The Forgotten History of Submarine Cables: How Glass Threads Replaced Copper Wires

Almost every byte of intercontinental internet traffic moves through glass fibers laid on the ocean floor. Satellite links carry less than five percent of traffic and would be impossibly inadequate at modern volumes. The transatlantic fiber network is one of the most consequential pieces of infrastructure in human history, and almost nobody outside the cable industry can name a single cable. The story of how it got built is one of the great unloved engineering arcs of the modern world.

The first attempt

The transatlantic telegraph cable of 1858 was funded by Cyrus Field, a New York paper merchant who had retired at thirty-three and was looking for a project. The technical lead was Charles Tilston Bright, a young British engineer, and the scientific advisor was William Thomson, later Lord Kelvin. The cable was 2500 nautical miles of single-strand copper wire wrapped in gutta-percha and armored with iron, weighing 2000 pounds per mile.

The first three laying attempts in 1857 and 1858 failed: the cable broke and was lost. The fourth attempt in August 1858 succeeded, and the first message ("Glory to God in the highest; on earth, peace and good will toward men") was sent on August 16. Queen Victoria and President Buchanan exchanged 99-word congratulatory messages that took 16.5 hours each to send, mostly because the cable was already failing.

The cable went completely silent on September 1, after about three weeks of operation. The investigation showed that the chief electrician, Edward Whitehouse, had been applying voltages of 2000 volts to push signals through, on the theory that higher voltage would penetrate the cable's electrical resistance. He had ignored Thomson's calculations showing that the cable's capacitance dominated its behavior at telegraphic frequencies, and that high voltage would actually damage the insulation. He was right that signals were weak. He was wrong about why.

The Thomson mirror galvanometer

The next attempt in 1865-1866 used a much heavier cable (5800 pounds per mile) laid by the Great Eastern, then the largest ship in the world, and used Thomson's mirror galvanometer for signal detection. The galvanometer detected currents of a few microamps, which let the cable operate at voltages a hundred times lower than Whitehouse had used. Speed went from a few words per hour to about eight words per minute, and the cable lasted decades.

The mirror galvanometer is one of the great unsung instruments of nineteenth-century physics: a light beam reflected off a tiny mirror attached to a coil that rotated in a magnetic field. The displacement of the light beam on a distant scale amplified small currents into visible motion. The technique survived in physics laboratories well into the twentieth century, and the underlying principle (small motion amplified by optical leverage) reappears in scanning probe microscopes and laser interferometers.

The cable empire

By 1900, a network of telegraph cables spanned every major ocean, mostly operated by British companies and routed through British colonial possessions. The Pacific cable from Canada to Australia was completed in 1902, and the global telegraph network was effectively complete. London became the world's information capital not because of any natural geographic advantage but because the cables converged there.

The strategic implications were significant enough that the cable network became a target in both World Wars. The British Royal Navy's first operational act in August 1914 was to dredge up and cut the German transatlantic cables, isolating Germany from direct telegraph contact with the Americas for the duration of the war. The American CIA's cable-tapping operations in the Cold War were the direct descendant of this strategic recognition.

The transition to telephone cables

The first transatlantic telephone cable (TAT-1) was completed in 1956 and carried 36 simultaneous voice circuits at a construction cost of about 47 million dollars (1956 dollars). The technical challenge was repeater amplifiers that could operate underwater for decades without maintenance: any failure meant losing the whole cable. AT&T's Bell Labs developed valve-based repeaters with calculated lifetimes longer than the expected cable lifetime, achieved by overdesigning every component and by running the valves at conservative duty cycles.

The submarine telephone cables ran for thirty years, gradually being replaced by satellite links in the 1970s and 1980s for long-haul voice. By 1985, satellite carried about half of all transatlantic voice traffic. The conventional wisdom was that submarine cable was a mature technology with limited future. This was wrong.

The fiber revolution

The first transatlantic fiber-optic cable (TAT-8) was completed in 1988. The technical breakthrough was the optical amplifier: a length of erbium-doped fiber that boosted signals without converting them back to electrical form. This allowed all-optical regeneration over long distances. TAT-8 carried 280 megabits per second, equivalent to about 40000 simultaneous voice calls. Within five years it was congested.

The capacity progression after that was extraordinary. TAT-14 in 2001 carried 640 Gbps. MAREA in 2017 carried 160 Tbps. The 2024 generation routinely carries 250+ Tbps. Each generation has cost less per bit than the last, while the global demand has grown faster than the capacity. The 2020s buildout is funded primarily by the hyperscalers (Google, Meta, Microsoft, Amazon), who collectively own or lease most modern transatlantic capacity. The telecom carriers that ran the previous generations are minority partners now.

The cable-laying ships

There are currently about 60 cable-laying ships in the world, and they are mostly owned by a handful of specialized companies. The work is physically demanding and dangerous: cables are loaded into onboard tanks holding 4000 nautical miles of cable each, and laid at speeds of 6-8 knots. A cable break in deep water can take weeks to repair, and the repair ship has to find a kilometer of cable on a featureless ocean floor, raise it, splice in a replacement section, and lower it back to the bottom.

The biggest cause of cable breaks is fishing trawlers and anchors, accounting for about 70 percent of incidents. Submarine earthquakes account for another 10 percent, with the 2006 Hengchun earthquake near Taiwan cutting nine cables simultaneously and disrupting Asian internet for weeks. The remaining incidents are equipment failure and the occasional intentional cable cut.

The deeper observation

The submarine cable network is one of the canonical cases of infrastructure that nobody notices when it is working. The five-percent satellite share is the headline; the 95 percent cable share is the load-bearing reality. Internet outages large enough to make the news are almost always cable-related, but the news rarely identifies the cable or the cause. The infrastructure is largely invisible to its users and to the political processes that affect it.

The 170-year arc from Cyrus Field's wire to Meta's optical fiber is the same as many other technology arcs: a brief heroic era of obvious public attention, followed by a century of incremental improvement during which the technology becomes invisible by becoming reliable, followed by a regulatory and strategic moment when the infrastructure becomes visible again because it is now load-bearing for everything else. The 2024 attention on cable cuts in the Baltic is the third act of a story whose first act started in 1858. The next century will be the long fourth act, and most of it will happen quietly.