Satellite Internet: How It Works

Podcast Transcript

When the Internet was first launched, it was only available on a few computers at a few research institutions.

Over the last 50 years, internet access has expanded to cover more institutions and more computers. Eventually, it was available in our homes and even in our pockets.

Recently, the final step in creating a fully ubiquitous internet was taken, enabling access from any point on the Earth’s surface.

Learn more about satellite internet and how it works on this episode of Everything Everywhere Daily.

In previous episodes, I discussed how satellites work and the evolution of the communications satellite. In this episode, I want to zoom in specifically on satellite-based internet, which is very different from the communications satellites that came before it.

Just to recap, the idea of a communications satellite was developed in the 1940s by science fiction author Arthur C. Clarke. He realized that if you put a satellite in orbit high enough, it could reach a point where its orbital period matched the rotation of the Earth, one day.

From this point, you could aim a satellite dish at a spot in the sky and receive signals beamed down to the surface, where anyone within the radio signal’s footprint could receive the transmission.

Soon after Sputnik was launched, communication satellites were being put into orbit. These satellites worked fine for broadcasting media such as television and radio, and even for limited two-way communication.

Satellite television became common starting in the 1980s, and it worked because many people received the same satellite signal. A small number of signals could be sent to the satellite, and millions of people could then receive them on the way down.

When the internet exploded in popularity in the 1990s, using satellites for internet access was obvious, but there were many problems with this idea.

Geosynchronous satellites sit at 35,786 km or 22236 miles above Earth. That distance is enormous compared to terrestrial networks or even low Earth orbit satellites.

A single data request doesn’t just go up and back once. In a typical internet interaction, it often involves multiple round-trips between your device and a server. Even a single round-trip introduces a delay of roughly 240 milliseconds, and real-world connections are often closer to 500–700 ms.

That delay is called latency, and it’s the fundamental weakness of geosynchronous satellite internet.

Geosynchronous satellites also have limited total capacity. They cover huge portions of the Earth, which sounds like an advantage, but it means many users share the same satellite bandwidth.

As more people connect, speeds can drop, especially during peak usage times. Compared to fiber or cable networks, scaling capacity is much harder and more expensive.

That isn’t to say you can’t access the internet over these sorts of satellites; it’s just slow and expensive.

During my travels, I’ve accessed the internet several times via these satellite connections. In 2007, I had to go to the communications company’s main office on Majuro, in the Marshall Islands, to get internet access. It was so slow that anything beyond looking at a simple website was impossible.

It was on a par with a slow dial-up modem, and you had to pay a premium to use it. Likewise, I used a satellite connection on the island of Saint Helena. It was so slow that I was only able to log on once over the course of three weeks.

If satellites were to be a part of the internet, a different approach would be required.

The first serious attempt to create a dedicated satellite internet service was the founding of Teledesic in 1994. It was the brainchild of two billionaires, Bill Gates of Microsoft and Craig McCaw of McCaw Cellular, which was later purchased by AT&T.

I have been heavily involved with the internet since I started an internet company in 1994. When I first learned about Teledesic, I followed it religiously because I found the idea of satellite internet so compelling.

Teledesic was designed as a low Earth orbit broadband network, not a traditional satellite system. Instead of acting as simple relay satellites, each satellite would function as a node in a space-based packet-switched network, communicating with neighboring satellites and routing data globally.

The goal was to provide “fiber-optic-like” broadband anywhere on Earth, including for real-time applications such as video conferencing and multimedia.

The constellation went through several iterations. The original 1994 proposal called for 840 active satellites, but later designs reduced that to 288 after cost-cutting and redesigns.

The large number was required because, unlike geostationary satellites, low Earth orbit satellites move quickly across the sky and each covers only a small portion of Earth at any given time.

The initial design plan was to have the satellites at an altitude of about 700 km in near-polar orbits, which was revised to a higher altitude of bout 1,300–1,400 km.

Teledesic ultimately failed because it tried to execute a technically sound vision before the economics and infrastructure could support it. Launch costs in the 1990s were prohibitively high, meaning deploying hundreds of satellites would have required tens of billions of dollars, far beyond what investors were willing to sustain after the dot-com bubble collapsed.

The required technologies, including cheap mass-produced satellites, phased-array antennas, and efficient inter-satellite networking, were not yet mature or affordable.

Teledesic was a good idea that was before its time.

In 1996, Hughes Network Systems launched DirecPC, the first consumer satellite internet service.

This, and similar systems, were one-way connections. Users downloaded data via satellite but still needed a dial-up modem for uploads. To be fair, download bandwidth is usually much greater than upload bandwidth, but anyone who was using such a satellite connection was usually doing so because they had no other choice.

They probably lived in an area without cable or DSL service

The business case for satellite internet never disappeared. As the world became more connected and people’s lives increasingly depended on the internet, the case for universal access across the planet became more compelling.

Advances were slowly being made that made a low Earth orbit satellite constellation more feasible. The cost of computing continually decreased, enabling the construction of smaller, cheaper satellites. Likewise, the cost of solar panels dropped, allowing for cheaper, more efficient power for satellites.

The missing piece was the cost of launching a satellite into space. Very little progress had been made in terms of reducing launch costs.

The radical innovation that reduced the cost of reaching orbit was the reusable rocket, pioneered by SpaceX.

While SpaceX’s stated goal was to reduce the cost of spaceflight and ultimately enable human settlement of Mars, the company quickly encountered a fundamental economic problem: launching rockets is capital-intensive, cyclical, and dependent on external customers.

