Scalable Photonic Quantum Computer Aurora
Design Proposed
Schematic of the architecture. Precursors to GKP states are generated from multimode Gaussian states produced probabilistically with GBS chips by heralding particular PNR patterns. Many precursor states are sent to each refinery chip (via optical fibre delays represented by blue lines), which use a combination of multiplexing and breeding implemented in a binary tree of beamsplitters (represented by the wedge-like shapes labelled ‘B’), and squeezing, to create a pair of high-quality GKP sensor states. For the fault-tolerant architecture, the binary tree is augmented with homodyne detectors. A Bell pair is then generated by applying a 50/50 beamsplitter (black solid lines). The spatial routing and temporal delays of the modes in each pair are set by the desired cluster-state graph, such that each graph macronode corresponds to an individual QPU chip and these chips share entangled pairs if they are neighbours on the cluster-state graph. In the fault-tolerant architecture, multiple pairs are created per edge, but only one pair is selected per graph edge by a multiplexer at the beginning of the QPU. Then, each QPU interferes with the selected pairs using static beamsplitters and these modes are measured using homodyne detection. Loss paths are shown with red dashed lines while classical feedforward is represented with black dashed lines. Middle: table showing the internal structure of each submodule in the fault-tolerant architecture and in the Aurora experiment. Bottom: legend for optical component diagrams.
1. Overall Architecture
- The system consists of 35 photonic chips arranged as modular, rack-mounted units.
- These modules are interconnected via fiber-optic links.
- The architecture is designed for fault-tolerant and scalable quantum computing.
System Architecture – 6 Core Subsystems:
- Laser System – Provides coherent pump and local oscillator beams, plus reference beams for phase stabilization.
- Sources Array – Generates squeezed light and two-mode Gaussian states as building blocks for quantum computation.
- PNR Detection System – Uses photon-number-resolving (PNR) detectors to herald useful non-Gaussian quantum states.
- Refinery Array – A binary tree-based multiplexing system that selects and entangles the best-quality quantum states.
- QPU (Quantum Processing Unit) Array – Forms the spatiotemporal cluster state and performs homodyne measurements to execute quantum operations.
- Fiber Buffer Network – Provides phase- and polarization-stabilized delay lines, ensuring synchronized quantum states between different modules.
The entire system, apart from the cryogenic detection array, fits into 4 standard 19-inch server racks. Here we summarize the main features of these subsystems, as well as our method for verifying the multimode entanglement present in our cluster-state benchmark experiment;
3. Networking and Scalability
- The system is interconnected using fiber-optic delay lines to synchronize quantum states.
- Components are deployed in rack-mounted modules, making scaling straightforward.
- The architecture uses cluster-state quantum computing with error correction.
4. Loss and Fault-Tolerance Considerations
- Photon loss is the biggest challenge for scalability.
- The architecture tolerates optical losses up to 1% in key paths.
- Optimized multiplexing and low-loss waveguide materials are essential.
5. Key Innovations
- Room-temperature operation (except for PNR detectors).
- Use of Gottesman-Kitaev-Preskill (GKP) states for encoding.
- Multiplexed Bell-pair generation for improved entanglement rates.
Implementation
The implementation can be found in this PAPER- Aghaee Rad, H., Ainsworth, T., Alexander, R.N. et al. Scaling and networking a modular photonic quantum computer. Nature (2025). https://doi.org/10.1038/s41586-024-08406-9
Sources & citation
Aghaee Rad, H., Ainsworth, T., Alexander, R.N. et al. Scaling and networking a modular photonic quantum computer. Nature (2025). https://doi.org/10.1038/s41586-024-08406-9