Julian Kelly

Timeline

1. 2014–2015

   Surface-code experiments at UCSB

   During his PhD at UC Santa Barbara, he helps co-develop the planar Xmon transmon architecture. He is a coauthor of an April 2014 Nature letter that reports a two-qubit gate fidelity of up to 99.4 per cent, at the surface code threshold. In March 2015, he is first author of a Nature paper in which a nine-qubit array cuts the failure rate of retrieving a state by a factor of 8.5 through repetitive error detection.

2. 2014–2018

   Designing the Bristlecone processor

   He joins Google in 2014 as a founding member of its quantum hardware effort and creates the Optimus automated calibration framework. In March 2018, Google presents Bristlecone, a 72-qubit gate-based superconducting processor, at the American Physical Society meeting in Los Angeles; he designs the chip and writes the announcement for Google's research blog.

3. 2019

   Quantum supremacy on Sycamore

   He is among the authors of the October 2019 Nature paper in which the Sycamore processor creates quantum states on 53 qubits and samples a circuit a million times in about 200 seconds, a task Google's benchmarks put at approximately 10,000 years for a classical supercomputer. The experiment is Milestone 1 in his team's progression of hardware and error-correction milestones.

4. July 2021

   Cyclic quantum error correction

   In a July 2021 Nature paper, his team demonstrates exponential error suppression in cyclic error correction. Increasing the number of qubits from 5 to 21 reduces the logical error per round more than 100-fold, with error suppression remaining stable over 50 rounds.

5. February 2023

   Scaling a logical qubit

   In a February 2023 Nature paper, his team achieves the first error suppression by scaling a surface-code logical qubit, Milestone 2. Across 25 cycles, a distance-5 surface-code logical qubit outperforms distance-3 logical qubits on average, with a logical error per cycle of 2.914 per cent compared to 3.028 per cent.

6. December 2024

   Below threshold on Willow

   In December 2024, he is a coauthor of a Nature paper that presents two below-threshold surface code memories on the Willow processor, a 101-qubit distance-7 code and a distance-5 code integrated with a real-time decoder. The logical error rate falls by a factor of 2.14 each time the code distance grows by 2.

7. 2025

   A five-year quantum outlook

   As director of hardware at Google Quantum AI, he tells CNBC in March 2025 that Google is about five years from a practical application that only a quantum computer can solve, with early uses in simulating cutting-edge physics. In April 2025, he describes quantum computers generating new data on which AI models could train, with AlphaFold as his example.

8. As of October 2026

   Senior director of hardware

   He is Senior Director of Hardware at Google Quantum AI, where he leads the development, integration and operation of Google's superconducting quantum computers on the path to large-scale, fault-tolerant systems. Following 2026 work on fault-tolerant magic state cultivation on Willow, he helps architect the Quantum AI Hardware Roadmap, and his team's current focus is long-lived logical qubits, Milestone 3.