Kuiper looks like a disaster
Launch costs determine the economics and capabilities of satellite constellations, and Starlink's huge edge will remain for many years to come
Amazon wants to disrupt the satellite internet business with Kuiper but unfortunately you can't build disruption on imaginary services. Kuiper is not just far behind Starlink, it’s continuing to fall farther behind every year. At this point it seems very unlikely it can even launch fast enough to make the commitments necessary to retain its FAA license (1,800 satellites within two years and 8 months from now, requiring close to 100 launches). Even if it does achieve that, it has a serious structural impediment to ever being competitive, its launch costs.
The costs and capabilities of LEO satellite constellations are directly correlated with launch costs. This is because the network capabilities are directly proportional to the satellite mass in orbit. Bigger satellites are more capable than smaller satellites, and more satellites offer more capacity. Worse, in LEO it's hard to keep satellites from de-orbiting, that requires fuel to overcome atmospheric drag that can significantly reduce orbital lifetimes as high as 1,000 km(1).
Lift costs will determine who can offer the most capable and powerful satellite constellation network. Right now Kuiper is at a huge disadvantage to SpaceX/Starlink in lift costs. Its first launches are using the last Atlas V rockets, and its waiting for Ariane 6 (first flight April) and ULA Vulcan Centaur (first test flight in December) to come on line.
So Kuiper's current launch costs are probably roughly $10M per ton w/Atlas V (2), and in a few years Vulcan and Ariane 6 may be able to reduce be able to reduce it as low as $6M per ton (3)(4). But today a reusable Falcon 9 is priced at about $3.6M per ton to LEO (5). SpaceX’s cost may be half that, but I’ll use $3M for this comparison. Thats a huge difference but why is it so important? Each v1 Starlink sats weighed 1/3 of a ton and estimated to cost less than $500k to build, but $1M to launch.
This makes launch costs the dominant portion of the cost of building a satellite constellation. The v2 satellites are going to weight 1 & 1/4 tons, with a launch cost of $3.8M. First gen Kuiper satellites are going to weigh a little over a half ton, so they are probably costing close to $5M each to launch, similar to Starlink Gen2 but with half the mass they are likely far less capable.
This difference means that SpaceX spent roughly $2B launching its first 1,800 satellites, while Kuiper is going to have to spend at least $7B on launch costs for its first 1,800 satellites. And then likely replace them in 5-10 years, compounding that launch cost differential every replacement.
And worse, Amazon has been scrambling to reserve 83 launches over the next 5 years to build the barebones version of its constellation, but that aggressively assumes no delays from complex new rockets. All are using hydrogen upper stages, which is a big reason they are so expensive. But also Hydrolox is a tough fuel to handle without leaks so it tends to lead to lots of delays (SLS for example).
Falcon 9 uses a single dense easy to handle fuel for both stages, which is one reason it is so cheap, and why it has such a high cadence. Falcon 9 is about to launch its 80th launch of the year this week. It’s on pace for 100 launches this year with about 60 being Starlink launches. SpaceX has massively more launch capacity than the rest of world combined, which means every year for the foreseeable future they will be launching multiples of the Starlink satellite mass more than Kuiper can achieve.
So the popular answer to Kuipers launch costs problems is Blue Origin’s New Glenn. Its going to reuse its first stage using the same technique as Falcon 9, retropropulsive landings. And its going to be much bigger than Falcon 9, with a bigger fairing. This should make it more efficient, for similar pad and personnel costs for a single launch you’ll get significantly more mass (45 tons) to orbit. And that bigger fairing should mean they can get to use more of that potential launch capacity because often launches max out fairing space before they max out payload mass..
But New Glenn has its inefficiencies that break some of the rules that made Falcon 9 so cheap. It doesn’t use the same dense fuel for first and second stages, in fact its second stage is a Hydrolox stage. So now they are back to handling two different fuels with two widely different thermal and handling requirements. And dealing with hydrogen leaks. Falcon 9 was already cheap before they mastered reuse because each F9 used 10 merlin engines that were made using mass production techniques on assembly lines. BO should get some of those efficiencies building 7 BE-4s per stage but they won’t build as many stages starting with reusable stages. And they have to build BE-3Us for the second stage on an entirely different assembly line in much lower volumes.
Lastly, BO is late. It’s always late. It has never put anything into orbit. The BE-4 was late and has held up Vulcan’s first launch. Its unclear if New Glenn can launch in 2024 at all. It’s certainly at least 3-4 years away from operating at a cadence high enough to give Amazon costs and launch capacity on par with Falcon 9. And hanging over it like a vulture is SpaceX’s Starship.
Starship is designed to put over 100 tons into orbit at a cost per launch lower than the Falcon 9. It’s a huge rocket designed to be fully re-usable. Even if the upper stage (Starship) doesn’t achieve cost effective re-use like planned, an expendable configuration is still likely to significantly reduce launch cost per ton over Falcon 9. It will be more efficient because of its greater size and using far less costly construction material (stainless steel). And if the upper stage works, launch costs will drop below $1/2M per ton. BO is also working on a reusable upper stage for New Glenn (Jarvis), but it’s unlikely to be nearly as efficient because of its smaller size. And again BO’s track record is full of claims, nearly empty of accomplishments. Starship already has a launch attempt, meaning it’s not only more advanced and larger, but appears to be ahead of New Glenn in development.
So for the next decade at least Kuiper is very likely to have a constellation that is much smaller and lower capacity than Starlink, while costing billions more to build and maintain each year.
Note that the chart is only rough estimates, heavier, more compact (aerodynamic) objects will live a lot longer in a particular orbit than lighter objects with large solar panels. SpaceX currently plans de-orbiting Starlink sats within 5 years or so since in the near future they can be replaced by much improved hardware. I'm not sure how long that strategy will continue, if lift costs continue to decline and satellite evolution slows it will make sense to extend their life span into decades by their increasing fuel mass.
Atlas V cost roughly $109M/launch in a configuration that can put 8 metric tons into LEO ($13.5M/ton), and roughly $150M maxed out with add-on solid rocket boosters to lift 18 tons ($8.3M/ton).
Its waiting for Ariane 6 to come on line (first flight April), which claims 10 tons for $80M (A62 model, $8M/ton) and 22 tons for about $120M (A64, $5.5M/ton).
When the ULA Vulcan Centaur comes on line (first test flight in December), it will lift 11 tons for $100M ($9m/ton), and 27 tons for $200M($7.4M/ton).
Currently a reused Falcon 9 is priced at $67M to put 18 tons into orbit, or about $3.6M per ton. Independent analysis of the benefits of reusing the first stage have estimated launch costs of roughly $30M, with $15M for expending first stage and $15M for refurbishment/amortization of first stage. SpaceX has never broken these costs out publicly to confirm this as far as I know.
I've long had similar thoughts so thank you for putting this article together. Launch costs will make or break companies (Oneweb / ULA Ariane).
It's borderline fraudulent that Amazon public shareholders are subsidizing Bezo's private venture and there hasn't been more backlash. Blue Origin is years behind in development and without steady launch cycle and partner basically is DOA. They get that in Kuiper. Amazon shareholders get to pour billions into an un-competitive system to bail out their chairman's other project and will likely not see a return on their investment.