35 ms·
Posting for Cunningham's law :) * Green energy suddenly becomes way more viable. Megaprojects in the most efficient sites can send energy long-distance and sto
by losteric 3y ago
Posting for Cunningham's law :)
* Green energy suddenly becomes way more viable. Megaprojects in the most efficient sites can send energy long-distance and store it with effectively no loss, somewhat mitigating regional variations (especially if we have a high-trust world order where a united global grid is viable). (I read LK99 might have some limitations carrying lots of current but presumably other approaches would do better)
* EVs: improved performance of motors, batteries, charge time, and weight - huge shift for the market. Much safer than most current car batteries too.
* Big breakthrough for computing in the form of fast, cool, and efficient zero-resistance transistors. Step change for cutting-edge component performance, all the cloud hyperscalers would completely revamp their compute. TSMC / ASML probably get huge volumes of new orders.
Obviously the first bet is following the patents. Otherwise, my play would be the industrial companies that build things that build things, like factory automation companies, followed by companies that would see a surge in demand from products incorporating room-temp superconductor technology, like TSMC, ASML (maybe Apple/AWS).
- jakeinspace 3y agoSorry, but how does one create a transistor from a superconductor? Maybe I’m missing something here.
- bpye 3y agoI don’t think we’d necessarily have to use it for transistors, but could use it in place of the metal interconnect that is significant in terms of resistive loss.
- junofan 3y agoI think Josephson junctions. Also maybe you could eliminate copper loss by using LK-99 interconnect layers.
- floxy 3y agoMost superconducting logic families aren't using transistors at all. They use Josephson junctions, which are just two pieces of superconductor separated by a non-supercondutor. RSFQ (Rapid Single Flux Quantum) uses millivolt-high picosecond-long pulses to represent logic 1 and their absence as logic zero, instead of using voltage level as in CMOS. https://ieeexplore.ieee.org/document/783712 https://ieeexplore.ieee.org/document/783712 (770 GHz toggle flip-flop in 1999) https://www.physics.sunysb.edu/Physics/RSFQ/Projects/WhatIs/rsfqre2m.html https://www.physics.sunysb.edu/Physics/RSFQ/Projects/WhatIs/...
- Culonavirus 3y agoSooo, a 770 GHz processor? Do I understand it correctly? Jesus.
- yetihehe 3y agoIf using that flip-flop, whole processor will be closer to 100ghz (typically there are multiple transistors which need to stabilise before you have a result of computation). But probably those superconductors could enable even faster transistors and maybe we could get 1THz processors.
- aitchnyu 3y agoLike IDE ribbon cables gave way to sleek SATA cables, we could have another parallel to serial transition from silicon multicore crap to superconducting single core.
- jacquesm 3y agoNo, you'll get superconducting multicore because all that will happen is that the bloat will expand to consume the new cycle budget. This has already happened many times.
- dtx1 3y agoI'm looking forward to Chromium version 4201337 needing 20% of my 1Thz CPU to render simple text.
- regularfry 3y agoWe'll get superconducting multicore because as a first step everyone will just map what they already know to the new space. Which is fun.
- MayeulC 3y agoI suspect you'd still need very low temperatures to make this work, even with a high temperature superconductor: low temperatures reduce thermal noise, which may be an issue at such time scales (unless you pump a lot of energy per bit, which means high voltages (limits scaling) or current/capacitances (also limits scaling).
- rezonant 3y agoWell you can apparently make diodes with them: https://scitechdaily.com/mit-develops-superconducting-device-to-radically-cut-energy-use-in-computing/?expand_article=1 https://scitechdaily.com/mit-develops-superconducting-device...
- u320 3y agoTransmission losses aren't really a big problem for the grid. Cost, geopolitics, and resiliency matters more. I don't expect superconductors to change much here.
- bscphil 3y agoThis seemed likely the case to me as well. I'd be interesting in hearing any counterarguments (or better still, actual studies) concerning distribution efficiencies.
- almostnormal 3y agoThe price for electricity is mainly driven by cost of distribution not production. Reducing losses on distribution would further increase the cost of the expensive part (upgrade to SC net) and decrease the cost of the cheap part. Storage would help, the cheapest production cannot run continuously.
- nine_k 3y agoTransmission losses are a concern because they require building huge, very high-voltage structures. Superconductors could make transmission lines much more compact, sturdy, weatherproof, and less vulnerable to sabotage. You could run a thick armored cable instead of a set of open-air wires on tall towers.
- jacquesm 3y agoYou would no longer need high voltages at all. We only use those to cut down on those very transmission losses because we're trying to reduce the amount of current. So you can drop all of the step-up-step-down stuff. On the down side: a short is then practically unlimited in current.
- nine_k 3y agoExactly the point. We will still need some step-downs, because current density is limited. But we will need fewer of them, and they won't need the monumental cooling they have now.
- flangola7 3y agoEvery doctor's office could have an MRI/fMRI machine, brain scans become routine.