When The Economist called Stanford Ovshinsky “the Edison of our age,” the name might have been unfamiliar to most people, but the comparison was apt. Like Edison, Ovshinsky was a prolific self-educated inventor. Among his more than 400 patents were those for the nickel-metal hydride battery, which still powers many hybrid cars, and for a way of mass-producing affordable thin-film solar panels.

These and many other technologies came out of Ovshinsky’s remarkable R&D laboratory, Energy Conversion Devices (ECD), an invention factory like Edison’s Menlo Park. It now appears, however, that Ovshinsky’s most lasting legacy will be something he achieved before the growth of ECD, when he was working as an isolated independent inventor—and, unlike Edison’s creations, it came from a fundamental scientific discovery he himself made.

Ovshinsky began as a machinist and toolmaker in the shops and factories of Akron, Ohio, where he was born in 1922. His first significant invention was an innovative automated lathe, in 1946, and he went on to use automation in other devices. Following the principle of Norbert Wiener’s cybernetics, which looks at self-regulatory systems in machines and organisms, he pursued analogies between machine control devices and animate nervous systems, which resulted in the invention of a powerful electrochemical switch in 1959.

This device depended on thin oxide films covering its tantalum electrodes, which for Ovshinsky were the analogue of nerve cell membranes. When the settlement of a lawsuit with a former partner barred him from using the same materials in developing this switch, Ovshinsky began a systematic search for new materials, a search that resulted in his most important discovery.

He focused on the chalcogens, elements grouped under oxygen in the periodic table (that is, sulfur, selenium, tellurium and polonium) and experimented with thin films of tellurium alloyed with neighboring elements like arsenic and antimony. The result, in 1961, was what is now known as “the Ovshinsky effect”—an almost instantaneous and reversible switching between resistive and conducting states. The effect yielded a threshold switch that turned on or off when the voltage reached or fell below a certain magnitude.

It also yielded a bistable electrical memory, a switch that stayed conductive until a stronger pulse returned it to the resistive state. These semiconducting devices were composed of amorphous (non-crystalline) materials, a feat that had not been considered possible. In the 1960s, solid-state physics dealt almost exclusively with crystals, and it was believed that semiconductor devices like the transistor could only be made from crystalline materials. Coming from an unaccredited outsider and contradicting current scientific assumptions, Ovshinsky’s discovery met with strong initial resistance, but in time it became accepted.

The growing scientific and commercial importance of amorphous and disordered materials is partly Ovshinsky’s legacy, but there is also a more specific contribution that is only now emerging. Phase-change memory, which works by changing from the amorphous to the crystalline state and back, was first successfully commercialized in an optical version, where a laser pulse triggered the change, the basis of rewritable CDs and DVDs. These were in use by the 1980s, but developing the electrical version of the memory took much longer.

To succeed as an information technology, phase-change memory had to improve its performance. Reducing its power requirements was accomplished by several small modifications, while increasing its speed came from one major advance. Research on the optical memory had yielded a chalcogenide alloy (Ge2Sb2Te5) with a very fast switching speed. Ovshinsky believed the alloy would perform even better in the electrical memory, as experiments by his staff spectacularly confirmed.

But even with this success, phase-change memory faced the formidable competition of flash memory, in which chip manufacturers had heavily invested. Phase-change was clearly superior: much faster, requiring much less power, and capable of orders of magnitude more rewrite cycles. But it was also more expensive, and what the market wants, as one of Ovshinsky’s scientists observed, is “cheap and good enough.” For a long time, flash memory has been cheap and good enough.

When Ovshinsky died in 2012, phase-change memory was still waiting for its time to come. Researchers in the field knew that eventually silicon-based flash memory would reach its limit for scaling down and believed that chalcogenide phase-change memory, which works even better as it scales down, would replace it.

Most thought this might take decades. It was therefore a surprise when in 2015 Intel and Micron, which had acquired Ovshinsky’s patents, announced their 3D XPoint chip, calling it “a major breakthrough in memory process technology and the first new memory category since the introduction of NAND flash in 1989.” As more details about the chip emerged, it became clear that it was based on Ovshinsky’s phase-change technology and used a design essentially the same as his researchers had created many years earlier.

As 21st-century information technology advances, Ovshinsky’s phase-change memory, discovered over 50 years ago, seems likely to be his most important legacy.