The MoatWide moat

SpaceX (SPCX) — moat facet

SpaceX owns the road to orbit — reusability made it the world's launch monopoly and Starlink its first mass-market product — a genuinely wide moat with an erratic profit record, priced at about 91 times sales.

SpaceX is the most formidable competitive position built in the hard, capital-hungry world of atoms in a generation, and it went public in June 2026 at a valuation — briefly north of two trillion dollars — that assumes the moat is not only real but nearly bottomless. Both halves of that sentence deserve a sober look. The moat is genuine, and it rests on a thing the aerospace establishment spent fifty years insisting was impossible: a rocket that lands, is refurbished, and flies again. By making the most expensive part of a launch reusable, SpaceX drove the cost of putting a kilogram into orbit down by an order of magnitude, and it did so a full decade before any serious rival even landed a booster. That is not a feature a competitor copies over a weekend; it is a decade of failures, telemetry, and hard-won manufacturing know-how baked into a culture.

Segment income from operations, 2025 ($M)+$4,423Connectivity-$657Space-$6,355AISpaceX IPO prospectus (Form 424B4), segment note and key business metrics
One segment earns the money: Connectivity made $4.4 billion in 2025 while Space and AI lost $7.0 billion between them.

From that cost advantage flows the first pillar of the moat: cadence. Because each launch is cheap and each booster flies many times, SpaceX launches more mass to orbit than the rest of the world — governments and companies combined — put together, and by a wide margin. Cadence is itself a moat, because it feeds on itself: more flights mean more data, faster refurbishment, lower cost, and a launch manifest so reliable that customers who need to reach orbit have nowhere else serious to go. NASA, the Pentagon, and nearly every commercial satellite operator on earth are, to a first approximation, captive.

The second pillar is the one that now pays the bills: Starlink. SpaceX used its own cheap launch to do the one thing its rivals could not afford — loft a constellation of thousands of satellites into low orbit and sell broadband from the sky. By the time it went public, Starlink was serving over ten million subscribers and generating the majority1 of the company's revenue, and it has the classic shape of a network business: a first mover with the orbital shells and the radio spectrum staked out, scale economics no latecomer can match, and a service that works in places no cable or tower ever will. The rocket company quietly became a telecom, and the telecom is where the profits are inflecting.

Underneath both pillars sits the real, unglamorous engine: vertical integration and a rate of iteration that has no equal in aerospace. SpaceX designs and builds nearly everything itself — engines, avionics, the satellites — and it iterates on hardware at a pace that makes traditional contractors look geological. That speed is a cultural asset, not a patent, which is precisely what makes it so hard to copy: you cannot buy it, you have to become it.

Now the sober half. SpaceX's profit record is thin and erratic: on the combined basis of its filings it lost $4.6 billion in 2023, earned $791 million in 2024 and lost $4.9 billion in 2025 on $18.7 billion of revenue, though it threw off positive adjusted cash earnings and the Starlink unit is solidly in the black.2 It burns staggering sums building Starship, launching satellites and, since the xAI merger, building AI data centres: capital spending was $20.7 billion in 2025 and $28.5 billion in the first half of 2026 alone,34 and it is bound tightly to one extraordinary and divisive founder whose attention is spread across several companies and the political arena. And the stock, at about ninety times trailing sales, prices a great deal5 of a future that has not yet arrived. So the honest appraisal is this: the competitive moat is one of the widest I can point to in any industry — but the price paid for it leaves no room for the ordinary disappointments that even wonderful businesses eventually deliver. The castle is real. The question is what you paid at the gate.

Moat trajectory: Widening

On balance this moat is still widening, and briskly. The reuse cost lead extends with every flight while rivals are only now learning to land a booster; Starlink keeps compounding past ten million subscribers and its economics are turning from cash-furnace to profit-engine; and Starship, if it lands, widens the lead by another order of magnitude. The one place the water pushes back is the founder — his attention is divided and his controversies bite — but the competitive castle keeps growing faster than that risk erodes it.

The number that tests this moat
Third-party estimate
Share of worldwide mass launched to orbit
~80%+

The overall moat is a near-monopoly on access to space, and the cleanest proof is the tally: in recent years SpaceX has lofted the large majority of all the mass humanity put in orbit — more than every other company and country combined. (Note: unlike the other companies here, there's no meaningful ROIC — SpaceX still runs a GAAP loss.) Watch this share erode as Blue Origin and China ramp.

