Fiber Optics: How Glass Wires Replaced Copper

A strand of glass thinner than a human hair now carries almost everything that travels on the internet. This video traces how that happened, from a Victorian lecture-hall trick to a research record set in London this year.

It starts with the physics: total internal reflection, first shown in public by John Tyndall at the Royal Institution in 1854, trapping light inside a curving jet of water. Early 20th-century glass fiber could not use the trick for anything real — the glass itself was too impure, losing almost all its light over a short distance, so long-distance phone and television networks ran on copper cable and point-to-point microwave relay instead.

In 1966, Charles Kao and George Hockham argued the problem was purity, not physics, and set a target: fiber loss below 20 decibels per kilometre. On 7 August 1970, Corning's Robert Maurer, Donald Keck and Peter Schultz hit it, doping silica glass first with titanium and then with germanium. By 1979, NTT had pushed loss down to about 0.2 decibels per kilometre, near the theoretical floor, and commercial fiber telephone links followed.

TAT-8 crossed the Atlantic on fiber in 1988, the first ocean-crossing fiber cable, carrying roughly 40,000 telephone circuits over a consortium route built by AT&T, British Telecom and France Télécom. Through the 1990s and 2000s, falling costs let fiber displace copper and microwave relay across the whole backbone — the long routes between cities, countries and continents, and eventually the data centers behind the modern internet.

The video then covers the engineering that keeps stretching fiber's capacity: wavelength-division multiplexing, which carries many separate streams of data as different colours of light on one strand, and the repeater stations and optical amplifiers that keep a signal readable over thousands of kilometres. It closes with where the records stand now — a joint NICT/University College London test reported 450 terabits per second over installed fiber in London in 2026, a figure this video is careful to explain correctly: it is a theoretical capacity estimate, not the 418 terabits per second the test actually decoded — and where the glass itself is heading next, with hollow-core fiber's lower-loss promise and the unfinished task of reaching the regions fiber has not.

Every figure is traced to a named, dated source in the production fact sheet, including two corrections made against primary sources after the original research brief did not hold up.

Educational documentary. Not financial or investment advice.