Two numbers travelled together across the timeline this week. Starlink has signed 48 airlines covering more than 7,000 aircraft — about 23% of the world's active commercial fleet. And only about 1,400 aircraft actually have it installed and working.
The pairing is doing real analytical work: it says the aviation business is four-fifths pre-revenue, and that the interesting question is conversion rather than size. That framing is right. Both numbers are wrong in ways that matter, and neither is what a reader would assume from how they are being quoted.
What the 7,000 actually counts
The original announcement language is careful, and almost every repost drops the careful part. Starlink's contracts cover more than 7,000 aircraft "that are already equipped, undergoing installation, or under contract."
That is three categories in one figure, and the first of them is the installed base. So 7,000 is not a backlog sitting behind 1,400 already flying — the 1,400 is inside the 7,000. The remaining conversion pool is roughly 5,600, not 7,000, and anyone subtracting the two numbers as though they were separate is double-counting the fleet that already works.
The figure is also moving faster than the reposts. Coverage from July put it at over 7,500 aircraft across 46-plus airlines, and the book kept growing this week: American Airlines announced it will install Starlink on more than 500 narrowbody Airbus aircraft beginning in Q1 2027, a commitment that lands entirely in the contracted column and not one aircraft of which will be flying before next year.
The fleet-share claim, at least, holds up. IATA counted 30,300 active commercial aircraft in June 2025, within a total fleet of 35,550 including stored airframes. Seven thousand against 30,300 is 23.1% — so "roughly a quarter of the world's active fleet is contracted" is fair, provided the word contracted survives the retelling. It usually does not.
The installed number is a 2025 figure
Here is the part that should stop anyone building an argument on the pair.
There is no current, disclosed count of Starlink-equipped commercial aircraft. The "more than 1,400" that everyone is quoting traces to SpaceX's own 2025 progress report. It was a company statement, it was accurate when published, and it is now the better part of a year old — during which the programme roughly quadrupled its airline count, from two partners at the end of 2025 to 48 today.
What exists instead is per-airline tracking, and it is partial. One public tracker covers 671 equipped aircraft across just three airlines. That is not a global total; it is three carriers that happen to be well documented. Individual counts are firmer:
| Airline | Equipped | Share of fleet | Note |
|---|---|---|---|
| United | 530 | 32% | +53 in the trailing 30 days |
| Alaska | 99 | 28% | All 90 regional E175s done; mainline later in 2026 |
| Hawaiian | 42 | 69% | Rollout complete — all A330 and A321neo |
United alone is now well beyond a third of the entire 1,400 figure, and Qatar has completed its 777 fleet with A350s in progress while Southwest has stated an intent covering more than 300 aircraft. The true installed base today is materially above 1,400. Nobody publishes what it is.
That matters more than it first appears. An argument built on "1,400 of 7,000" is comparing a stale company figure against a current one, and it will overstate the gap it exists to measure. The gap is real. Its size is not knowable from public sources to better than a few hundred aircraft.
Conversion is measured in years, and that is the whole story
The useful number is not the book. It is the rate at which contracted aircraft start billing, and for the first time this week there is enough disclosure to see the shape of it.
- United is adding roughly 53 aircraft a month and targets somewhere between 800 and 1,000 equipped by the end of 2026, against a 1,817-aircraft fleet.
- Lufthansa Group put its first Starlink A320neo into passenger service on August 19, with a rollout to roughly 850 aircraft by 2029 — one aircraft to 850 in four years.
- American starts in Q1 2027 on 500-plus narrowbodies.
- Emirates targets 150 aircraft by the end of 2026; Royal Air Maroc its full fleet by 2027.
United is the fastest-moving operator in the programme and it is the exception, not the pace. Lufthansa is the honest base case: a large European group with a four-year installation schedule, because aircraft downtime, supplemental type certificates and sub-fleet prioritisation set the speed, not demand and not satellite capacity. Published rollout schedules across the programme stretch to 2029.
So the aviation book converts to revenue over roughly four years, in a queue governed by maintenance slots. That is a very different asset from a signed contract, and it is the reason the "23% of the world's fleet" framing flatters the business.
What an aircraft is actually worth, and what nobody will tell you
The revenue question has a hard ceiling and a soft floor.
The ceiling is public, because SpaceX raised it three weeks ago. In August 2026 Starlink doubled its business-aviation pricing: Aviation Global Unlimited went from $10,000 to $20,000 a month, a new Aviation Regional Unlimited tier landed at $12,500 a month, the Regional 25GB plan doubled to $4,000, and the hardware and installation package rose from $145,000 to as much as $200,000. Existing business-aviation customers move to the new plans from August 7.
Two things follow. First, in a year when the price of nearly everything adjacent to AI and bandwidth has been falling, Starlink doubled a list price and kept selling — which is a genuine signal about how contested this market is not. Second, and less convenient for the bull case: those prices are for private and business jets, and they do not apply to airlines.
Commercial carriers sign opaque enterprise contracts, and the service reaching passengers is free — Lufthansa's is free with Miles & More or a Travel ID, across every cabin class, and United, Emirates and Qatar are the same. The airline pays a bulk rate for the aircraft. That rate is not public at any carrier, and it is certainly well below the business-jet list, because a 500-aircraft narrowbody order does not price like a single Gulfstream.
