“Any sufficiently advanced technology is indistinguishable from magic.” Arthur C. Clarke, the English science-fiction writer and author of 2001: A Space Odyssey. Compared to the 1960s when Kubrick’s film based on Clarke’s book was released, in today’s world, microchips have become the tiny but incredibly powerful base of any technological instrument, from your smartphone to the landers used to explore Mars. They can be as small as a fingernail while containing hundreds of thousands of transistors. The exponential growth of microchips does have limits though, and making them more powerful while continuing to reduce the size of their components has become a challenge. With new Artificial Intelligence machines, the ability of chips to transfer data is also limited. Several scientists have therefore begun looking for a solution in a different direction: light.
Optical fibres, in which the flow of electrons is replaced by beams of light, have been around for decades. But it is not that easy to adapt this technology to microchips and there is an increasing interest, at an investment level, in start-ups that seek to use light instead of electrical energy. This is a highly innovative technology that could have a huge impact on many tech environments, from the self-driving automotive sector to AI. For example, the Californian start-up, Ayar Labs, has recently received funds of over 130 million dollars to develop a technology called Silicon Photonics that allows data to be transferred between computers via cable-free optical rays. Lightmatter, another American start-up, has designed a microchip called Passage that uses light to allow different kinds of chip, from processors to graphic cards, to communicate at an incredible speed. Passage will reduce both energy consumption and the costs that are normally required to construct complex systems.
Replacing electrons with photons is a challenge that has been taken up by numerous researchers, including an MIT team. This team is currently creating components for microchips that use signal processing techniques to reduce both energy consumption and the space available on the chip, a breakthrough that allows them to become considerably more powerful than their electrical predecessors. Once again in Boston, the Lightelligence start-up is developing accelerator chips that use light to power models of artificial intelligence. Compared to electricity, this method will allow algorithms to be processed literally at the speed of light.