Starlink — the cash engine.
The world's largest satellite broadband network — the business SpaceX reports as its Connectivity segment. Starlink had 13.6 million consumer subscribers at Sept 30, 2026, up from 12.0 million at June 30 and 7.2 million a year earlier, per SpaceX's Q3 2026 8-K. Connectivity produced $11.4 billion of FY2025 revenue with $4.4 billion of operating income — the only segment with operating income that year — and $5,248M of revenue and $2,367M of operating income in Q3 2026 alone. See how that lands on the consolidated SpaceX earnings statement.
From 1M to 13.6M subscribers in under four years.
Starlink's most recently reported number of paying customers is 13.6 million — consumer subscribers at Sept 30, 2026, per SpaceX's Q3 2026 8-K (Oct 20, 2026). That is up 1.6 million in the quarter from 12.0 million at June 30, 2026, and from 10.3 million in Q1 2026, the last count in the S-1. For the two year-end counts people usually want: Starlink closed 2024 with roughly 5.2 million paying subscribers and 2025 with roughly 9.4 million — about 4.2 million net adds in a single year. The S-1 traces the trajectory from launch:
| Period | Subscribers | Revenue (annualized) |
|---|---|---|
| End 2020 | ~10,000 (beta) | De minimis |
| End 2021 | ~145,000 | ~$200M |
| End 2022 | ~1.0M | ~$1.4B |
| End 2023 | ~2.6M | ~$4.2B |
| End 2024 | ~5.2M | ~$7.7B |
| End 2025 | ~9.4M | $11.4B |
| Q1 2026 | 10.3M | Implied $13.5B run rate |
Subscribers and ARPU since the IPO
SpaceX now reports Starlink subscribers and ARPU every quarter. Subscribers doubled year over year in Q2 2026 and nearly doubled in Q3, while ARPU fell from $85 a month in Q2 2025 to $66 in Q2 2026 before ticking up to $69 in Q3:
| Quarter | Subscribers | ARPU / month | Year earlier (subs · ARPU) |
|---|---|---|---|
| Q1 2026 | 10.3M | $66 | — |
| Q2 2026 | 12.0M | $66 | 6.0M · $85 (Q2 2025) |
| Q3 2026 | 13.6M (consumer) | $69 (consumer) | 7.2M · $83 (Q3 2025) |
Connectivity segment revenue and operating income
The $11.4 billion of FY2025 revenue and $4.4 billion of operating income usually quoted for Starlink are the totals for SpaceX's Connectivity segment ($11,387M and $4,423M). Since the IPO, the 8-Ks also split Connectivity revenue between consumer and enterprise & government customers:
| Period | Connectivity revenue | Consumer | Enterprise & government | Operating income |
|---|---|---|---|---|
| FY2025 | $11,387M | — | — | $4,423M |
| Q1 2026 | $3,257M | — | — | — |
| Q2 2026 | $4,291M | $2,485M | $1,806M | $1,656M |
| Q3 2026 | $5,248M | $3,121M | $2,127M | $2,367M |
Sources: S-1 (FY2025 and history); SpaceX Q2 2026 8-K (Aug 4, 2026) and Q3 2026 8-K (Oct 20, 2026). Q1 2026 figures are comparatives from the Q2 8-K; “—” means not shown in the figures summarized here. The Q3 release labels subscribers and ARPU as consumer figures.
Net adds were 1.6 million in Q3 2026, per the 8-K. The residential market in the United States, Australia, and Western Europe is now mature; growth is increasingly mix-shift toward business, maritime, aviation, and Direct-to-Cell — all of which carry materially higher ARPU than the residential consumer tier.
Five tiers, very different margins.
Starlink is not one product. The S-1 segment footnote identifies five distinct product lines, each with its own ARPU, gross margin profile, and competitive setup.
| Line | Target customer | Approx. ARPU | Notes |
|---|---|---|---|
| Residential | Consumer households | $80–120 / month | The original tier. One-time hardware kit + monthly service. Largest line by subscriber count. |
| Roam (RV) | Nomadic / pause-able use | $165 / month (active) | Premium consumer line. Significantly higher ARPU than residential; many subscribers pause off-season. |
| Business | Enterprise & SMB | $250–500 / month | Enterprise SLAs, throughput priority, multi-site billing. Higher equipment costs. |
| Maritime & Aviation | Cargo, cruise, commercial & private aircraft | $1,000–10,000+ / month | The industry-changing line. Per-vessel ARPU is multiple orders of magnitude above residential. SpaceX reported 3,800+ aircraft equipped at Q3 2026. |
| Direct to Cell (D2C) | Existing LTE handsets via MNO partners | Revenue split with MNO | Launched 2024 with T-Mobile US as anchor. SMS first, voice and data rollout in progress. |
ARPU bands are illustrative — the 8-Ks report consumer ARPU ($69 in Q3 2026) but not per-line ARPU. Bands are inferred from published price points and known mix.
