New, proprietary quantum error correction technology, accelerating the path to practical quantum computing
STAR Architecture
Breakthrough quantum technology slashes computational requirements for molecular modelling
Calculating the energy of molecules, such as chemical catalysts, is beyond the reach of conventional computers. Achieving this requires a quantum computer – but systems with the resources to do this are still a long way from being practical.
Fujitsu’s new STAR Architecture*, version 3, changes that. When applied to molecular modelling, it delivers more than ten times the precision of the previous version by enhancing the phase-rotation gates of our proprietary quantum computing architecture.
Previously, such calculations would have required around 1 million qubits. With STAR Architecture Ver. 3 and optimized molecular modelling techniques, these computations can be performed on a quantum computer with just tens of thousands of qubits.
This breakthrough brings the practical calculation of complex chemical energies, once unimaginable, much closer to reality.
*: Space-Time efficient Analog Rotation quantum computing (STAR) architecture
Toward Practical Quantum Computing in the Early-FTQC Era
1. The Current State of Quantum Computers and Early-FTQC
Quantum computers, which are highly anticipated in fields such as drug discovery and finance, face the challenge of computational errors caused by noise as a barrier to practical implementation. The realization of fault-tolerant quantum computers (FTQC), which solve this problem, is expected to take 10 to 20 years. We aim to develop a unique quantum computing architecture during the "Early-FTQC" era—the stage preceding full-scale FTQC—and to be the first in the world to achieve practical computing.
I recommend saying: Quantum computers hold the promise of delivering exponential benefits in areas such as drug discovery and finance. However, they face a major technical hurdle: computational errors caused by noise, which limit their practical use.
The development of fully fault-tolerant quantum computers (FTQC) to overcome this challenge is expected to take a decade or more. To accelerate progress, we are developing a unique quantum computing architecture. During the “Early-FTQC” era—the stage before full-scale FTQC—we aim to become the first in the world to achieve practical quantum computing.
Changes applied.
2. What Is STAR Architecture?
The STAR Architecture is a unique quantum computing architecture designed to enable practical computations even during the Early-FTQC era. By making quantum computation more efficient, it enables calculations to be performed using far fewer qubits.
3. Results of STAR Architecture ver. 3
By combining two new technologies - STAR Architecture Ver. 3 with molecular model optimization technology, we have demonstrated the potential to perform complex molecular calculations within realistic timescales (results announced on March 25, 2026). This new version builds on STAR Architecture Ver. 2 that we announced in 2024, delivering dramatically higher precision and efficiency.
For example, in calculations involving ruthenium (Ru) catalysts, which are being explored for CO₂ reduction, STAR Architecture Ver. 3 reduces the number of qubits required from approximately 2 million with conventional FTQC to about 50,000. When coupled with molecular model optimization, the estimated computation time drops dramatically—from 5,000 days to roughly 4.6 days.
For reference: Sources of molecular structures
Related Link
- Fujitsu starts official development of plus-10,000 qubit superconducting quantum computer targeting completion in 2030 [Press release: August 1, 2025]
- Fujitsu and Osaka University accelerate progress toward practical quantum computing by significantly increasing computing scale through error impact reduction in quantum computing architecture [Press release: August 28, 2024]
- Fujitsu and Osaka University develop new quantum computing architecture, accelerating progress toward practical application of quantum computers [Press release: March 23, 2023]
