In October 2023, Amazon.com (AMZN 0.55%) took one small step toward chasing SpaceX's giant leap... of building a broadband satellite internet constellation circling the Earth.

On October 6 that year, a United Launch Alliance (that's Boeing and Lockheed Martin) Atlas V rocket blasted off carrying Amazon's experimental Kuipersat-1 and Kuipersat-2 to a 500-kilometer orbit above Earth, aiming to test the satellites' performance preparatory to beginning full-scale satellite production, and building a full constellation of 3,236 internet satellites. At the time, Amazon anticipated beginning full-scale deployment of Project Kuiper in H1 2024 and beginning "early customer pilots in the second half of 2024."

But Amazon is behind schedule.

Plans and practice

As I wrote back in December 2023, it took SpaceX 15 months to move from initial testing of its own experimental Starlink satellites to kicking off full-scale production and launching its first full payload of operational Starlink satellites. It took SpaceX another 18 months to announce the start of beta service on Starlink in October 2020 -- 32 months from experimental launch to beta service.

Suffice it to say that when Amazon promised investors it would go from experiment to beta service in just 14 months, I was skeptical.

I was also right. Last week, Amazon finally launched its first batch of operational Kuipersats aboard another ULA Atlas V. The April 29, 2025 launch was roughly a year behind schedule and took several months longer to accomplish than SpaceX's first operational Starlink launch.

And beginning Kuiper beta service will take even longer.

Image source: Getty Images.

Kuipersat versus Starlink

Consider: SpaceX initially believed it needed 800 Starlinks in orbit before it could cover enough territory to begin beta service. Ultimately, the company waited until it had nearly 1,000 satellites in orbit before announcing a price for Starlink beta service.

Today, SpaceX has more than 7,100 Starlinks in orbit, generating billions in annual revenue, and is already on its third or fourth generation of development. Most Starlinks going up are V2 Mini models, significantly upgraded from the original Starlink.

In contrast, after last week's launch, Amazon now has just 27 operational Kuipersats. Weighing perhaps 1,260 pounds each versus 1,760 pounds for SpaceX's more robust V2 Minis, Amazon's satellites are presumably less capable than SpaceX's. But are they good enough to compete with Starlink?

Amazon's looming FCC deadline

Amazon is now in a race against time to get 50% of its constellation (1,618 satellites) completed by a Federal Communications Commission (FCC) deadline that tolls on July 30, 2026. At the rate of 27 satellites per launch, this implies the company needs to launch 59 more times over the next 14 months, or roughly once per week.

In the meantime, Amazon must spend heavily on its $10 billion Project Kuiper. It must spend money to build satellites. It must spend money to launch satellites on ULA rockets, on Arianespace rockets, on Blue Origin rockets, and even on SpaceX rockets. Basically, Amazon has to buy up (and indeed, already has bought up) all the rocket rides all around the world for the next year or three to get its constellation completed.

For that matter, just to reach a point where Amazon can announce beta service, or as it terms it, "early customer pilots," will be tricky. Amazon's Kuipersats orbit higher than SpaceX's Starlinks. Each Kuipersat can cover more square footage down on Earth than can an equivalent Starlink (albeit at the cost of a slower "ping" signal).

This means Amazon won't need as many Kuipersats to offer the same coverage as Starlink or to begin beta service. Still, even if we assume Amazon can begin beta with, say, 700 satellites instead of SpaceX's 800 or 1,000, Amazon probably needs to launch 25 more batches of satellites before it can announce beta service.

If Amazon tries to get at least beta service up and running by July 30, 2026, hoping this will be enough to convince the FCC to extend its license deadline, it would still need to launch nearly every other week between now and then -- and even then, winning FCC approval of a deadline extension would not be certain.

(Note: Amazon says it actually plans to begin beta service later this year. But once again -- I'm skeptical.) 

A second issue dogging Amazon is production capacity. Amazon eventually wants to be building five Kuipersats per day, but The Wall Street Journal reports it's currently building only about one a day. This implies Amazon might not have more than 450 satellites built by its FCC deadline -- far fewer than the 1,618 it needs.

The upshot for Amazon investors

According to data from S&P Global Market Intelligence, Amazon.com boasts $650 billion in annual revenue and nearly $95 billion cash in its war chest. If anyone can compete with SpaceX in satellite internet, it's Amazon.com. Despite all the challenges it faces, I wouldn't count Amazon out of this race yet.

But winning it is not going to be easy. Like everything else in space, I think it's going to be very hard, indeed.