Solving the problem of reusable rockets wouldn’t mean much if they didn’t have enough customers who wanted to put satellites in orbit. Launches had traditionally been so expensive that there were very few satellite launches per year.

Starlink emerged in the mid-2010s as a solution to that problem, a way to generate steady, recurring revenue by leveraging SpaceX’s launch capabilities to build a global communications network.

The first prototype satellites were launched in 2018, followed by the first operational batch of 60 satellites in May 2019, marking the beginning of the largest satellite constellation ever deployed.

From the beginning, Starlink was conceived not as a traditional satellite system but as a low Earth orbit broadband network designed to overcome the latency limitations of geostationary satellites. They were going to achieve the dream first envisioned by Teledesic in the 90s.

By operating at altitudes around 550 kilometers, Starlink could deliver latency low enough for real-time applications like video calls and gaming, something earlier satellite systems struggled to achieve.

The relationship between SpaceX and Starlink is unusually tight and mutually reinforcing, to the point that each arguably exists to sustain the other. SpaceX makes Starlink possible primarily through its reusable rocket technology, especially the Falcon 9.

By dramatically reducing the cost per launch, SpaceX enabled the deployment of thousands of satellites at a scale that would have been economically impossible in previous decades.

Starlink launches are now among the most frequent missions flown by SpaceX, turning the company into not just a launch provider but the world’s largest satellite operator.

As of the time of recording, there are more Starlink satellites in orbit, a little under 10,000, than have been put in orbit by everyone else in history.

At the same time, Starlink makes SpaceX viable by providing a massive, recurring revenue stream. Launch services alone are sporadic and competitive, but Starlink subscriptions generate continuous income from consumers, businesses, and governments worldwide.

This revenue helps fund SpaceX’s more ambitious projects, particularly the development of the fully reusable Starship system, which I’ve covered in previous episodes.

This was the piece of the puzzle that previous satellite and communications companies lacked. Because they didn’t control the launches, they couldn’t reduce costs as much as they wanted.

The Starlink system is composed of three primary elements: satellites, ground infrastructure, and user terminals. The satellites operate in low Earth orbit, moving rapidly across the sky and handing off connections seamlessly from one to another.

Unlike earlier systems, Starlink satellites increasingly communicate with each other using laser inter-satellite links, creating a mesh network in space that can route data without always relying on ground stations.

While it might not seem like much, the fact that Starlink lasers can operate in the vacuum of space actually makes them faster than fiber-optic connections on Earth.

Users connect via a flat, electronically steered phased-array antenna, often erroneously called a “dish,” which can track satellites automatically without mechanical movement.

A phased-array antenna is a group of many small antennas that electronically adjust the timing of their signals to steer and focus a radio beam in different directions without physically moving the antenna. It is flat, not concave like a typical satellite dish.

Ground stations link the constellation to the terrestrial internet, although over time, the system is evolving toward more space-based routing as laser links expand.

Most Starlink satellites operate at around 540 to 570 km in their primary orbital shell, whereas the International Space Station orbits Earth at an altitude of about 400 to 420 km.

Starlink satellites are relatively small and inexpensive by traditional space standards: early versions weigh about 260 kg or 573 pounds and are roughly the size of a table, while newer V2 satellites are larger at around 800 kg or 1700 pounds and a few meters across.

Mass production has driven prices down dramatically, with estimates of roughly $250,000 to $500,000 per satellite for early versions and closer to $1 million for newer, larger, more capable models, excluding launch costs.

The impact that Starlink has already had has been dramatic.

In many rural and remote regions, traditional broadband was never economically viable. Running fiber across mountains, deserts, or sparsely populated areas simply doesn’t pay off. Starlink changed that overnight by making high-speed internet available anywhere with a clear view of the sky.

This has been especially impactful in places like rural North America, parts of Africa, remote islands, and isolated communities where people went from dial-up or no connection at all to broadband capable of video calls and streaming. It has effectively collapsed the geographic barrier to connectivity in a way no previous system has.

Starlink has also had major geopolitical and military consequences. Its use in Ukraine demonstrated that a decentralized satellite network can provide resilient communications even when terrestrial infrastructure is destroyed or jammed.

This has forced governments and militaries to rethink their communications strategy. Instead of relying solely on centralized systems, they now have access to a distributed, rapidly deployable network that is difficult to disable. At the same time, it has raised concerns about private companies controlling critical infrastructure during conflicts.

When hurricanes, earthquakes, or wildfires knock out local networks, Starlink terminals can be deployed quickly to restore communication. Emergency responders have used it to coordinate relief efforts, reconnect hospitals, and provide temporary internet access to affected populations.

Currently, Starlink has a de facto monopoly on satellite internet, but other companies are planning to compete.

OneWeb already has hundreds of satellites in orbit and focuses on enterprise, aviation, and government connectivity rather than direct consumer service.

Amazon is developing Project Kuiper, a planned constellation of over 3,000 satellites intended to deliver global broadband, leveraging Amazon’s cloud and logistics ecosystems.

Traditional geostationary providers like Viasat and SES S.A. are also evolving, investing in higher-capacity satellites and hybrid networks that combine GEO and medium-Earth-orbit systems.

Meanwhile, China is pursuing its own large-scale constellation projects, signaling that satellite internet is becoming a globally competitive and strategic industry rather than the domain of a single company.

Satellite internet is not for everyone. If you live in a place that can get DSL or fiber, it is probably a better option. However, for much of the rest of the planet, including places such as the South Pole and remote wilderness areas, satellite internet offers the promise of connectivity for everything, everywhere.