Cloud-led Future · Intelligence-driven Hotline +86 21 6228 0217 中文 | EN

Quantum Computers Get a High-Speed Memory: Zhejiang Team Achieves First Real-Hardware QRAM

Overview

Teams from Zhejiang University’s schools of physics, computer science and software, together with the Ningbo International Innovation Center, have achieved the world’s first real-hardware implementation of quantum random-access memory (QRAM) on the self-developed superconducting quantum chip “Tianmu-2.” The results were published in the international journal Nature Physics and covered by People’s Daily under the headline “Quantum Computers Get a High-Speed Memory.”

Classical computers store information in bits; quantum computers exploit superposition of qubits for multi-path amplitude parallelism, theoretically outperforming classical machines on certain problems. Yet when quantum algorithms process large classical datasets, they need an efficient data channel — otherwise computing is “as fast as lightning while reading data is as slow as a snail.” The QRAM concept was proposed in 2008 to let quantum machines point precisely to and retrieve multiple data items simultaneously via superposition, but for over a decade it remained theoretical: real chips had circuits too long and noise too high, so quantum states decohered before data could be read.

The team ran 4-bit and 8-bit QRAM prototypes on Tianmu-2, measuring accuracies of 80.9% and 60.4% respectively, with circuit depth over 30% lower than unoptimized schemes — confirming the architecture’s local noise resilience.

Background & Interpretation

1. Technical background and evolution

QRAM is a key device for moving quantum algorithms from theory to practice. Regarded as a quantum computer’s high-speed memory, it provides an efficient interface between classical data and quantum states, analogous to the role memory plays for classical compute. Since its 2008 proposal, the lack of hardware scale and coherence time kept it untested on real machines. Improvements in superconducting qubit count, fidelity and control have now made hardware validation possible. The result continues the field’s shift from demonstration experiments toward verifiable prototypes.

2. Core drivers and mechanisms

The team began work in 2022. The central challenge was running an ideal tree structure on a two-dimensional grid chip. They reconfigured the routing architecture and node-switching logic, turning detours into straight channels. Because any node error in a tree distorts the result and is hard to trace, routing qubits double as monitors: a level transition or logic error triggers a signal that lets the system identify and discard invalid queries. To handle the many swap gates needed between root and leaf qubits — and the resulting noise — the team used quantum teleportation to build high-speed logical tunnels between distant qubits, bypassing physical wiring constraints. Together, these methods improved both accuracy and circuit depth.

3. Market structure and value-chain impact

The work was completed on Tianmu-2, designed, fabricated and controlled in-house, so core capabilities from chip design to fabrication are domestically held. At the industry level, scalable QRAM could ease the data-access bottleneck and support scenarios requiring massive classical data — financial big-data analytics, AI training, drug discovery and materials simulation. Demand for upstream links such as dilution refrigerators, microwave coaxial cables and micro/nano fabrication should also become clearer as prototypes advance.

Implications & Outlook

Near term, the 4-bit and 8-bit prototypes remain proof-of-concept, with accuracy and scale far from practical use. Medium-term variables include whether accuracy holds as qubit count grows, whether circuit-depth optimizations are reusable, whether local noise resilience replicates on larger chips, and whether peers independently verify the results. The industry remains cautious about overall quantum maturity, with no consensus on a timeline for general-purpose quantum computing.

Takeaways for Industry Participants

Quantum supply-chain players should track the transition from theoretical architecture to hardware validation, using fidelity, circuit depth and scalable design as hard metrics rather than qubit count alone. Potential users in finance, pharma and materials can map high-value scenarios with genuine demand for quantum data interfaces. Research institutions should note that in-house chip fabrication and control capability is becoming a precondition for validating quantum algorithms.

This column compiles industry information and shares technical perspectives; it does not constitute investment advice.