Researchers at Sydney Nano Hub at Sydney University may have cracked two problems at the same time. They’ve built a nanochip using photons instead of electricity.
This chip is a strange-looking creature, nothing like standard chips. It’s “sort of recognizable” as obviously being an electronic component. There, the resemblance ceases. Compared to a normal chip, it looks like a map of some desert country.
Photonic computing is based pretty much on the same theory as conventional computing, but far more efficient in all aspects of processing. You don’t need endless miles of nano wires and the conductivity that generates so much inefficient heat. Just think of the ridiculous levels of energy usage in crypto operations. AI could be far more demanding.
Heat and water usage are AI’s key vulnerabilities at this stage. The headlines are quite rightly full of rage against the high water usage requirements of AI data centres, with good reason.
The “world’s most mismanaged critical resource” doesn’t need any more burdens on volumes of usage. It’s also a huge logistics problem for AI data management.
Add to this the fact that the AI sector definitely does NOT need a big, expensive, and likely to be very unpopular, highly regulated addition to the balance sheets. It’s a train wreck in plain sight.
Note: Nobody’s actually saying that photonic computing will solve this problem, but less conductivity should mean less heat in the systems. It should also mean less attrition on the components.
Check out this AI overview regarding photonic computing, lower energy usage and most importantly, easy integration with existing systems. For a nice change, a shift to a new technology could be relatively painless to the infrastructure.
The key selling point to nanophotonic computing is speed. This is another core inefficiency that could be on its way out. At the consumer level, it’s all good. Faster games, no lag, no poor little fans gasping their way through another session.
Well done, Sydney Nano Hub and Sydney University. You might have just headed off decades of added costs and inefficiencies.
There’s a lot to like about nanophotonic computing.
_______________________________________________________
Disclaimer
The opinions expressed in this Op-Ed are those of the author. They do not purport to reflect the opinions or views of the Digital Journal or its members.
