Satellite vs Fiber Latency Physics

The two speeds of light

Light in glass fiber travels at roughly 2/3 the speed of light in vacuum because the glass's refractive index slows it down. Per kilometer, fiber takes about 50% longer than vacuum.

GEO satellite: the worst case

A geostationary satellite is at altitude ~35,786 km. A signal goes up to the satellite and back down — round-trip distance ~71,600 km plus the terrestrial ends. At the speed of light in vacuum:

71,600 km / 299,792 km/s = ~240 ms minimum latency

Add ground-side processing and the typical fiber backhaul, and total RTT through a GEO satellite link is 500-700 ms. This is why GEO satellite internet has historically been associated with painful lag — the physics is unavoidable.

LEO satellite: much closer

A LEO satellite at 550 km altitude has a round-trip path of ~1,100 km plus the slant distance to the user. Typical end-to-end latency:

~1,100 km / 299,792 km/s = ~3.7 ms (the pure RTT to satellite and back)
Plus ground-side and processing: ~20-40 ms total ground-to-ground via one satellite hop

This is comparable to or better than terrestrial fiber over similar distances. The 240 ms GEO penalty is gone.

Long-distance routes: the satellite advantage

For an intercontinental route (e.g., New York to London):

• Fiber: trans-Atlantic cable plus routing within each country. ~5,500 km of cable distance, perhaps 6,500 km of actual signal path. At 5 μs/km: ~32 ms one-way, ~65 ms RTT plus processing.
• Satellite (LEO with ISL): straight-line great-circle distance is ~5,500 km. Going up to LEO, hopping across satellites in vacuum, and coming back down: extra ~3,000 km of slant + ISL distance. But traveling that in vacuum at full speed of light: ~28 ms one-way, ~56 ms RTT.

The satellite path can be ~10-15% faster. For specific high-value applications (financial market data, real-time trading), this matters.

Why fiber wins on short routes

For a route within a city or even a country, fiber wins for one reason: it has no up-and-back-to-space overhead. The ~1,100 km of up-and-down to LEO is fixed; for a 100 km city-to-city link, fiber's ~500 μs path is way shorter than satellite's ~5 ms minimum.

Fiber dominates for short and medium-distance routes. Satellite only catches up when the great-circle saves enough kilometers to offset the vertical detour.

The crossover distance

Roughly speaking, satellite starts to beat fiber when the great-circle route is more than ~3000-4000 km. Below that, fiber's direct path wins despite the medium being slower. Above that, the satellite's vacuum-speed and direct routing advantages compound.

The "fiber routes aren't straight" factor

Real-world fiber rarely follows a great-circle path. Cables follow geography (along coastlines, around mountains, through specific peering POPs). Add: routing through specific gateways, congestion-related path changes, and the fact that data centers aren't on the great-circle line. The effective distance through fiber is often 1.2-1.5× the geographic great-circle distance.

A satellite path going through vacuum can be essentially straight-line, modulo the up-and-down detour. This compounds the satellite's per-km speed advantage.

Why this matters for trading

High-frequency trading firms care intensely about latency. A few milliseconds can be the difference between a profitable trade and a missed opportunity. For decades, the industry invested in dedicated fiber links straight-line across continents to minimize latency.

LEO constellations with inter-satellite laser links can theoretically beat fiber on transatlantic routes. As constellation density and routing optimize, satellite will become competitive for the most latency-sensitive workloads. Most internet traffic doesn't need this; for trading, it's a real shift.

Why fiber still dominates aggregate traffic

• Bandwidth. A single submarine fiber cable carries tens of Tbps. A whole LEO constellation carries low-Tbps aggregate. Per-bit cost on fiber is far lower.
• Reliability. Fiber doesn't have rain fade, satellite handoffs, or pass timing. Once installed, it's predictable.
• Cost. Operational cost per bit is much lower on fiber.

The right framing is: satellite for latency-sensitive applications over long distances where bandwidth is modest; fiber for bulk bandwidth and shorter routes.

What the user sees

Most consumer satellite-internet users on LEO services see ground-to-ground latencies of 30-50 ms — better than the 500+ ms of GEO, similar to terrestrial broadband for most measurements. Long-distance connections (intercontinental) can be better than fiber for users on Starlink-class services routed through their ISLs.

The physics also caps how good the latency can be. The speed of light in vacuum is fast but finite. LEO satellite isn't faster than instant — it's a few tens of milliseconds, not single digits.

Air links and microwave

For specific routes, dedicated terrestrial microwave links also beat fiber. Microwave travels through air at near-vacuum speed; for line-of-sight paths between tall buildings or microwave towers, it can be slightly faster than fiber. Used historically by trading firms for Chicago-to-NJ links. The technique doesn't scale to oceans, but reinforces the principle: any medium with a higher speed of light than fiber wins on latency.

Frequently Asked Questions

How can satellite be faster than fiber?

Two physics facts. First, light travels about 30% faster in vacuum than in glass fiber (300,000 vs 200,000 km/s). Second, LEO constellations with inter-satellite laser links can route great-circle paths over the Earth instead of following winding fiber routes. For long-distance links (intercontinental), the combination means a properly-routed satellite path can beat fiber by 30-40%.

Why is GEO satellite always slower than fiber?

Because the round trip to geostationary orbit is ~72,000 km. At the speed of light, that alone takes ~240 ms — added to whatever ground-side processing happens. Even though light travels in vacuum, the sheer distance dominates. GEO inherent latency is far worse than fiber for any terrestrial distance. LEO satellites (550-2000 km altitude) avoid this because they're much closer.

What is the speed of light in fiber?

Approximately 200,000 km/s — about 2/3 the speed of light in vacuum. The slowdown comes from the refractive index of the silica glass; light interacts with the medium and effectively travels slower. The same is true of any transmission medium denser than vacuum. For most internet calculations, "5 microseconds per km of fiber" is a useful rule of thumb.

For what routes does satellite win?

Intercontinental routes where the satellite path is shorter and through vacuum. For example: New York to London by fiber is ~5,500 km of trans-Atlantic cable plus routing inside each country. By satellite, the great-circle path is ~5,500 km of straight-line distance traversed mostly in vacuum. Satellite can beat fiber by tens of milliseconds. For short routes (within a city, within a country), fiber wins because it has no up-and-back-to-space overhead.

Why doesn't satellite always replace fiber?

Two reasons. Bandwidth: a single fiber pair carries tens of Tbps; a satellite carries Gbps. For total volume, fiber dwarfs satellite. Cost: terrestrial fiber, once built, has very low per-bit cost; satellite has higher per-bit operational cost. Latency wins for some specific use cases (financial trading, video calls across continents), but for most internet traffic the marginal latency improvement isn't worth the bandwidth and cost tradeoffs.