SpaceX Starlink is a technological marvel, and on its way to becoming a commercial success.

From 2015, when it was but a twinkle in Elon Musk's eye, with no customers and no satellites in orbit to support having customers, Musk's Starlink high-speed satellite internet system has grown into an orbital behemoth: More than 5,000 satellites circling in Low Earth Orbit (LEO), and 2.2 million internet users down here on Planet Earth.

But here's something you may not know about Starlink and its 5,000 satellites: Every year, 1,000 of them die and fall out of the sky to burn up in Earth's atmosphere. And this number is only going to grow with time.

Stylized image of a global internet network over Earth.

Image source: Getty Images.

Planned obsolescence

This isn't necessarily a bad thing.

SpaceX famously develops its products through iterative design, constantly upgrading them to lower cost, increase performance, or both. For example, a SpaceX Falcon 9 rocket ship in 2016 was advertised as launching less than 4.9 tons of payload to geostationary transfer orbit (GTO) for $61 million. Today, upgraded "Block 5" Falcon 9s cost 10% more -- but carry 70% more cargo to GTO.

Likewise with Starlink satellites. As Payload Research explains in its 2024 "The Starship Report," early versions of Starlink communications satellites massed just two or three hundred kilograms. Each such satellite could support about 23 gigabits per second-worth of download speed.

Newer "V.2 Mini" Starlinks are larger (as much as 800 kg), and boast four times the bandwidth. Musk has promised that as Starlink evolves toward a future full-sized "V.2" model, the mass will increase 5x from the earlier V.1 Starlink -- but bandwidth will grow "almost an order of magnitude," a 10x increase.

The downside to fast turnover

Now, one reason SpaceX products increase in capability so quickly is that SpaceX doesn't keep some them around very long. Starlinks in particular have only a five-year lifespan, and are expected to fall out of LEO pretty regularly, to be replaced by new and improved models.

But there's a downside to this.

Simple math tells you that if a product has a 5-year lifespan, then 20% of that product (100 / 5) ages out of service every year. Now, this wasn't too noticeable back when Starlink started beta service with just 800 satellites. It's become more noticeable -- though still manageable -- today, with 5,000 Starlinks circling the globe.

If 1,000 of these birds age out and need to be replaced every year, and Starlink can replace them at a rate of 23 satellites per Falcon 9 launch, then 44 launches per year should do the trick. (Hint: SpaceX launched more than twice that number of flights last year.)

The problem is that SpaceX is progressing toward 12,000 Starlinks in orbit, requiring 2,400 satellites, meaning 104 launches per year to replace old satellites. At an estimated cost of $500,000 to build each satellite, plus $67 million per Falcon 9 to launch them, this implies an annual cost of about $8.2 billion per year -- just to keep the satellite constellation at full strength.

The cost of progress

Now SpaceX, as you've probably heard, is in the process of planning an IPO for Starlink, just as soon as its "revenue growth is smooth & predictable."

As investors begin weighing whether they're going to want to participate in that IPO when it happens, they may want to ask themselves: Can SpaceX afford to pay that kind of money, and still earn a profit from Starlink?

Surprisingly, it can -- but the profit might not be as big as SpaceX hopes.

Consider: According to the latest data from Payload Research, Starlink generated $4.2 billion in revenue for SpaceX last year, and might double its 2023 numbers and collect $10 billion in revenue in 2024. By the time SpaceX has all 12,000 Starlink satellites in orbit (I estimate this might happen in late 2026 or early 2027), SpaceX internal documents suggest that Starlink revenue could exceed $32 billion.

That much revenue should easily cover the cost of building and launching as many Starlinks as SpaceX needs to keep the constellation fully stocked.

This, as the saying goes, is the good news. Now here's the bad news for SpaceX and its planned Starlink IPO:

Although Starlink is on course to generate more than enough revenue to cover its equipment replacement costs. I can't help but notice that the $8 billion (ish) that Starlink will need to spend on maintenance capital expenditures (i.e. satellite replacement) still represents roughly 25% of the company's anticipated $32 billion revenue. Compare that to a terrestrial cable internet company like Comcast, though, and, well ... it turns out that Comcast's annual capex averaged less than 10% of its annual revenue over the last five years (according to data from S&P Global Market Intelligence).

Now, it makes total sense that infrastructure on Earth would be cheaper to build and maintain than infrastructure in space. That's just logical. However, this still suggests that SpaceX's Starlink business may end up being less profitable for Elon Musk, than a more down-to-earth internet company like Comcast.

This may not be a reason to avoid a coming Starlink IPO entirely. But it does inject a note of caution -- and it's a good reason for investors to think twice before investing in any such IPO.