Amazon’s Small Reality-Check: Kuiper Launches With SpaceX
Amazon’s contract with SpaceX for three Falcon 9 launches is only the beginning…
I wrote an article for Astralytical related to this topic: High Ambitions/Low Bar: Initial Rocket Launch Goals. It’s about the industry’s tendency to sign up customers for ghost rockets despite a history of delays from initial launch estimate to actual launch. While there might be good reasons for some companies to do so, they seem shortsighted to the potential consequences.
Regarding Amazon Kuiper and launching its satellites, it was a matter of WHEN Amazon would contract with SpaceX instead of IF.
I based that assumption on believing Amazon was serious and committed to building and deploying a space-based internet relay system. If it was serious, its business plan included having a certain number of KuiperSats deployed by a particular date (or date range). Adding to the pressure applied to Kuiper was that external factors (such as Federal Communications Commission (FCC) compliance) drove its satellite deployment schedule: 1,618 Kuiper satellites by July 2026–although the FCC is likely to be lenient.
Because it is serious about its business, Amazon’s announcement last week of its decision to contract with SpaceX for three launches to deploy its KuiperSats was, if not inevitable, certainly logical. Until that announcement, the company appeared to have tried to launch with any company that wasn’t SpaceX.
Marching From Realistic to Hopeful
One of the first companies Amazon contracted launches with was ULA. In April 2021, the company announced a 9-launch contract using ULA’s Atlas V. That announcement made sense, as ULA had the Atlas V (an extremely reliable rocket) available. It made so much sense that I never thought the announcement was interesting enough to cover.
However, Amazon’s launch selection announcements following that first ULA announcement were so much less sensible that they were worth writing about. Its decisions to use…alternative…launch systems seemed based on naively optimistic assumptions. Or the company’s desperation drove the decision to use any rocket but the Falcon 9.
Maybe both.
One of the first alternative launchers Amazon announced a launch contract with was ABL Space Systems. In its November 2021 news release, Amazon highlighted the following reasons for choosing ABL:
With a one-ton-plus payload capacity, RS1 delivers the right capacity and cost-efficiency to support our mission profile. ABL’s RS1 and GS0 launch system are also fully containerized and mobile, providing the speed and flexibility to support these early launches. We have been working closely with the ABL team for several months and already completed two integration design reviews—including plans for a novel adapter design. We will conduct an initial fit check early next year.
Amazon was super-serious in its belief that ABL would have the capability to launch its prototypes in 2022. It assigned people for those design reviews and made time for designing an adapter to deploy from ABL’s rockets.
The problem with Amazon’s announcement? When Amazon announced its contract with ABL, the launch startup had yet to launch any rocket. ABL hadn’t demonstrated it could launch a rocket, much less launch it reliably. My observation of the risks in Amazon’s decision at the time:
But the choice of ABL’s rocket is challenging for PK because launching prototypes on an unproven rocket is not ideal, as RS1 (if it gets operational) is an unknown risk factor. This is in addition to the very nature of prototype satellites, which will be unproven and contain a level of risk. And to be clear, the engineers for each system--rocket and satellite--will have tried their best to reduce the risk for both. It’s likely the PK’s engineers would have rather seen a proven, operational launch system selected for their prototypes, but their options are limited.
ABL was an odd choice because there were other options as well. In that article, I noted that no other smallsat-dedicated launchers could launch the higher-mass Kuipersat prototypes. That meant that while Amazon had options, there weren’t many, and one of them was off-limits:
If PK is determined to keep things local, then that leaves three launch service providers. ULA and Northrop Grumman are expensive and infrequently launch, while SpaceX provides inexpensive launch services at less than half of the per kg cost for the RS1. But the fact that PK didn’t choose SpaceX, instead choosing to pay more than double SpaceX’s per kg rideshare cost ($5,000/kg), indeed comes as a surprise to no one. Although, Amazon shareholders should question that decision at the next meeting.
The Bleakening
About one year later, in October 2022, Amazon announced that its two prototypes would launch on the United Launch Alliance’s (ULA) Vulcan instead. It expected ULA to launch its two prototypes in “early 2023.” In the meantime, ABL conducted a launch (without Kuipersats on board) close to the date it was supposed to have launched Amazon’s prototypes, in January 2023. The launch failed, and ABL hasn’t conducted any launches since that failure.
Still, Amazon’s move to ULA for its prototypes was a jump from one ghost rocket to another. Because Vulcan still wasn’t ready, Amazon launched its two KuiperSat prototypes using one of the nine Atlas V’s it contracted with ULA.
Vulcan still hasn’t launched yet and won’t until December 24, 2023. About nine months earlier, in April 2022, the company had announced contracts for 83 launches. Amazon would not just rely on ULA’s rockets but those of Arianespace and Blue Origin. However, as with ABL, none of those companies had operational rockets–at least the rockets they were contracted to use to deploy Kuiper.
