Our engineering and research teams publish regularly on topics spanning quantum error correction, hardware advances, hybrid algorithm design, and platform capabilities. Subscribe via RSS at https://blog.kandz.co/feed.xml or follow us for the latest updates.
Crossing the Threshold: Real-Time Quantum Error Correction on k&z 72-Qubit Processors
We report the first demonstration of real-time surface code error correction running continuously on k&z production hardware. By implementing a distance-5 surface code across 49 data qubits and 24 syndrome qubits on our Zurich-based 72-qubit processor, we achieved a logical error rate of 1.2 x 10^-4 per round — a 47x improvement over the physical qubit error rate. This milestone brings us closer to fault-tolerant quantum computation and has immediate implications for customers running variational algorithms, where error-corrected expectation values dramatically improve convergence reliability.
Hybrid Quantum-Classical Optimization: Benchmarking QAOA on Max-Cut Problems at Scale
We benchmarked the Quantum Approximate Optimization Algorithm (QAOA) on Max-Cut instances ranging from 20 to 70 nodes using k&z hybrid infrastructure. Our results show that at circuit depth p=3, the quantum solver matches the best-known classical heuristic (Goemans-Williamson) for graphs up to 50 nodes and exceeds it for irregular, high-connectivity graphs at 60+ nodes. Critically, our hybrid runtime reduced the classical-quantum data transfer overhead by 82% compared to a naive implementation, enabling 12x more optimizer iterations per hour. This post details the benchmark methodology, reproducible code, and implications for combinatorial optimization workloads.
Introducing the k&z SDK v2.0: Framework-Agnostic Circuit Submission
Today we release k&z SDK v2.0, a major update that introduces native support for Qiskit, Cirq, and PennyLane circuits without requiring manual transpilation. The SDK now accepts circuits in any of these frameworks and handles conversion to our hardware-native gate set automatically. Version 2.0 also includes a new asynchronous job manager that supports batch submission of up to 10,000 circuits in a single API call, automatic retry on transient failures, and streaming result callbacks. Upgrade with pip install kandz-sdk --upgrade and check the migration guide in our documentation.
Cryogenic Engineering at Scale: How We Cool 9 QPU Blocks to 15 Millikelvin
Running nine dilution refrigerators simultaneously presents unique engineering challenges that go far beyond simply ordering more hardware. In this post, our cryogenic engineering team shares the design decisions behind k&z's cooling infrastructure: custom pulse-tube pre-cooling stages that reduce helium-3 consumption by 35%, vibration isolation mounts that decouple mechanical noise from qubit coherence, and a centralized monitoring system that detects thermal anomalies within 200 milliseconds. We also discuss our approach to planned maintenance — how we take individual cryostats offline for servicing without interrupting customer workloads through automatic QPU rerouting.
Post-Quantum Cryptography on Quantum Hardware: Benchmarking Lattice-Based Schemes
As NIST finalizes post-quantum cryptographic standards, understanding how these algorithms perform — and how they might be attacked — on real quantum hardware is essential. We used k&z QPUs to implement Grover-based key search attacks against CRYSTALS-Kyber and CRYSTALS-Dilithium, measuring the actual qubit resources, circuit depths, and error rates required. Our findings confirm that current 72-qubit devices are far from threatening these schemes, but they provide the first empirical calibration of the quantum resources needed, replacing theoretical estimates with measured values. This data is critical for government agencies and enterprises planning their cryptographic migration timelines.
Sovereign Quantum Infrastructure: Designing for Data Residency and Regulatory Compliance
National governments and defense organizations increasingly require that quantum computing infrastructure — including the physical QPU hardware, control electronics, and all associated data — remain within their sovereign territory. In this post, we detail k&z's approach to sovereign deployments: air-gapped control planes, on-premises QPU installation with remote calibration support, and compliance with data residency regulations across jurisdictions including the EU, Switzerland, Singapore, and the United States. We share lessons learned from deploying our first sovereign cluster for a European defense research agency and outline the architectural patterns that make sovereignty compatible with cloud-scale operations.
Stay Updated
New posts are published every two to three weeks. To stay informed:
- RSS Feed:
https://blog.kandz.co/feed.xml - Email Digest: Subscribe in your Console under Settings → Notifications → Blog Updates
- Community Forum: Discuss posts and ask questions at
community.kandz.co