Experimental Progress in Scaling Surface Code and Autonomous Cali | Alexis Morvan, Google Quantum AI

The transition of quantum error correction (QEC) from theoretical promise to experimental reality has reached a pivotal turning point. Recently, milestones on 105-qubit processors have demonstrated that logical error rates can be suppressed exponentially, exceeding the "break-even" point where a logical qubit outperforms its physical constituents [Nature 638, 2025]. This progress extends to the realization of distance-preserving logic in the color code, achieving suppressed error rates on a superconducting processor [Nature 645, 2025]. We will discuss the challenges of scaling these results on real-world hardware. We first show how exploiting the "time-dynamics" of QEC circuits allows us to break free from strict gate scheduling and circumvent hardware constraints like reduced connectivity. We explore how these dynamic circuits utilize non-traditional gate sets such as iSWAP and periodically swap the roles of data and measure qubits to mitigate leakage [Nature Physics, 2025]. We then show how error syndromes can be used as a learning signal for Reinforcement Learning (RL) agents to autonomously steer analog control parameters [arXiv:2025]. This allows the system to stabilize against environmental drift in real-time, establishing a path toward self-stabilizing, fault-tolerant quantum processors. Quantum Series & Program: https://ivado.ca/evenements/seminaire... Subscribe for more high-level quantum and AI content : ivado.ca #QuantumComputing #GoogleQuantumAI #ErrorCorrection #SurfaceCode #QuantumAI