On October 15, various news outlets, including SpaceNews, reported that SpaceX submitted filings for up to an additional 30,000 Starlink satellites—on top of the ~12,000 satellites the company already filed to deploy and late last year received regulatory approval for. The filings were by way of 20 individual coordination filings to the ITU, by way of the U.S. FCC, proposing a system having 11 discrete orbital altitudes ranging from 328.3 km to 580 km (22,500 of the 30K are between 328 and 499 km).
SpaceX has not provided a rationale for why an additional 30,000 satellites might be needed or desired. One analyst speculated the filing could be intended to gum up the ITU and create FUD amongst competitors. Interesting speculation, indeed, but our immediate question was, “Is this real?”
This question set in motion certain ideas and a few thought experiments. It’s interesting to see by momentarily accepting these astronomical constellation sizes where some of the logic leads.
Is it real?
There is a likelihood that rules around NGSO “mega-constellation” coordination are soon set to change, possibly as soon as the upcoming World Radio Congress (WRC) starting later this month. With little clarity on how the rules will evolve, it is perhaps prudent for SpaceX to “plant a flag” for maximum optionality in terms of future system design, orbital altitude, and architecture. Further, erecting early competitive blocks wherever possible makes sense in light of SpaceX’s potential LEO broadband competitors (even creating perceived competitive blocks tends to inhibit fundraising efforts, on the margin).
However, we are incredulous a 42,000-satellite system is actually envisioned, let alone would be constructed anytime in the next decade. There are a variety of reasons for skepticism, not the least of which being demand dynamics and high cost. (That said, SpaceX is nothing if not ambitious.)
For example, admittedly using grossly simplified assumptions, if one assumes 5 Gbps of typical throughput realized per Starlink satellite (highly conservative versus theoretical downlink capacity of 17-23 Gbps per Starlink satellite), then 30,000 incremental satellites represents additional system capacity of +150 Tbps. For reference, total international terrestrial Internet capacity (worldwide) currently stands at 466 Tbps. In other words, while bandwidth demand and its growth are seemingly insatiable, 42,000 satellites would saturate the globe with a shockwave of new bandwidth supply.
Meanwhile, cost and production volume factors also come into play. Assuming a nominal on-orbit spacecraft life of 3 years, a system of 42,000 satellites would require production of ~40 satellites per day (i.e., ~30 per day for the 30,000). This requires a production volume that will need to take cues from higher-volume industries (automotive anyone?). And putting aside the enormous ground system costs, the spacecraft cost for 30,000 incremental satellites, even factoring in production economies of scale, is likely in the ~$10 billion range.
How could SpaceX even contemplate this plan? Launch?
Another thought experiment derived from accepting this unrealistically high number of SpaceX satellites involves how the company can even afford to deploy an additional 30,000 satellites to an already-large constellation.
Though we aren’t privy to the details of SpaceX’s plans (although it would be nice), regardless of the ultimate constellation size, we strongly suspect that the company’s new Starship rocket (which we covered in an earlier Research Flash) must play a key role in SpaceX’s Starlink vision. In addition to its massive lift capacity (150 tons to LEO), the key feature that sets Starship apart is the fact that the rocket is fully reusable. In contrast, the partially reusable Falcon 9 must replace the second stage after every launch.
If we ignore the Starship’s one-time construction cost ($89-177 million), development cost ($2-3 billion) and all fixed overhead (launch facility, solar farm, etc.) to focus solely on the Starship’s incremental launch cost, the (highly speculative) conclusion is quite jarring.
- 42,000 Starlink satellites would require 636 Falcon 9 launches, at 66 satellites per launch.
- Assume fuel and non-reusable second stage cost $20 million per launch (note – the Falcon 9 consumes ~$200,000 of kerosene per flight)
- Based on just that $20 million, it would cost SpaceX ~$13 billion to launch the full constellation
- 42,000 Starlink satellites would require 93 Starship launches, assuming 450 satellites per launch
- Starship and Super Heavy are 100% reusable
- Assuming SpaceX builds a solar farm (as Elon Musk has indicated), the marginal cost of propellant production (oxygen and methane) is nominal – making the launch overhead costs, which are comparatively small, the most significant contributor to cost
In turn, Starship launches to deploy the whole Starlink could cost something on the order of 1% of the costs of using the already-miserly Falcon 9 and perhaps 50% less than a single Ariane 5 launch. In other words, Starship is a game-changer for launch – and satellite system cost economics as a whole.