Analog, iterative machines ditch binary transistor switches for photons and electrons to process at the speed of light.
Presenting its findings as "Unlocking the future of computing" Microsoft is edging ever closer to photon computing technology with the Analog Iterative Machine (AIM). Right now, the light-ba
Optical components are already used in some parts of servers, in interconnects. But I don't expect them to replace silicon for general purpose processing ever. One thing that's never noted in these scientific press releases is that optical components are huge. The wavelength of light is about a micron, i.e. a thousand times larger than the feature sizes of silicon electronics. This limitation can't be easily overcome.
The wavelength of light is about a micron, i.e. a thousand times larger than the feature sizes of silicon electronics. This limitation can’t be easily overcome.
"Can't easily be overcome" is true but it's worth bearing in mind that the features in silicon electronics you're referring to are actually etched by light in many cases. There may be ways to cheat, even if it doesn't sound intuitive.
IBM has been working on optical chips since the early 2000s, but still hasn't released one yet. Maybe this news will get them to show off what they've got now if they're still working on it
Knowing IBM they could have got these things working in the 2010s and then just never done anything with it; that company has a history of pre-empting major breakthroughs and then just leaving them collecting dust in a basement.