Value chain · CLM-0877
In plain terms
A US regulator's order repeats a spaceflight news site's count of more than 3,500 Starlink satellites launched by October 2022. If that count is right, doubling from one unit to 3,500 takes nearly twelve doublings, and the company must have built at least as many as it launched, so that is a floor rather than an estimate - but the number is a journalist's, carried by the regulator rather than established by it.
The claim as recorded
An FCC order of December 2022 states, CITING A SPACEFLIGHT TRADE PUBLICATION AND NOT A LICENSEE FILING OR A COMMISSION FINDING, that as of October 2022 SpaceX had launched more than 3,500 of its Gen1 Starlink satellites. IF that count is not overstated, then since log2(3,500) is 11.77 (our arithmetic) and cumulative units PRODUCED is greater than or equal to units launched - units lost, deorbited or never flown still consumed the learning - at least 11.7 doublings of cumulative units, counted from a single unit, had been completed by October 2022. It establishes no upper bound and says nothing about any later date.
Evidence behind this: CONTENT T5, CHANNEL T3, AND THE CONTENT GOVERNS. The order is a government document read directly, which would be T3, and its AUTHORISING limb is T1 as a normative instrument. But the sentence this record cites is neither: footnote 16 sources the 3,500 figure to a spaceflight trade publication, NASASpaceFlight.com, 20 October 2022. So the number is trade press arriving through a regulatory channel, and the Commission's imprimatur does not launder a news count into a regulatory finding. This is the SRC-197 content-versus-channel distinction running the other way: there a T1 instrument reached the register through a non-T1 republisher; here a T5 fact reaches it through a T3 document. THE STRONGEST COUNTER-ARGUMENT IS ANSWERED HERE RATHER THAN LEFT FOR SOMEONE TO REDISCOVER. It is not the weak form, that a regulator adopting a figure lends it standing. It is this: the Commission asserts the proposition IN ITS OWN VOICE - "SpaceX has launched more than 3,500", not "SpaceX reports that" - and it CHANGED THE PROPOSITION'S SHAPE, since the cited article counts a point value, the 3,500th satellite, while the order states a FLOOR. A floor is something the agency that licenses these satellites could assert safely from its own records whatever the article's precision, and that conversion is the Commission's. The answer is narrower than "channel does not launder content": the NUMBER is the article's and the Commission performed no count, so its own-voice framing is endorsement rather than measurement, and a background recital is not a finding. The citation form supports this - footnotes 10 to 15 and 17 cite Commission orders by FCC Rcd, grants and filings by IBFS File No., petitions for reconsideration and one court decision, and footnote 16 is the only trade-press citation among them, so the Commission itself distinguishes its regulatory and legal record from what it mentions.
Scope and limits
A FLOOR AT ONE DATE, AND NOTHING ELSE. Four things this record does NOT establish. First, ANY FIGURE FOR 2026: SRC-438 cannot speak past October 2022, and no later filing is cited. Second, AN UPPER BOUND: "more than 3,500" bounds below only, and the larger numbers in the same document - 7,500 granted, 7,518 V-band, 4,408 Ku and Ka band Gen1 - are AUTHORISATIONS and not counts of anything built. Third, A PRODUCTION COUNT: launched is a lower bound on produced, not a measurement of it, and no first-party production figure has been found by any route. Fourth: A REGULATORY FINDING. The number is not one. It sits in Section II BACKGROUND, in the paragraph reciting Gen1's regulatory history, and "3,500" occurs exactly twice in the whole document - there and inside the cited article's title. Nothing in the order's analysis or its ordering clauses turns on it, so the figure is load-bearing for nothing the Commission decided. It entered the order from a news article, so a reader following this citation reaches trade press with a regulator's cover sheet on it. The floor is stated as 11.7 and not rounded up: by our arithmetic 2 to the 11.8 is 3,566, which "more than 3,500" does not reach, so 11.8 would not be a floor.
What it assumes
That the trade-press count of 3,500 is NOT OVERSTATED. Only an overcount breaks the floor; an undercount leaves it untouched, so assuming the count accurate would concede a fragility this record does not have. Nothing here checks even the weaker version: no filing, no Commission finding and no independent count has been read. That doublings are counted FROM A SINGLE UNIT, which is the whole magnitude and was unstated: log2(3,500) is 11.77 from one unit and 5.87 from a first batch of sixty, so a reader taking a different baseline gets a different answer with no warning. basis records the convention; this field now names it as an assumption. That Gen1 is ONE PRODUCT for learning-curve purposes - v0.9, v1.0 and v1.5 are materially different vehicles, and doublings of a heterogeneous population are not doublings of a product. That "launched" counts satellites placed in orbit rather than launch attempts, which the cited article's own title supports by counting a 3,500th satellite. That the order's "Gen1" qualifier, which the cited article does not use, is correct. That cumulative production is at least equal to cumulative launches. And that cumulative production at October 2022 is the right denominator for a learning curve the surrounding claims apply to later dates.
Where it comes from
, read in the held file. The sentence splits across PDF pages 4 and 5 with footnotes 10 to 15 intervening, and reconstructs as "As of October 2022, SpaceX has launched more than 3,500 of its Gen1 Starlink satellites." Its footnote 16 is the provenance and it is a news article: the note cites a piece titled "SpaceX Launches the 3,500th Starlink Satellite" on NASASpaceFlight.com, 20 October 2022. The Gen1 attribution is the order's, not the article's, whose title says only "Starlink"; it is supported from inside the document by footnote 17, which states SpaceX "has not launched any of the 7,518 V-band Starlink satellites". Arithmetic: log2(3500) = 11.773.
Basis: dimensionless (doublings of cumulative units)
Related records
- Answers question
- QST-041 not published in this release
Cite this
energyaperture.com/c/clm-0877
CLM-0877, Energy Aperture, as of 2022-10.
Connected records
Other published records linked to this one. A list, not a judgement.