tunable coupler

Two qubits sitting next to each other on a chip are a bit like two tuning forks on the same table: strike one and the other starts to hum. That whispering between neighbors is exactly what you need when you want them to interact during a two-qubit gate, and exactly what you do not want the rest of the time, when each qubit is supposed to sit quietly and hold its state. A tunable coupler is an extra circuit element placed between two qubits that acts like a volume knob on their conversation, so you can turn the interaction up to do a gate and turn it down to near silence the instant the gate is done.

In a superconducting chip the coupler is usually its own little circuit, often a small SQUID loop or an extra transmon, threaded by a magnetic flux you control with a nearby current line. Changing that flux shifts the coupler's frequency, and at one special setting the unwanted always-on interaction between the two qubits cancels out, leaving an effective coupling of essentially zero. Nudge the flux away from that point and the coupling reappears, strong and fast, so a two-qubit gate that once took hundreds of nanoseconds can finish in tens. Because the off state is genuinely off, idle qubits stop slowly poisoning each other through the residual ZZ interaction that fixed couplers leave behind.

The catch is that you have traded one hard problem for a more manageable one, not for none. Every coupler needs its own control line and flux bias, which adds wiring, heat, and another knob that has to be calibrated and kept stable; the flux line itself can leak noise into the qubits. Tunable couplers are now standard on several leading superconducting processors precisely because suppressing idle crosstalk matters so much for fidelity, but they are not magic: they help most where qubit frequencies are well spread out, and crowded, noisy chips remain the real bottleneck to scaling.