch capacity, and deepens its moat from scale.
Internal customer
Starlink enables SpaceX to launch economically even when third-party demand is lumpy, underpinning consistent learning curves and economies of scale.
Capacity flywheel
More satellites create better service, better service creates more demand, and more demand supports more launches.
Government and enterprise
Government and enterprise customers buy managed connectivity for aviation, maritime, remote sites, public services, and defense use cases through negotiated contracts.
Starlink is the profit engine
Launch is the core of the moat, but SpaceX monetizes through Starlink.
Current earnings driver
Connectivity generated $4.4B of operating income in 2025, while the whole company reported a $2.6B loss from operations.
Positive mix shift to enterprise and government
Enterprise and government work can move Starlink beyond consumer broadband into aviation, maritime, fixed-site backup, mobile, and defense. No large enterprise or government customer has churned since 2023.
Starship is the next jump in scale for Starlink
Falcon was phase one. Starship promises to make the network much cheaper and bigger.
- SpaceX expects Starship payload delivery to orbit to begin in the second half of 2026.
- Starship V3 is designed to carry 100 metric tons to orbit with full reuse.
- Falcon rockets cannot deploy V3 Starlink satellites or V2 Mobile satellites.
- Starship turns Starlink into a broader broadband-and-mobile network, rather than just a rural cable substitute.
downlink capacity per V3 satellite
V3 Starlink satellites per Starship launch
more downlink capacity per launch versus Falcon 9
SpaceX has its eyes on bigger opportunities
SpaceX's dominant franchises, launch and Starlink, are also much smaller TAMs according to the company's TAM breakdown. The real prize is in AI Infrastructure and Enterprise Applications.
solutions
broadband
mobile
infrastructure
subscriptions
advertising
applications
market
Proven franchises are a narrow slice of the TAM
Launch and space-enabled services are where SpaceX has the clearest lead, but that TAM is small compared with the AI categories.
Suggesting SpaceX is focused on enterprise applications
Can SpaceX turn Starlink, X, and Grok into a real position in enterprise services?
Part 2: the debates
Key points of tension: How much growth + margin is ahead for Starlink? Can SpaceX execute on its newer businesses and big bets?
Starlink
- Subscribers are growing quickly.
- Monthly revenue per subscriber is falling.
Starship
- Starship reduces launch costs and powers the next stage of Starlink.
- It still has major execution risk.
Mobile, enterprise, and government
- Increases the TAM, and improves the customer mix.
- They also need spectrum, approvals, and trust.
AI
- xAI, if it succeeds in building a leading frontier lab, can pay for the whole EV.
- Execution is paramount. Near-term, it burns a lot of cash.
Debate 1: Starlink growth versus price
Starlink is growing fast, but ARPU is falling. To grow the market, ARPU may have to fall further, which requires costs to come down as well.
Bull case
Lower price can be rational if launch, satellite, terminal, and support costs fall faster than ARPU.
Bear case
If support, ground ops, and capacity costs scale with subscribers, Starlink starts to look more like a utility than a software-margin network.
Debate 2: Will Starship work?
SpaceX needs Starship for the next stage of Starlink scale, direct-to-cell satellites, AI compute in orbit... and yes, Mars.
What needs to go right
- Full reusability
- Rapid turnaround
- High launch cadence
- Timely FAA and launch-site approvals
Why it matters
- V3 Starlink deployment gets cheaper and faster
- Direct-to-cell capacity can scale
- Orbital AI compute becomes more realistic
- Without it, growth is slower and more expensive
Debate 3: mix shift to mobile, enterprise, and government
Can Starlink become critical infrastructure for phones, companies, and governments? How large is this?
Mobile
Mobile service reached 7.4M monthly unique devices across about 30 countries at Q1 2026.
Spectrum
SpaceX agreed to buy 65 MHz of U.S. spectrum and global mobile-satellite licenses from EchoStar.
Enterprise
Use cases include aviation, maritime, hospitality, retail, fixed-site backup, and remote worksites.
What the $740B mobile TAM implies
Basically the whole world uses mobile, so say 6.2B users available. $740B per year is about $62B per month: roughly 6.2B users at $10 monthly ARPU.
How much of this is within Starlink's ideal user profile, and which segments are economic for Starlink to serve?
Debate 4: xAI adds a huge option, but also huge spending
SpaceX's AI buildout turns SpaceX into a consumer of capital.
Current financial drag
AI generated $3.2B of 2025 revenue, a $(6.4)B operating loss, and $(1.2)B of Segment Adj. EBITDA.
The most vertically integrated AI provider?
SpaceX is trying to combine X distribution, Grok, data centers, launch, Starlink, and satellites into one AI stack. And then there's Terafab.
Orbital AI compute: Yes, they're serious about it
SpaceX argues that rockets, satellite manufacturing, Starlink, and xAI infrastructure could support AI compute satellites in orbit.
Elon Musk: "In 36 months, the cheapest place to put AI will be space." A hedge against terrestrial regulation?
- As early as 2028 SpaceX expects to begin deploying orbital AI compute satellites.
- $26.5T AI TAM SpaceX sizes AI across infrastructure, subscriptions, ads, and enterprise applications.
- Starship unlocks orbital compute AI compute satellites at scale need full Starship reuse and rapid turnaround.
Musk gets paid for Mars and orbital compute
Musk's big awards only pay out if SpaceX hits valuation targets and science-fiction operating milestones.
SpaceX CEO Award
- 1.0B restricted Class B shares.
- 15 SpaceX valuation tranches from $500B to $7.5T.
- No tranche vests on valuation alone: each also needs a self-sustaining Mars colony with at least 1M people.
AI CEO Award
- 302.1M restricted Class B shares.
- 12 separate SpaceX valuation tranches from $1.065T to $6.565T.
- No tranche vests on valuation alone: each also needs non-Earth data centers capable of 100 terawatts of compute per year.
The operational milestones are inspiring for humanity, but are they aligned with public shareholders?
A recap of key investor questions
Moat
- How far ahead is SpaceX on cost, cadence, reuse, and reliability?
- Does Starlink keep widening that lead?
Starlink
- Does subscriber growth offset lower pricing?
- How much do mobile, enterprise, and government improve the mix?
Starship
- How much execution risk is there in Starship, really, and when does it become useful capacity?
- How quickly does it lower the cost of adding network capacity?
AI
- Can xAI earn a return on massive compute spending?
- Can xAI become a frontier AI lab? How much will this cost?