Undersea Cables vs. Satellites: What Really Carries the Internet?

The common misconception

Satellites make for a compelling mental image: a dish on a roof, a signal beamed to space and back, data flowing invisibly through the sky. It's an understandable picture, but it isn't how most international internet traffic actually travels. When you video call someone on another continent, or a website hosted overseas loads in your browser, that data almost certainly spends most of its journey inside a fiber-optic cable running along the ocean floor, not bouncing off a satellite.

The mental image persists partly because satellites are visible and dramatic, while submarine cables are, quite literally, out of sight. There's no equivalent of a rooftop dish to notice for fiber — the infrastructure carrying most of the world's international traffic simply disappears beneath the waves a short distance from shore and doesn't reappear until it reaches the next continent.

Ownership has also shifted over the years. Where submarine cables were once built almost exclusively by telecommunications carriers and consortiums of national operators, large cloud and technology companies now co-invest in or directly build some of their own cable systems, reflecting how much of today's internet traffic — video, cloud storage, and large data transfers between their own facilities — flows between their networks specifically.

How submarine cables actually work

Submarine cables are, at their core, the same fiber-optic technology covered in how fiber-optic cables carry the internet as light, engineered to survive years on the seabed. A cable is typically only as thick as a garden hose, but it's built in layers: optical fibers at the center, surrounded by protective steel and polymer armoring, with heavier armor near shore where fishing and anchors pose the greatest risk. These cables run for thousands of kilometers between continents, carrying an enormous share of the world's intercontinental data.

Building and laying one of these cables is a significant undertaking. Specialized cable ships lay the fiber along a carefully surveyed seabed route chosen to avoid the roughest terrain and the highest-risk areas for anchors and fishing gear, then bury the cable in shallower coastal waters for extra protection. In deep ocean, simply resting on the seafloor is usually protection enough — few things down there are likely to disturb it.

Landing stations and redundancy

Each submarine cable comes ashore at a landing station, a facility that connects the cable to terrestrial networks. Along many routes, branching units let a single cable system serve more than one landing point, and periodic repeaters boost the optical signal over the cable's full length, similar in principle to the amplifiers used on long-haul terrestrial fiber. Because a single cable can be damaged — most often by ship anchors or fishing equipment rather than anything dramatic — operators build redundancy into the system: multiple cables along different routes, so traffic can shift to alternate paths if one is cut, while a specialized repair ship locates and fixes the fault.

Cable faults happen more often than most people assume, precisely because fishing and anchoring occur constantly along busy coastlines and shipping lanes. The internet mostly rides through these events without visible disruption exactly because redundancy is built in from the start — the goal isn't a system that never fails, but one where any single failure is absorbed elsewhere on the network while it's repaired.

Geostationary, MEO, and LEO satellites

Not all satellites operate the same way. Geostationary (GEO) satellites orbit far above the equator at a distance that lets them match Earth's rotation, so they appear fixed in the sky — useful for broadcast and some fixed internet links, but their great distance adds real latency. Medium-Earth-orbit (MEO) satellites orbit closer, offering a middle ground between coverage area and latency. Low-Earth-orbit (LEO) satellites orbit much closer still, in large constellations that work together to provide continuous coverage as individual satellites move quickly across the sky — reducing latency substantially compared to GEO, at the cost of needing many more satellites to maintain coverage.

That trade-off between coverage and satellite count is a real engineering constraint, not an afterthought. A single geostationary satellite can cover a huge, fixed portion of the globe on its own, which is efficient in terms of hardware, while a usable LEO service needs a large constellation working in coordination so that as one satellite moves out of view of a ground terminal, another is already positioned to take over.

Why distance to orbit adds latency

Latency over any link is shaped heavily by distance, since even light has a finite speed. A satellite path doesn't just cover the horizontal distance between two points on the ground — it also travels up to the satellite's altitude and back down again. For a geostationary satellite, that round trip to orbit and back adds a substantial, unavoidable delay on top of whatever ground distance is involved. LEO satellites, orbiting far closer to Earth, cut that added distance dramatically, which is a major reason newer LEO satellite services can offer noticeably lower latency than older geostationary satellite internet.

It's a useful reminder that "satellite internet" isn't one single experience. Two services both described as satellite internet can feel completely different in practice — one closer to a typical wired connection's responsiveness, the other closer to what older geostationary satellite internet has historically delivered — depending entirely on which kind of orbit is involved.

Where satellite internet genuinely excels

None of this makes satellites unimportant — it just clarifies their role. Satellite links are often the best or only practical option for rural and remote regions where running fiber is prohibitively expensive, for ships and aircraft that can't connect to any fixed line, and for disaster recovery when terrestrial infrastructure is damaged or unavailable. In these cases, satellite internet provides connectivity that would otherwise not exist at all, which matters far more than any latency gap versus fiber.

Satellite links can also serve as a practical backup for locations that do have fiber, for exactly the redundancy reasons discussed above with submarine cables: if a terrestrial link is cut or a natural disaster damages ground infrastructure, a satellite connection can keep critical services reachable while repairs are underway, even if it's slower than the fiber connection it's temporarily replacing.

Why routes can still be indirect

Even with dedicated submarine cables between two countries, traffic doesn't always take the most direct route available. Business relationships between networks, available capacity, and current routing policy — covered in more depth in how routers find a path across the internet — can send traffic through a third country's infrastructure even when a more direct cable exists. This is a routing decision, not evidence that cables don't matter; it's simply how a network of independently operated systems behaves.

What this means for your connection

If you're testing a connection to a server on another continent, expect meaningfully higher latency than testing a nearby server, largely because of real physical distance over fiber — and if you're specifically using satellite internet, expect that gap to be larger still, especially over a geostationary link. Neither is a flaw in your setup; it's physics. Choosing a nearby test or content server, where available, is one of the most effective ways to minimize latency regardless of which physical medium ultimately carries your traffic.