A room-temperature superconducting interconnect just cleared independent replication, and three chipmakers are already redesigning motherboards around it.
For a decade, "room-temperature superconductor" has been the physics world's boy who cried wolf. Every few years a paper claims one, headlines run wild, and the result quietly fails to replicate within a year.
This time, three independent labs — in Grenoble, Daejeon, and Ann Arbor — have reproduced the core result behind the Kessendra interconnect, a copper-lattice material doped with a rare-earth compound that carries current with zero resistance at 22 degrees Celsius, under ambient pressure.
Why interconnects, not chips
The material isn't fast enough, yet, to replace silicon transistors. Its value shows up somewhere less glamorous: the wiring between chips. Modern server racks lose a meaningful fraction of their power budget to resistive heating in the copper traces connecting processors to memory. A zero-resistance interconnect doesn't make a chip smarter, but it makes a data center dramatically cheaper to cool.
"Everyone wanted this to be about faster CPUs," said materials scientist Ingrid Solheim-Okafor, who led the Ann Arbor replication. "The actual near-term win is a rack that draws 30% less power for the same throughput. That's boring and it's also enormous."
The skeptics haven't gone quiet
Not everyone is convinced the manufacturing process will survive contact with a real fab line. The doping process currently requires a controlled cooldown period measured in days, not the milliseconds a production line needs. Semiconductor analyst Devrim Kowalczyk-Otieno puts the odds of a commercial product before 2029 at "maybe one in three, and that's if nothing else goes wrong — something always goes wrong."
Three chipmakers have licensed early access to the replication data regardless. None will confirm which product lines they're targeting.