> LK-99’s purported superconductivity drew immediate scrutiny from scientists. “My first impression was ‘no.’” says Inna Vishik, a condensed matter experimentalist at the University of California, Davis. “These ‘Unidentified Superconducting Objects’, as they’re sometimes called, reliably show up on the arXiv. There’s a new one every year or so.” Advances in superconductivity are often touted for their potential practical impact on technologies such as computer chips and maglev trains, but Vishik points out that such excitement might be misplaced. Historically, progress in superconductivity has had tremendous benefits for basic science, but little in the way of everyday applications. There’s no guarantee a material that is a room-temperature superconductor would be of practical use, Vishik says.
As someone who loves Nature, I would like to commit a dash of heresy. A manufacturable, room temperature superconductor with a low current density wouldn't lead to maglevs, but it would revolutionize the world. It would be one of the starkest turning points in human history.
Most of the heat generated by a CPU is generated shuttling electrons back and forth. A superconductor instantly changes the calculus and, depending on whom you ask, makes processors 500x more efficient. And that's just the start.
The before/after is so stark that it's really hard to game out all of the consequences, but it's quite obvious that it instantly changes the cost of running these massive AI models as well as the cost of doing highly detailed FEM simulations.
Even if some of the properties are replicated, the strangeness of this material puts us on track to that world. It's incredible what it could be. I'm extremely excited!
> it's quite obvious that it instantly changes the cost of running these massive AI models as well as the cost of doing highly detailed FEM simulations.
This seems pretty overstated. Developing the tooling to build chips takes years, there is good reason to think that even if this material does revolutionize chip efficiencies, it will be years at least before we start producing those chips and probably years more before that production scales to fully supplant the existing production infrastructure.
> A superconductor instantly changes the calculus and, depending on whom you ask, makes processors 500x more efficient. And that's just the start.
I'm excited too, but "instantly" is not how I'd describe retooling semiconductor fabrication to include a new material. Besides, nobody has claimed that you can dope silicon with it.
But since we're dreaming, I'm hoping it'll make doing plasma physics cheaper (via simpler superconducting magnets) which will then make nuclear fusion power plants feasible.
Aren’t CPUs prone to quite a lot of current flow even in their off state? If I’m not mistaken that’s the biggest energy draw in case of large servers, but I am no expert on hardware nor physics.
As someone who loves Nature, I would like to commit a dash of heresy. A manufacturable, room temperature superconductor with a low current density wouldn't lead to maglevs, but it would revolutionize the world. It would be one of the starkest turning points in human history.
Most of the heat generated by a CPU is generated shuttling electrons back and forth. A superconductor instantly changes the calculus and, depending on whom you ask, makes processors 500x more efficient. And that's just the start.
https://en.wikipedia.org/wiki/Superconducting_computing
The before/after is so stark that it's really hard to game out all of the consequences, but it's quite obvious that it instantly changes the cost of running these massive AI models as well as the cost of doing highly detailed FEM simulations.
Even if some of the properties are replicated, the strangeness of this material puts us on track to that world. It's incredible what it could be. I'm extremely excited!