Google Quantum AI builds on Transmon qubits, the same family as IBM Quantum, arranged in a 2D grid with each qubit tunably coupled to its neighbors. Its Sycamore processor produced the first widely publicized claim of quantum computational advantage in 2019, running a random circuit sampling task. That claim was later narrowed as classical simulation techniques improved.
More recently, Google's Willow chip showed that adding physical qubits to a surface-code logical qubit reduces the logical error rate. This was a long-sought milestone: error correction actually getting better as it scales, not just staying even.