Elon Musk Is Lying About the True Cost of Starlink

Elon Musk tells the public and his investors a very simple story: Starlink-Starship duo is going to print billions of dollars and fund the colonization of Mars. His core argument is that his new Starship rocket makes launching payloads so incredibly cheap that Starlink simply cannot fail.

He is lying.

Media dutifully copies Elon’s rhetoric:

“How Starlink Will Fund Elon Musk’s City on Mars”

“Starship’s Game-Changing Launch Costs Will Make Satellite Internet Unstoppable”

“Why Starlink V3 Could Make Ground-Based Fiber Obsolete”

“SpaceX Is About to Turn Orbit Into a $30 Billion-a-Year Cash Cow”

“The Mega-Rocket That Will Bring Gigabit Speeds to Every Corner of the Planet”

Musk is obfuscating the true costs of his avant-garde projects. He wants you to obsess over “cheap launch”, ignoring true costs. Myth of “cheap launch” shatters when you see the true costs.

The “Cheap Launch” Distraction

Starship is the load-bearing pillar of modern SpaceX. Its the next-generation mythical rocket that is supposed to crush launch costs and make scaling out Starlink cheaper.

Elon’s favorite Twitter pastime is hyping launch costs of Starship. This is a seductive distraction. It conveniently ignores the mountain of non-marginal costs:

• R&D of Starship itself. Starship is 15 years into development. And still unfinished
• Launch infrastructure
• Manufacturing facilities

And so on. These are not a direct cost in launching, but these have to be paid. This is simply a cheap parlor trick to generate hype and impress masses.

And that’s just Starship. Starlink satellites are also accompanied by similar costs. And cost of manufacturing them. In-house does not mean free.

“Launch 100 satellites for $100M”

Amazing. The satellites add another $100M. Elon just lied understated true costs.

Sky hates Starlink’s signals

Scaling traffic per satellite is hard. Ground links are a brick wall. SpaceX picked increasing frequencies to crush this wall. Next-generation Starlink V3 satellites use 70-110 GHz radio. Claimed speeds skyrocket to “1Tbps”, more than 10x faster than previous versions.

There is just one trap: sky absolutely hates these frequencies.

Your home Wi-Fi collapses within 5 metres. 5G cell signals get blocked by wet leaves.

At these frequencies, even on a cloudless, bone-dry desert sky becomes an impenetrable brick wall.

Additionally, “weather” is a lot more than human-visible clouds/rain. A clean perfect line-of-sight from space is impossible.

SpaceX’s highly publicized “satellite Lasers” do not fix this. Bouncing the signal across continents just to find a clear patch of sky is counter-productive. Modern ground backhaul is exclusively fiber-optic—it is naturally direct, boasts near-infinite bandwidth, has massive wide-area coverage, is highly redundant, and routes data at the speed of light without atmospheric interference. Their only value is acting as an expensive band-aid for the fact that satellites physically cannot touch fiber everywhere. It is a massive, inefficient detour pretending to be a futuristic feature.

Military Have Already Rejected This Hardware

The Pentagon has bottomless black budgets and a mandate for absolute technological survival. They rely on space communications for nuclear targeting, drone operations, and naval positioning. They rely on bleeding-edge radars for targeting and detection. Yet, the military actively avoids 100GHz frequencies. Globally, not a single military uses such frequencies in regular operations (or at least to the best of our knowledge)

Why? Because it is completely unreliable. The military sticks to lower, reliable frequencies because they cannot coerce weather.

If the defense industry—with infinite money and the smartest aerospace contractors on Earth—shuns this technology because it is too practically fragile, the idea that a commercial internet provider is going to conquer it is laughable. Musk is pretending he can casually solve roadblocks that DARPA refuses to touch.

Myths vs Stark Reality

To try and punch through that atmospheric brick wall anyway, SpaceX needs alien technology.

The absolute best standard commercial chips top out around 30 GHz. 100GHz hardware does not exist on the open market.

The kinds of hardware needed for 100GHz space links are pretty much sci-fi. The capabilities needed exceed commercial tech by several orders of magnitude

“SpaceX is backed by classified U.S. defense block grants and massive federal subsidies”

Unfortunately, bureaucracy and money does not change physics. It does not make sci-fi tech practical.

“In-house”

In software, doing things “in-house” saves money [1]. In bleeding-edge aerospace hardware, doing things “in-house” is a death sentence. The research and development costs explode into the billions before a single satellite ever reaches the launchpad.

To even try to make this work, SpaceX has to design bespoke, highly experimental microchips from scratch. They have to partner with hyper-niche, ultra-expensive military foundries. Occasionally, they have to build custom plants. Costs will just skyrocket

To say nothing of talent. There are exactly 10 people on earth who can dream of building 100GHz analog space links. Standard research and PhDs does not universally target 100GHz RF

[1] (well that’s also a myth, but pretend it does)

Laughable claims of achieving the impossible

What would it mean if SpaceX actually solved this? The sheer engineering capability required is astronomical. Look at standard consumer RF: 5 GHz Wi-Fi struggles to penetrate two walls in a house. 5G mmWave (20–30 GHz) barely lasts a single neighborhood block and is completely stopped by a glass window or a wet leaf.

Now jump to 100 GHz. As established, the sky is an impenetrable black hole. The kinds of hardware needed for 100GHz space links are pretty much sci-fi. The capabilities needed exceed commercial tech by several orders of magnitude

The mixed-signal titans (Analog Devices, Linear Tech, Hittite, Maxim, Ti) spent five decades mastering this dark art, and their absolute best commercial tech barely reaches 30 GHz. To believe that SpaceX—an aerospace integration company—magically leapfrogged them to build a perfect 100 GHz analog chain demands a literal tech God. It is completely non-credible and laughable.

The looming starlink saturation

All of this brings us to the final, fatal reality of Starlink’s business model.

SpaceX is spending hundreds of billions of dollars. They are designing bespoke military-grade silicon. They are trying to force radio waves through an atmospheric black hole. They are building the largest rockets in human history to lift these massive, hyper-complex satellites into orbit.

And who are they doing this for?

The addressable market for Starlink is incredibly small. People living in cities and suburbs will always use fiber optics and 5G. Ground networks are infinitely faster, cheaper, and more reliable. Starlink’s actual market is a hyper-niche demographic: rural farmers, cruise ships, and remote cabins.

SpaceX is building a trillion-dollar infrastructure to serve a market that is fundamentally capped.

And then the final boss: competition. SpaceX is currently living in a golden age. Competition is light years behind. However, it is slowly but surely creeping closer. Competition from China (most notably, LandSpace) and even West (Rocket Labs, Blue Origin, ULA) are starting to gain ground. LandSpace almost has a Falcon-9-equivalent ready. China is building ambitious space ports and launch infrastructure.

Inevitably this golden age will come to an end.

You Cannot Fake the Math Forever

That Starlink is an “infinite-money-glitch” is a myth.

Musk wants the public and his investors to believe that venture capital momentum can bully reality into submission. He loves to spam how cheap marginal costs of his rockets are. He loves to bury true costs of his avant-garde experiments.

SpaceX is forced to build a larger rocket just to launch heavier, vastly more expensive satellites. Burn billions in R&D and NRE for identical capability. Launch costs might be seductive, but the true cost is bankrupting.