Kuipersats Will Be Ready, But What About the Rockets?

There appeared to be much breathless coverage of Amazon’s announcement this last week that the company would build a satellite processing facility near the Kennedy Space Center in Florida. Why all the excitement for a large warehouse? It’s not as if Amazon isn’t familiar with building massive, sprawling structures that can process things for shipping. The $120 million investment for the satellite processing facility isn’t even the most that Amazon has spent on distribution centers. This one in Alabama, for example, was built for $325 million.

So, okay, Amazon is building yet another of its many large structures in Florida–but this time, it’s different because it’s for satellites–specifically Kuiper internet relay satellites. The company hopes to get it up and processing around January 2024. At full steam, Amazon hopes the facility can push through enough Kuipersats to support launching 80 monthly.

Of course, the inevitable question arises: launch with what?

A Testing Time

The answer to that appears effortless. Amazon contracted 77 launches among three launch service providers. Aside from Atlas V’s, none of the rockets that will launch Amazon’s Kuipersats have been test launched. Not once. Nary a one. Nada. Zero and zilch.

Maybe ULA will launch its Vulcan before the end of 2023. It could be that Arianespace finally launches an Ariane 6 in early 2024. And Blue Origin with New Glenn? Maybe sometime after that–the company has just embraced that whole being “shrouded in uncertainty” thing. Frankly, those projected first launches should be taken with a grain of salt.

Such uncertainty should be causing some sweatworthy moments for Amazon’s Kuiper group. The company hasn’t even been able to launch its two test satellites. Ideally, the test satellites should be launched and tested before Amazon starts deploying satellites in groups of 40 or so. The company initially had its two demonstration satellites contracted with ABL Space (two launches–1 per satellite). Then it switched to ULA and its first Vulcan launch when it appeared ABL wouldn’t be ready.

But the two satellites haven’t been launched and deployed yet, which implies that even if Amazon’s satellite manufacturing facility in Kent, Washington, and its processing facility in Florida can get 80 satellites out per month, the company may have to wait a while as it tests its two satellites. It would make sense not to manufacture more satellites until the two demonstration satellites get thoroughly tested. Of course, it would make sense to use the most available and currently operational launch service provider, too–but Amazon went with hope and imagination instead for its contracts.

Learning From The Competition

But that gets us to the comparisons with Kuiper’s competitor–Starlink. SpaceX first deployed its two demonstration satellites in February 2018. Called TinTin, they looked nothing at all like the flatpack Starlink satellites SpaceX uses today. The initial deployment of the 60 operational Starlink satellites occurred nearly a year and a half later. Again, the differences between the two TinTin satellites and Starlink were significant.

It’s doubtful that the operational Kuipersats will undergo as much of a change, but since SpaceX has set a precedent, perhaps they will. If they don’t, my initial estimate of Kuipersats having a mass of ~500 kg might hold true (which will come into play further in this analysis).

Amazon can’t wait a year and a half from demonstration satellites to operational satellite deployments. If it did, it would only have another year and a half to get half its initial Kuiper constellation into orbit. That’s 1,618 Kuiper satellites in orbit by July 2026. Amazon must do so to fulfill the Federal Communications Commission (FCC) requirements.

Compare that requirement with SpaceX’s Starlink deployment activities. In 2019, SpaceX conducted two Starlink-dedicated launches, deploying 120 satellites that year. In 2020, the company deployed 833 Starlinks (using ~14 launches). SpaceX deployed 989 Starlink satellites by the end of 2021. The company took 2.5 years to get nearly 2,000 satellites in orbit. By the end of 2022, SpaceX had deployed another 1,722 Starlink satellites (using 34 launches).

That Starlink ramp-up is fast. SpaceX is an aggressive company that tends to move quickly, with the thousands of Starlink satellites in orbit a testament to the company’s assertiveness. None of SpaceX’s launch competitors have yet to demonstrate that they are equally, or more, aggressive. They seem content to go slowly, even when SpaceX eats their lunches. Their lackadaisical rocket development pace might bleed over into their rocket launch cadence.

That’s a problem for Kuiper.

Rocket Math!

As it is, Kuiper has yet to test its satellites in orbit. Almost all of the rockets it’s hoping to use have yet to be launched, which means their reliability is unknown. It’s unclear how these companies will conduct half of the 77 contracted launches to meet the FCC deadline. The following analysis is based on many assumptions, which themselves are based on company history and performance. You’ve been warned.

Take Arianespace and its Ariane 6. The company is replacing its Ariane 5 with the Ariane 6. Arianespace launched the Ariane 5, on average, about six times per year. It may be that the company can only launch the Ariane 6 on the same launch cadence–about six times per year. The Ariane 6 can carry 35-40 Kuiper satellites (we’ll split the difference and say 37). Using my 500 kg per satellite estimate, that’s about 18,500 kilograms per Ariane 6 launch–well within the rocket’s projected capabilities for LEO.

