[Tech Products]+[Breakthrough]: QRAM Commercial Quantum Memory Chip Addresses Long-Standing Quantum Computing Storage Bottleneck

Article Body

An underestimated bottleneck in quantum computing lies in storage. Traditional quantum computer qubit states are extremely unstable, typically maintaining coherence times only at the microsecond to millisecond level, meaning intermediate results generated during computation must be used or lost within extremely short timeframes. The QRAM chip jointly released by PsiQuantum and TSMC is attempting to solve this problem.

The QRAM-1 chip released in May 2030 uses a photonic qubit approach, leveraging TSMC's advanced packaging technology to integrate 1,024 addressable optical quantum storage units on a single 12-square-millimeter chip. Each unit can store the state information of one photon qubit, with a coherence time reaching an impressive 1.2 seconds—three orders of magnitude longer than the best superconducting qubits.

PsiQuantum CEO Jeremy O'Brien said: "QRAM gives quantum computing true random access memory for the first time. This means quantum algorithms can freely read and write intermediate results during computation, just like classical programs, without needing to keep all information within the qubit coherence time."

QRAM-1's technical core is a "delay line storage" architecture. Photon qubits are injected into a fiber ring resonator composed of low-loss silicon nitride waveguides. By precisely controlling the ring cavity length and coupling coefficients, photons can circulate and be stored in the ring for up to several seconds. Read and write operations are performed through tunable directional couplers with latencies below 50 nanoseconds.

TSMC's head of advanced technology business development stated that QRAM-1 uses TSMC's N3E process to manufacture the photonic chip, with hybrid bonding technology vertically integrating the photonic layer with the classical control circuit layer. The single chip package measures 25mm x 25mm and can be directly integrated into PsiQuantum's existing photonic quantum computing platform.

In performance testing, QRAM-1 achieved a qubit storage fidelity of 99.3% and a read-write fidelity of 98.7%. These metrics meet the requirements of most quantum error correction codes. PsiQuantum expects that quantum computers equipped with QRAM-1 will see performance improvements of more than 10x when executing quantum programs requiring extensive intermediate storage, such as Shor's algorithm.

However, QRAM-1 still requires operation near absolute zero, with the cost and size of the cooling system limiting its application scenarios. PsiQuantum plans to release a room-temperature compatible version in 2031, but the specific timeline depends on progress in new quantum materials research.

Lukin, director of Harvard's Center for Quantum Science and Engineering, commented: "QRAM is a critical piece of the puzzle on the road to practical quantum computing. With it, quantum computers truly gain the ability to execute complex algorithms, rather than merely demonstrating quantum supremacy."