9,600+ satellites and counting.
Starlink operates in three FCC-approved orbital shells. The S-1 discloses the current active count at 9,600+, with two future expansion bands authorized:
- Generation 1 (V1.0 / V1.5). Original satellites at 540–570 km altitude. Mostly being deorbited or replaced as the V2 build-out scales.
- Generation 2 (V2 Mini and V2 full). Heavier satellites with significantly more downlink capacity, plus Direct-to-Cell payloads. Currently the bulk of monthly launch cadence.
- Generation 3 (V3). The next-generation satellite design, optimized for Starship's full payload capacity. Starship Flight 14 in late Sept 2026 — the first orbital Starship flight — deployed 26 Starlink V3 satellites. Deployment at scale depends on Starship reaching routine service.
Most launches are Falcon 9 ride-shares with Starlink as the only payload, in batches of 20–24 V2 Mini satellites or smaller batches of full V2. The S-1 explicitly notes that Falcon 9 is the gating factor on V2 deployment — and that V3 satellites cannot be deployed via Falcon at their planned size, which is part of why Starship matters so much to this segment.
For the launch vehicle side see Falcon & Dragon and Starship.
It's not the satellites. It's the launches.
The competitive moat is unusual for a connectivity business. It is not just the satellites — it is the cost of putting them up there. Starlink uses Falcon 9 to deploy its own constellation. Every competitor has to buy launches from a third party:
- Amazon Project Kuiper has contracted Atlas V, Vulcan, Ariane 6, and New Glenn launches. None of those are reusable at Falcon 9's cadence and price point.
- OneWeb (now Eutelsat OneWeb) historically relied on Soyuz and now Falcon 9 itself — paying SpaceX for the privilege.
- Telesat Lightspeed has contracted Falcon 9.
- AST SpaceMobile (ASTS) contracts Falcon 9 and Indian launch providers.
When competitor satellites cost $10–40M each on third-party launches, and SpaceX's marginal cost per Starlink V2 launch is dominated by reusable Falcon 9 economics that SpaceX itself controls, the unit-economics gap is structural and difficult to close. See our comparisons page for the full peer table.
Operating margin trajectory
The Connectivity segment earned $4,423M of operating income on $11,387M of revenue in FY2025, and $2,367M on $5,248M in Q3 2026. Three drivers the bull case points to for further margin expansion:
- Mix shift to higher-ARPU lines (business, maritime, aviation, D2C).
- Constellation amortization: satellites are depreciated over 5 years; the early-generation hardware is rolling off the depreciation schedule.
- Falcon 9 cost-per-kg continues to fall as reusability is extended.
Bull-case modeling has segment operating margin reaching 45–50% by FY2027 — a margin profile closer to a hyperscaler than a satellite operator.
What can break the Starlink story.
- Spectrum disputes. Starlink operates under FCC and international ITU spectrum coordination. EchoStar and Dish have separately filed objections to Starlink's terrestrial spectrum requests, and ITU member states can challenge constellation expansions. Spectrum disputes are slow and political.
- Orbital debris & capacity. Low-Earth orbit is finite. The FCC's most recent constellation orders include orbital debris mitigation conditions; future Starlink expansions require regulatory green-lights that are not guaranteed.
- Subscriber saturation. Growth has held above 30% YoY despite the law of large numbers, but the residential market in mature geographies is no longer the bulk of net adds. Maintaining growth depends on mix-shift segments executing.
- Direct-to-Cell economics. The MNO revenue-share model is favorable to the MNO, not Starlink. Headline subscriber count grows fast in this line but unit economics are still being defined.
- Geopolitical exposure. Starlink is unavailable in China, Iran, and Russia; service in Ukraine has been politically prominent. A larger geopolitical flare-up could force pricing or service decisions with material P&L impact.
- Competition at scale. Kuiper's full deployment is targeted for 2027–2029. By the time the constellation is operational, Starlink V3 should be deployed via Starship — but execution risk on both sides is real.
See the full risk factors page for cross-segment risk framing.