In 2019, Amazon.com (AMZN 3.89%) announced its next big business idea. Following in the footsteps of SpaceX founder Elon Musk and his Starlink satellite internet system, Amazon would build a competing satellite internet business: Project Kuiper.
The plan calls for the deployment of 3,236 satellites, which will require 92 separate rocket launches to put them in orbit. Yet this massive undertaking will actually be smaller in scope than Starlink, which will eventually boast 12,000 satellites, and will need to launch more than 100 times a year just to refresh its constellation as old satellites fall out of orbit. Even so, Amazon initially estimated the cost of Project Kuiper would be roughly similar to the $10 billion that Musk said it would cost SpaceX to create Starlink.
There's just one problem with that: It turns out Amazon may have been wrong. Deploying the Kuiper constellation might cost it as much as twice that amount.
NASDAQ: AMZN
Key Data Points
Quilty sounds the alarm on Project Kuiper costs
In a report titled "Amazon Project Kuiper ... A Quilty Space Deep Dive" last month, space sector business intelligence company Quilty Space warned that "costs are ballooning" on Project Kuiper, and forecast that it would take between $16.5 billion and $20 billion to deploy the first generation of the constellation.
Granted, with $89 billion in cash and equivalents on its books and more than $48 billion in annual free cash flow (according to data from S&P Global Market Intelligence), Amazon can easily manage those greater expenses if it needs to. But the higher the initial costs run, the harder it's going to be for Amazon to turn a profit on Project Kuiper. And the more money Amazon ends up losing on it, the more this venture will be a drag on company profits -- and the more investors may regret it.
Why is Project Kuiper looking likely to cost so much more than Amazon initially expected? It's hard to be certain, as the company is keeping mum. But a couple of possibilities suggest themselves.
First and most obviously, there's the fact that Amazon is an e-commerce company -- not a rocket company. Unlike SpaceX, which builds its own satellites and builds its own rockets to launch them (for a price that rounds down to free), Amazon must buy its rocket rides on the open market. To date, it has contracted with everyone from Boeing to Lockheed Martin to Airbus to Amazon founder Jeff Bezos's own pet rocket company, Blue Origin. Operating under a time crunch to get its satellites in orbit by a regulator-imposed deadline, Amazon even bought a few launches on SpaceX's Falcon 9.
A second factor, perhaps less well-known to the public, may be internal to Project Kuiper itself.
As space journalist Eric Berger notes in his recent book, Reentry: SpaceX, Elon Musk, and the Reusable Rockets that Launched a Second Space Age, Amazon has hired former SpaceX Starlink head Rajeev Badyal to run Project Kuiper.
But as Berger also reports, back when Badyal worked for SpaceX, he "clashed" with Musk. Badyal moved "too cautiously," wrote Berger, and "wanted to continue tinkering with the Starlink satellite design," while Musk is infamous for wanting everything done yesterday. Eventually, in June 2018, Musk fired Badyal.
If Amazon has SpaceX's second-string Starlink team running Project Kuiper, it makes sense that it might take longer to deploy and wind up costing more than Starlink. Furthermore, as Quilty points out in its recent report, it doesn't bode well that "Amazon placed Kuiper within its troubled Devices & Services division instead of its rockstar Web Services unit."
Given Amazon's "shaky" record scoring hits in consumer devices, and its demonstrated willingness to shut down money-losing projects, that almost seems like intentional foreshadowing.
What this means for Amazon
Could Project Kuiper end up costing twice as much as it was originally supposed to?
I wouldn't be surprised if it does. Most of the companies Amazon has contracted with to launch its satellites charge far more than SpaceX does for rocket launches -- $100 million and up. At those prices, just the cost of launching 92 times could easily burn through Amazon's $10 billion budget. And that's before accounting for the cost of building the 3,236 satellites.
At Quilty's estimated cost of $1.5 million to $2 million apiece, the satellites alone could cost Amazon $6.5 billion.
That's a heck of a big start-up cost Amazon will need to earn back as it begins launching its satellites to orbit and signing up customers. Yes, begins. So far, Amazon has only launched a pair of test satellites. Operational satellite launches aren't expected to start until later this year -- and the constellation must be completed by July 2026 or Amazon will violate the terms of its Federal Communications Commission license.
In a best-case scenario, Quilty's report posits that if Amazon can eventually sign up 100 million paying customers for, say, $30 a month, that would give Project Kuiper annual revenues of $36 billion -- about as much as Amazon reaps in annual subscription revenues today. But it's a long road from zero subscribers and two satellites to 100 million subscribers and 3,236 satellites.
Amazon is going to be losing money on this project for a long time.
