It’s a space industry version of the classic “That’s not a knife” scene from the very old movie, “Crocodile Dundee.” Amazon looked at what Bharti, Hughes, and the United Kingdom were doing with OneWeb, and basically chuckled, “That’s not an investment.” Then the company whipped out $10+ billion for Project Kuiper (PK) and added “Now, there’s an investment.”
Now, There’s an Investment
Amazon revealed the $10+ billion figure in its company blog post last week. The post was a self-congratulating message, celebrating the U.S. Federal Communications Commission’s (FCC) order/approval for Amazon to deploy 3,236 PK satellites. The justification provided for most of the FCC’s approvals in the Amazon request is that it will “serve the public interest.” This may mean Amazon’s low Earth orbit broadband service will provide fast network connectivity to underserved areas in the U.S. Or it may mean that the tax revenues from Amazon’s efforts are projected to be fairly fat.
There’s a lot of information to unpack, and there are no surprises. What Amazon’s blog and the FCC announcement did was verify Amazon’s Project Kuiper plans and commitment, with the U.S. government’s official blessing. Some of the easier bits to note about Amazon and PK follow:
Amazon is serious and confident about PK
The “over $10 billion” investment quote seems to confirm this
$10 billion seems realistic for manufacturing, launching, deploying and operating PK satellites
The $10+ billion should include not just satellites and launches, but the infrastructure
It appears Amazon will keep expenses down
If that $10+ billion includes everything necessary for satellite operations across 3,000+ satellites, then it’s not a bad deal
It doesn’t appear to be as inexpensive as SpaceX’s constellation
Can’t rely on legacy satellite manufacturers to meet timeline--too slow
Amazon has nine years to deploy all PK satellites
Manufacturers need to build over 360 PK satellites annually
Only SpaceX & OneWeb have the ability to meet/exceed that manufacturing rate
Big and Round
$10 billion seems to be “about right” for building out a LEO broadband constellation. It falls in line with estimates for Teledesic (an older constellation idea) and Starlink. However, Teledesic was planning to put up somewhere around 1,000 satellites, while SpaceX has released plans for over 42,000 Starlink satellites. PK would be in the middle of the bunch, but estimating it will invest around the same amount. This indicates that something in Amazon’s satellite production/deployment chain is more expensive than SpaceX’s Starlink production/deployment chain.
Is it that Blue Origin won’t be as inexpensive to use as SpaceX? It certainly isn't as operational as SpaceX. Or, perhaps like OneWeb, PK would be launched with Arianespace, which is also more expensive (than SpaceX)? Maybe the satellites are using more expensive components, including laser crosslinks? There are still too many unknowns to pin down why PK is more expensive to manufacture and deploy than Starlink.
To put the $10 billion number in perspective, that amount was nearly what Amazon invested in capital expenditures and leases in the latest reported quarter (Q2) this year. From Brian Olsavsky, during Amazon’s latest quarterly call:
“We continue to invest meaningfully, including $9.4 billion in capex and finance leases in Q2 alone, an increase of 65% year-over-year, primarily driven by investments in our fulfillment and logistics footprint.”
Considering that building out the PK constellation is part of capital expenditures and that the $10 billion the company is using to build it out will likely be spread out over 9-10 years, $10 billion--while it is a lot of money--doesn’t amount to much each quarter. Especially when comparing the total CAPEX investments the company makes quarterly. Guessing how Amazon may spend the $10 billion, it may mean that, at a $1 billion a year addition (spread out over 10 years), Amazon’s quarterly CAPEX investments may **only** increase by $250 million. In reality, the division of investments dropped in each quarter due to PK activities will not be that neat. The blog post certainly didn’t break down what Amazon will invest in for PK.
There are capital expenditures associated with PK that Amazon has already figured out for its day-to-day business. Those include building out data/communications infrastructures and data/networking hardware. Also, Amazon is, unlike OneWeb, a profitable company. It understands customer needs. It understands data. More importantly, it knows how to look for needs customers may not know they need (Prime being as an example) using data.
Who Will Build It?
What Amazon has never done (at least it’s never admitted this publicly) is to design and manufacture satellites. There are companies who were also in this predicament who appear to overcome this challenge, such as SpaceX. It’s not clear if Amazon will go SpaceX’s route and design/manufacture its PK satellites in-house. Or if it will contract with an established manufacturer to have that done.
What is clear is if Amazon goes with a legacy manufacturer, such as Boeing or Lockheed Martin, then it will not be able to meet the FCC’s deadline to deploy all 3,236 satellites by end of July in 2029. That deadline means that from now through July 2029, Amazon would need to manufacture over 360 satellites annually. Probably more each year, since no Amazon PK prototype satellite has been deployed (no public announcement has been made about contracting with a satellite manufacturer to do so).
As we’ve seen with SpaceX (and somewhat with OneWeb), it is possible to build a factory that can pump hundreds of satellites each year. But would Amazon approach those two companies to do so? It seems unlikely, but there don’t appear to be other manufacturers out there able to produce satellites in large quantities, annually. Which may mean that Amazon builds its own satellite manufacturing facility. That will take some time, at least a year, but if the factory can approach a SpaceX-like manufacturing rate (~7 per day), it would have no problem building enough satellites to meet the FCC deadline.
If Amazon does not build its own manufacturing plant and does not contract with OneWeb or SpaceX, then the choices facing it among legacy satellite manufacturers don’t just limit how many satellites would be produced annually. It would impact Amazon’s budget, as it would then probably spend more than $10 billion for the satellites alone since those legacy companies tend to offer very expensive satellite products.
In the meantime, both OneWeb and SpaceX are racing ahead to deploy their constellations. When Amazon finally deploys an operational constellation, it will be entering a field with established players. As it did when it offered its web services.
It didn’t take long for AWS to become one of the 800 lb gorillas in the tech sector.