Launch-cadence and share figures from third-party launch trackers; SpaceX's first 10-K is not yet filed.
Source: Third-party launch trackers (estimate) ↗
Aspects of the moat
DIG DEEPER 01
The Reusability Cost Advantage
A rocket that lands, refuels, and flies again — the cost revolution every rival is still chasing a decade later.
DIG DEEPER 02
Launch Cadence & Orbital Dominance
More mass to orbit than the rest of the world combined — a near-monopoly on the road to space.
DIG DEEPER 03
The Starlink Network
Ten million subscribers beamed from low orbit — 61% of revenue, and the first product that could pay for the whole dream.
DIG DEEPER 04
Vertical Integration & Engineering Velocity
Build almost everything in-house and iterate faster than anyone — velocity as the true competitive engine.
References
  1. ReportedBy the time it went public, Starlink was serving over ten million subscribers and generating the majority of the company's revenue, and it has the classic shape of a network business: a first mover with the orbital shells and the radio spectrum staked out, scale economics no latecomer can match, and a service that works in places no cable or tower ever will.
    SpaceX IPO prospectus (Form S-1 / 424B4) and FY2025 disclosures — revenue ~$18.7B (+~33%), GAAP net loss ~$4.9B, positive adjusted EBITDA; Starlink >$11B of revenue (the majority) and 10.3M subscribers (Mar 2026), the segment operating profitably — FY2025 / IPO June 2026 · publ. June 2026 · source ↗
  2. ReportedSpaceX's profit record is thin and erratic: on the combined basis of its filings it lost $4.6 billion in 2023, earned $791 million in 2024 and lost $4.9 billion in 2025 on $18.7 billion of revenue, though it threw off positive adjusted cash earnings and the Starlink unit is solidly in the black.
    SpaceX IPO prospectus (Form 424B4, 11 June 2026) - net income (loss) $(4,628)M, $791M and $(4,937)M for 2023-2025 (2024 aided by a $549M tax benefit and $985M of other income); mass to orbit 1,210, 1,699 and 2,213 metric tons (customer payloads 205, 282 and 312; internal 1,005, 1,418 and 1,901); launches 98, 138 and 170; Starlink subscribers 2.3M, 4.4M, 8.9M and 10.3M (March 2026); ARPU $99, $91, $81 and $66; NASA figures of $18,500 per kilogram historical average, about $2,700 for the first Falcon 9 (2010, about 85% less) and about $1,400 for the first Falcon Heavy (2018); Customer A 25.2%, 24.2% and 20.9% of revenue, across all three segments; backlog $27,621M at 31 March 2026; about 650 V1 Mobile satellites, about 30 MNO partners covering about 1.9 billion people; no inter-segment revenue on internal constellation deployments; Starship expected to begin payload delivery to orbit in 2H 2026, with next-generation V3 and V2 Mobile satellites dependent on it — FY2023-FY2025 and Q1 2026 · publ. June 11, 2026 · source ↗
  3. ReportedIt burns staggering sums building Starship, launching satellites and, since the xAI merger, building AI data centres: capital spending was $20.7 billion in 2025 and $28.5 billion in the first half of 2026 alone, and it is bound tightly to one extraordinary and divisive founder whose attention is spread across several companies and the political arena.
    SpaceX IPO prospectus (Form 424B4), segment note - revenue: Space $3,557M / $3,796M / $4,086M, Connectivity $3,869M / $7,599M / $11,387M, AI $2,961M / $2,620M / $3,201M, total $10,387M / $14,015M / $18,674M (2023-2025); segment income from operations Space $(1)M / $21M / $(657)M, Connectivity $469M / $2,006M / $4,423M, AI $(3,973)M / $(1,561)M / $(6,355)M; capital expenditures $4,415M / $11,163M / $20,737M; Starlink ARPU $99, $91, $81 — FY2023-FY2025 · publ. June 11, 2026 · source ↗
  4. ReportedIt burns staggering sums building Starship, launching satellites and, since the xAI merger, building AI data centres: capital spending was $20.7 billion in 2025 and $28.5 billion in the first half of 2026 alone, and it is bound tightly to one extraordinary and divisive founder whose attention is spread across several companies and the political arena.
    SpaceX Q2 2026 results release (Form 8-K exhibit 99.1, 4 August 2026) - revenue $7,814M (+92%); Space $962M, Connectivity $4,291M, AI $2,561M (Q2 2025: $746M, $2,588M, $737M); net loss $541M against $1,008M; adjusted EBITDA $3.5B; AI segment adjusted EBITDA $1,146M against $(609)M in Q1, operating loss $1,257M; capex $18,369M ($15,828M AI; $2,825M a year earlier), $28,476M in the first half; nameplate compute 1.4 GW from 1.0 GW; cloud services agreements of $14.1B contracted sales adding $1.6B of AI infrastructure revenue; advertising $367M and AI solutions & infrastructure $2,194M (Q2 2025: $426M and $311M); Starlink subscribers 12.0M (+1.7M in the quarter, 6.0M a year earlier), ARPU $66 against $85; Connectivity revenue +66%; launches 38 against 46 and 78 in the half against 84; mass to orbit 485 t against 652 t; Starship Flight 12 (May) and Flight 13 (July, 20 production V3 satellites deployed); agreement to acquire Cursor for $60B; IPO net proceeds about $85.7B; $25B bond; $100B of cash and marketable securities; backlog $47.5B — Q2 2026 · publ. August 4, 2026 · source ↗
  5. Third-party estimateAnd the stock, at about ninety times trailing sales, prices a great deal of a future that has not yet arrived.
    Market data (stockanalysis.com) - SpaceX closed at $154.72 on 22 September 2026, market capitalisation about $2.10 trillion, about 91 times trailing revenue of $23.04 billion — 22 September 2026 · publ. September 23, 2026 · source ↗
Sources
Generated September 23, 2026