That April announcement sent mixed signals–none positive. A few days after the announcement, I elaborated on a few concerns:
ULA and Arianespace have other customers, but they historically have maintained low annual launch rates (when compared with SpaceX’s four+ launches per month).
Amazon’s contracts with those companies takes up whatever launch slack they may have had. They need Amazon more than Amazon needs them. If necessary, because those launch companies may not produce rockets on time—Amazon could just turn to SpaceX. Worse for those companies, it basically forces other potential customers interested in space operations to go either with SpaceX or consider using smallsats and the corresponding smallsat launchers. As explained in the next section, there are a few characteristics in SpaceX’s current launch operations that allows it to rapidly respond to more customer demand. But neither option builds a diversified loyal customer-base for those three companies.
Even if they all managed to get new rockets operational, all three launch companies would be focusing on fulfilling their launch contracts for Kuiper. Their history shows they’d also do it slowly (despite optimistic claims). It’s entirely possible those contracts were already discounted because of the risks involved with using new rocket systems. And while those companies strove to clear out those less-profitable contracts, SpaceX would be getting the customers they couldn’t serve.
It already is, based on a Telesat announcement in September 2023. Or an ESA announcement in October 2023. And in December 2023, Amazon announced three launches with SpaceX’s Falcon 9.
Availability Trumps Hope
While some reasons have been floated for why Amazon might have this change of heart, there is no mystery: SpaceX is the only company with rockets that can launch Amazon’s satellites. There’s nothing special in this decision. The fact Falcon 9 is reusable is irrelevant. SpaceX’s lower launch pricing was never attractive since Amazon chose higher-priced launch competitors earlier. Reliability wasn’t a consideration because the company decided to go with untested rockets. It’s just that the Falcon 9 is available.
Since Amazon appears serious about the space-based internet business, it will use what’s available, contracts for ghost rockets notwithstanding. Reality seems to be dragging the company back from the brink of hope-based business decisions.
The company might also feel shareholder pressure, as Amazon is already starting from a significant numerical disadvantage. SpaceX has deployed over 5,500 Starlink satellites since 2019. Nearly 1,900 Starlinks were deployed in 2023 (so far). Amazon shareholders might be pushing for earlier and more Kuiper deployments to keep the company at the table.
Those shareholders don’t appear to care who owns the rocket so long as–and this is an important distinction–the rocket is available. It wouldn’t be surprising if Amazon secures more SpaceX launches based on the deadlines, the imbalance favoring the competition, and the uncertainty that ghost rockets will coalesce into real rockets. Blue Origin, for example, is a considerable unknown, unwilling to pin down a launch date despite already extending nearly five years from its initial launch estimate.
Some of the company’s issues have already impacted the other launch company, ULA. It’s not clear if it will continue to experience challenges that will impact ULA. That’s not great for ULA, even if it successfully launches Vulcan on December 24 this year.
It’s unclear if Blue Origin will be able to supply a steady stream of engines to ULA to launch. And, ULA’s CEO wouldn’t (probably couldn’t) give a date for the next Vulcan launch except that it might occur in the first half of 2024. However, the fact that either company is unwilling or unable to pin down a more realistic date doesn’t bode well for the U.S. launch industry.
ULA has the largest contract of ghost rockets to launch for Amazon’s Kuiper. Suppose ULA’s rockets keep ghosting Amazon because Blue Origin can’t supply reliable engines to the company (or supplies them slowly). In that case, Amazon has little choice but to turn to SpaceX for more launches. In the meantime, ArianeGroup is constantly shooting more holes in its feet than there are in Swiss cheese. Maybe its Ariane 6 will be ready by mid-2024.
More Amazon news releases will likely come out with more SpaceX launches, probably when it’s obvious the new rockets won’t be ready by next year. If their rockets do launch, then I’ll be pleasantly surprised.
For those interested, here are some of my analyses from the past few years about Kuiper and the global launch situation, from oldest to latest.
Project Kuiper’s Imaginary Rocket, Astra Space’s Broadband Distraction
Ample Capacity and Monopoly Markets: the U.S. Space Launch Paradox
And for Astralytical: High Ambitions/Low Bar: Initial Rocket Launch Goals
If you liked this analysis (or any others from Ill-Defined Space), any donations are appreciated. For the subscribers who have donated—THANK YOU!!
I missed the announcement that ESA signed a contract to launch Galileo satellites on SpaceX Falcon 9. I am wondering when ESA will contract with SpaceX to launch the Sentinel-1C satellite that is currently waiting for ArianeSpace to fix the Vega-C.