While I hate to do it, I will refer to the Ariane 6’s Wikipedia entry for its launch manifest. According to that manifest, there were supposed to be three Ariane 6 launches in 2023. Since the Ariane 6’s first launch won’t occur until early 2024, it’s likely those three launches get moved to the right of that time.

According to the entry, nine launches were already scheduled for 2024–optimistic for a company that launches an average of six large rockets annually, especially considering the Ariane 6 is new. If Arianespace can’t accommodate those three extra launches from 2023 into 2024, it will need to move three into 2025. So far, none of those twelve launches have anything to do with launching Kuiper satellites.

For 2025, there appear to be only three Ariane 6 launches, which may mean that Arianespace might be adding another six (for nine launches, based on 2024’s example) Ariane 6 launches for Kuipersats. Except, there might be three more launches scheduled initially for 2024 moved into 2025, bringing the possible Kuiper launches down to three. That would be 111 Kuipersats deployed by Ariane 6 rockets during 2025, about 7% of the total required to meet the FCC requirement.

Adding in Vulcan from ULA. Vulcan is supposed to lift 45 Kuipersats to orbit (an estimated total mass of 22,500). According to Vulcan’s launch manifest on Wikipedia, it will launch at least six times in 2024 and at least another six times in 2025. None of those launches are for Kuipersat. The CEO believes the company will launch the Vulcan every two weeks by the end of 2025. We’ll take his word for it, even if there is no sound basis for accepting that. Assuming the last six months are part of that ramp-up, at least another nine launches could be conducted–maybe for Kuipersat launches.

Nine launches, multiplied by 45 satellites each–that’s 405 Kuipersats put in orbit for 2025, 25% of what’s necessary to meet the FCC deadline in mid-2026. Adding those to the 111 from Arianespace means that about 32% of the required Kuipersats will have been deployed. That also means an estimated 1,103 Kuipersats must be deployed before July 2026.

Only one company has managed to deploy close to that many internet satellites in half a year–and only this year (2023) at that: SpaceX. Can ULA and Arianespace launch a comparable number of rockets (more, actually) to get the necessary number of Kuipersats in orbit? Even if they split the 1,103 number down the middle, it seems not likely. ULA would have to launch over 12 Vulcans, meaning the first half of 2026 is dedicated to Kuiper launches. Arianespace would have to launch 15 Ariane 6s in half a year, exceeding its annual overall launch average for all of its launches so far.

I know I’ve left out Blue Origin. The company is hard to take seriously, as its engine troubles and timelines pose a case of delays. Its position is worse than that of ULA and Arianespace in that it doesn’t have orbital launch experience. If it had already exceeded expectations, that lack of experience might not have posed as much of a challenge. But the company is not exceeding expectations. That makes me believe it also will not exceed expectations when it (maybe) begins launching.

In theory, Blue Origin’s New Glenn would be tremendously valuable to the Kuiper team. New Glenn is supposed to be able to carry 61 Kuipersats to orbit. But as noted in the beginning, there’s that shroud of uncertainty. We don’t know the projected launch manifest, aside from the fact that the rocket is supposed to launch Kuipersats. Since it’s a new team operating a new rocket, the whole system’s reliability is questionable.

Rockets Can Be Reliable, But New Rockets Can Be Interesting

Much of this analysis assumes everything will go smoothly with these new systems. But that’s unrealistic. The chances are that one of these new rockets will produce a surprisingly robust explosion instead of launching (even SpaceX has experienced those). Or the problem may be more subtle, such as not reaching the correct altitude for the satellites. Those challenges will slow a launch cadence down to zero sometimes.

Again, the analysis above is based on guesses. But those guesses are based on the numbers provided, launches conducted so far, anticipated launches, company histories, etc.

Ultimately, the circumstances of the companies Kuiper has contracted with make it unlikely to meet the FCC deadline for satellite deployments. They haven’t launched their rockets yet. When they do, there will be a period in which the companies will become familiar with these new systems. That means not racing to launch quickly since the companies would risk customer payload if they did. Eventually, they will understand the system they are working with, which is when Murphy steps in.

As noted earlier, it took quite a few years for an already aggressive company, SpaceX, to launch its Falcon 9 as often as it does now. One of the reasons it reached those high launch numbers is because its people were already familiar with the rocket. And it launches from three launch pads.

Neither ULA nor ArianeGroup has demonstrated a hunger close to that of SpaceX in their rocket development. Neither ULA nor Arianespace has the experience with their new rockets that SpaceX gained with the Falcon 9. While ULA and Arianespace could undoubtedly reach high launch cadences, they may not reach them as quickly as they advertise (SpaceX and Rocket Lab didn’t). A new company like Blue Origin will probably take longer to achieve a high annual launch cadence than ULA or Arianespace.

Then there’s the launch range deconfliction, which will be interesting to watch.

Is it possible for Kuiper to get 1,618 satellites into orbit by July 2026? Perhaps, but it seems unlikely.