Brian Hayes in American Scientist:
When Bell Telephone Laboratories announced the invention of the transistor in 1948, the press release boasted that “more than a hundred of them can easily be held in the palm of the hand.” Today, you can hold more than 100 billion transistors in your hand. What’s more, those transistors cost less than a dollar per billion, making them the cheapest and most abundant manufactured commodity in human history. Semiconductor fabrication lines churn out far more transistors than the world’s farmers grow grains of wheat or rice.
In this thriving transistor monoculture, can a new circuit element find a place to take root and grow? That’s the question posed by the memristor, a device first discussed theoretically 40 years ago and finally implemented in hardware in 2008. The name is a contraction of “memory resistor,” which offers a good clue to how it works.
Memristor enthusiasts hope the device will bring a new wave of innovation in electronics, packing even more bits into smaller volumes. Memristors would not totally supplant transistors but would supplement them in computer memories and logic circuits, and might also bring some form of analog information processing back into the world of computing. Farther out on the horizon is a vision of “neuromorphic” computers, modeled on animal nervous systems, where the memristor would play the role of the synapse.