Trouble is, existing superconductors require extreme cold to reach the required state, or they can work in warmer temperatures at very high pressures indeed. In MRI machines, the magnets are kept frigid with liquid helium.
That makes MRI machines very complex, cumbersome and expensive, with high power usage. They illustrate why researchers are frantically looking for superconductor materials that don’t need extreme cooling but work, ideally, at room temperature and ambient atmospheric pressure.
LK-99 is supposed to be such a superconductor.
You could even make the compound material in your kitchen, cheaply and, by the looks of it, quite easily.
The worldwide excitement around LK-99 has been palpable. Social media has been rife with videos of little specks of LK-99 hovering over a magnet, demonstrating one of the characteristics of superconductivity, called Meissner effect levitation.
This in turn saw breathless stories about the possibility of iPhones as powerful as current supercomputers that take up large amounts of space, batteries with massive capacity, nuclear fusion being made more feasible, thin and small electricity transmission grids that can handle enormous currents, and much more more, thanks to LK-99.
Unfortunately, despite many efforts from scientists and talented amateurs (and many fake levitation videos), LK-99 is probably a sad trombone rather than a superconductor.
So far, it seems the LK-99 replication efforts mainly show that humanity wants to believe, no matter how implausible the notion.
Never say never, though. What was written off as science fiction not so long ago has often come true, just not the way the writers imagined. If electronics based on high-temperature superconductive materials are an engineering conundrum and do not violate the laws of physics, there’s a chance they’ll come into existence.
Like building electronic and computer components that use light, which we have been doing for some years now. It’s referred to as photonics and optical computing, and opens mind-blowing new possibilities that simply weren’t possible before.
Light is faster than electrons, which by itself provides a performance bump that cuts down on the time a circuit is active; coupled with lower power usage as seen in, for example, optical fibre networks, moving away from electronics could get us several magnitudes ahead when it comes to computing prowess.
This is without increasing the size of the devices and losing energy as heat that has to be removed.
Some applications are blue-skying, but Intel and other chip vendors have already married semiconductors with optics to create commercially available components with higher performance and lower power consumption than traditional parts.
In June this year, Microsoft announced the Analogue Iterative Machine (AIM), which is designed to solve difficult optimisation problems for industries such as logistics, finance, healthcare and manufacturing and others.
AIM goes beyond binary logic with zeroes and ones and works with continuous value data, Microsoft says. That is, it’s an analogue device, more akin to humans than digital computers that approximate the world with 0 and 1 values.
Understanding how AIM works isn’t easy, and it’s early days, but opto-electronic devices are said to get around the diminishing returns when it comes to increasing computing capacity that we’re seeing currently.
Microsoft says the components to build AIM are miniaturised onto tiny centimetre-scale chips, and the whole thing fits into a rack, or shelf enclosure. It’s compact, and the low-cost technologies for AIM are already in consumer products, made by existing manufacturers.
There is even an AIM simulator available, which you can ask Microsoft to let you try out.
Importantly, if we’re going to continue down the path of artificial intelligence, it’s difficult to see how it can be achieved without better energy efficiency. The way to make AI less hallucinatory and perform faster is not by optimising and cleaning the giant amount of training data used.
Instead, it’s a matter of throwing more computing power at increasingly large systems, which won’t be sustainable in the long run without a fundamental rethink of the underlying technology used to power AI.
Pour one out for the advanced materials scientists and everyone else who makes possible technology that’s almost indistinguishable from magic; they deserve it, and we’re going to need what they come up with sooner rather than later.