This is where most analysis of Starlink aviation quietly invents a number. We are not going to. What can be said honestly is that the business-jet list price is an upper bound — no airline pays more per aircraft than a private owner does — and that the real figure is somewhere below $20,000 a month and above zero. Any revenue estimate for this business is a statement about where in that range the contracts sit, and nobody outside SpaceX and the airlines knows.
Does any of this change our model? No — and the size of the no is the point
Our SpaceX model does not carry an aviation line, and after working through the numbers above, it should not.
Aviation sits inside the Starlink vertical's otherRevenue — a single quarterly figure covering enterprise, maritime, aviation and rail together. It is worth being explicit that this figure is derived rather than disclosed: 12.0 million subscribers at $66 a month for three months is $2,376M, and SpaceX's reported Starlink revenue of $4,290M less that leaves $1,914M. Nobody at SpaceX published $1,914M. We subtracted it.
Bound aviation inside that residual using the only ceiling that exists — no airline pays more per aircraft than a business-jet owner, so $20,000 a month is a hard maximum and the real contract rate is well below it:
| Per aircraft, per month | ~1,400 installed | Share of otherRevenue |
~5,600 remaining | Share of otherRevenue |
|---|---|---|---|---|
| $3,000 | $12.6M/qtr | 0.7% | $50.4M/qtr | 2.6% |
| $8,000 | $33.6M/qtr | 1.8% | $134.4M/qtr | 7.0% |
| $15,000 | $63.0M/qtr | 3.3% | $252.0M/qtr | 13.2% |
Now set that against what the model already assumes. otherRevenue grows 9% a quarter, which adds $172M in the first projected quarter alone. So the entire remaining aviation book — every one of the 5,600 contracted aircraft installed and billing, four years of installation work delivered at once — is worth between 0.29 and 1.46 quarters of growth the model has already booked.
The valuation says the same thing more bluntly. Against a base case of $187.12 a share:
| Change to the model | Base fair value | Delta |
|---|---|---|
| Whole remaining book billing today at $15,000/mo | $191.85 | +$4.73 |
| …at $8,000/mo | $189.65 | +$2.52 |
| …at $3,000/mo | $188.07 | +$0.95 |
| Remove all of today's installed aviation, at $15,000/mo | $185.94 | −$1.18 |
otherGrowthQoQ 9% → 11% |
$201.34 | +$14.22 |
otherGrowthQoQ 9% → 7% |
$176.82 | −$10.31 |
A two-point change in the growth rate of the line aviation sits inside is worth three to fifteen times the entire aviation book. For aviation to matter as much as one year of that growth, all 7,000 aircraft would each have to pay about $37,500 a month — nearly double the $20,000 global-unlimited list price for a private jet, on contracts that are cheaper than a private jet's, for a service the passenger gets free.
That is the reconciliation, and it is a quiet result rather than a dramatic one: the model is in line with the data. Not because it forecast the aviation programme — it never modelled it separately — but because the programme is too small to reach the model at any defensible price. The 7,000 aircraft are real, the 23% of the world's fleet is real, and both are immaterial to what SpaceX is worth. What is material is whether the enterprise line as a whole keeps compounding at 9%, and aviation is one of four businesses inside it.
The honest change is a note rather than a number: the Starlink vertical's rationale should record that aviation is bounded at roughly 3% of otherRevenue today and cannot move this model on any pricing the public record supports. If SpaceX ever discloses an aviation line, or an airline discloses a contract value, that bound gets replaced by a fact — and this section is the thing to check it against.
What to watch
- Whether SpaceX ever discloses an installed aircraft count in a filing. It is now a public company. A current equipped figure would replace a 2025 press number that the whole market is still quoting, and it is the single most useful disclosure the aviation programme could make.
- United's run rate through year-end. It is the only carrier moving fast enough to measure, and 53 a month against a target of 800–1,000 by December is the test of whether installation capacity is the constraint everyone assumes.
- Whether any airline discloses contract value. One disclosed per-aircraft rate would collapse the entire estimation problem above into arithmetic. Until then it stays a range.
- The 2027 starts. American in Q1 2027 and Royal Air Maroc's full fleet are the first test of whether the contracted book converts on schedule or slips, which is the only question that turns 7,000 aircraft into revenue.
The 48-airline and 7,000-aircraft figures are from Starlink's August 2026 announcement as carried in press coverage; the "over 7,500 across 46+ airlines" figure is from July 2026 reporting. The "more than 1,400 equipped" figure is from SpaceX's 2025 progress report and is not a current count; no current aggregate is published. Per-airline counts are from public trackers as of August 18–19, 2026 — United 530 aircraft and +53 over the trailing 30 days, Alaska 99, Hawaiian 42 — and are third-party tallies rather than airline disclosures. Fleet totals of 30,300 active and 35,550 including storage are IATA's, as of June 2025. Starlink Aviation pricing is the schedule effective for new customers in August 2026, with existing business-aviation customers migrating on or after August 7. American Airlines' 500-plus narrowbody commitment beginning Q1 2027 is from the company's own newsroom; Lufthansa's first-aircraft service date and roughly 850-aircraft, 2029 programme reached us through social posts rather than a company release and should be treated as the least firm figure here. SpaceX's Q2 2026 Starlink revenue of $4,290M is the figure stored on this site, from its Q2 2026 release.
