Superconducting memory exists in prototype form: Josephson junction circuits, flux qubits, single-flux-quantum logic. All of these engineer memory from external circuit design — the superconductor carries the current, but the memory lives in the circuit topology. The material is the wire; the architecture stores the bit.
Wu et al. found that uranium ditelluride — a triplet superconductor candidate — has intrinsic memory. A DC current pulse switches the material between two states: high critical current and low critical current. The state persists after the pulse ends. Another pulse switches it back. No magnetic layers, no Josephson junctions, no external circuitry. The material itself is the memory element.
The mechanism involves two competing vortex species within the superconductor. A current pulse pushes the system into a metastable configuration where one vortex species dominates, creating stronger pinning and a higher critical current. The system stays there because the configuration is locally stable — it would need to climb over an energy barrier to relax. The other state has weaker pinning and lower critical current. Switching between states is controlled by pulse strength and duration.
This is memory as a property of quantum order, not of architecture. The superconducting condensate itself has two stable configurations distinguished by their vortex populations. The bit isn't stored in a circuit — it's stored in the topology of the vortex lattice. The material doesn't just conduct; it remembers which kind of conducting it's doing.
The distinction matters for superconducting computing. Engineering memory from circuits requires space, interfaces, and energy for switching. Intrinsic memory requires only the material. If the superconductor IS the memory, the component count for a superconducting computer drops by the number of external memory elements.
The wire isn't just a wire. It knows which way it was pushed.