A small Japanese quantum startup just scored a world-first achievement on one of the planet's most powerful supercomputers. The secret behind this David-vs-Goliath story is a management philosophy simple enough to fit on a sticky note: let your researchers do what they do best.

Who Is Quemix?

Quemix (pronounced "queue-mix") is a quantum computing startup headquartered in Nihonbashi, Tokyo. Part of the TerraSky Group, the company specializes in developing algorithms and software that maximize the performance of quantum computers.

The company is led by CEO Yuichiro Matsushita, a computational science specialist with a Ph.D. in Engineering from the University of Tokyo. His career path reads like a tour of Japan's top research institutions: he worked as a researcher at Germany's Max Planck Institute, served as an assistant professor at the University of Tokyo, and held positions at what is now the Institute of Science Tokyo. He currently holds concurrent appointments as an associate professor at the University of Tokyo and project chief at the National Institutes for Quantum Science and Technology (QST).

What sets Quemix apart from many quantum startups is its laser focus on FTQC (Fault-Tolerant Quantum Computer) algorithms. While many competitors rush to build applications for today's noisy, intermediate-scale quantum machines (known as NISQ devices), Quemix has bet on developing foundational technology for the fully error-corrected quantum computers of the future.

The Fugaku Breakthrough

Before founding Quemix, Matsushita made his mark on one of Japan's most prestigious national projects. In 2015, while still an assistant professor at the University of Tokyo, he was invited to develop applications for Fugaku, Japan's flagship supercomputer jointly developed by RIKEN and Fujitsu.

Fugaku is no ordinary machine. It topped the TOP500 global supercomputer rankings for four consecutive terms starting in 2020, making it one of the most powerful computing systems ever built. Matsushita's assignment was to create applications for the successor to the "K computer" (Fugaku's predecessor) that could only run on this next-generation hardware.

His ambitious goal: apply quantum chemistry calculations, traditionally used in the chemistry domain, to physics problems, specifically to create high-precision models of the electronic states of solid materials. If successful, it would be a world first.

The challenge was immense. High-precision calculations predicting electron behavior required enormous computing time, even on the K computer. But the next-generation machine promised such a massive leap in performance that Matsushita was confident the barrier could be broken.

And break it he did. His team successfully developed large-scale, high-precision quantum chemistry calculation applications on Fugaku, achieving a world-first result that would later form the technological backbone of Quemix.

The Management Secret: Let Researchers Be Free

What makes this story particularly interesting is that the breakthrough wasn't just about technology, it was about management philosophy.

Matsushita has spoken candidly about the pressure of leading a team of people he considered more talented than himself. His solution? Give researchers maximum autonomy and minimum bureaucracy.

This approach runs counter to traditional Japanese organizational culture, which tends toward hierarchical top-down management. National projects, in particular, often come with mountains of paperwork, mandatory meetings, and administrative obligations. Matsushita deliberately minimized these management burdens, believing that researchers perform best when they can immerse themselves freely in their work.

This "researcher-first" philosophy carries over to Quemix today. The company's team is composed of elite researchers from institutions including the University of Tokyo, the Institute of Science Tokyo, RIKEN, and the National Institute of Advanced Industrial Science and Technology (AIST). Among them is a recipient of the ACM Gordon Bell Prize, one of computing's highest honors.

Partnerships with Honda, Asahi Kasei, and Sumitomo Rubber

Quemix's technical prowess has translated into concrete results through collaborations with some of Japan's most prominent corporations.

In May 2025, Quemix and Honda's R&D subsidiary jointly developed a groundbreaking new technique for reading quantum states. One of the fundamental challenges in quantum computing is that attempting to read a quantum state causes "wavefunction collapse", the quantum information is destroyed in the process. Quemix and Honda solved this by developing a method that "scans" the characteristic features of a quantum state without directly reading (and thus destroying) it. Using this technique, they achieved the world's first successful X-ray Absorption Fine Structure (XAFS) calculation on an actual quantum computer, a milestone for materials development.

With chemical manufacturer Asahi Kasei, Quemix conducted verification experiments for new materials development. They ran Quemix's proprietary algorithm PITE® (Probabilistic Imaginary-Time Evolution) on Quantinuum's ion-trap quantum computer, marking the first time an FTQC-designed algorithm was successfully executed on actual quantum hardware.

The partnership with Sumitomo Rubber Industries focuses on dramatically accelerating fluid dynamics simulations using quantum computing, essential for tire development and other applications where understanding fluid behavior is critical.

The Global Trend: Quantum-Classical Hybrid Computing

Quemix's work sits at the forefront of a major global trend: quantum-classical hybrid computing.

Today's quantum computers are still in their early stages and cannot handle all types of calculations independently. The emerging solution is to combine traditional supercomputers with quantum machines, assigning each type of computer the tasks it handles best.

In the Honda collaboration, for example, the XAFS calculations were performed using a hybrid setup combining Quantinuum's quantum computer with the University of Tokyo's Institute for Solid State Physics supercomputer. Problems were divided based on which computing platform could handle each component most effectively.

RIKEN announced in 2024 that IBM's Quantum System Two would be installed in the same building as Fugaku, signaling Japan's commitment to quantum-HPC integrated computing. Fujitsu has also developed its own quantum simulator leveraging Fugaku's technology, achieving world-leading performance in 36-qubit quantum circuit simulation.

Japan's Quantum Startup Ecosystem

Quemix's success reflects the broader maturation of Japan's quantum startup ecosystem.

QunaSys is another notable player, focusing on algorithm research and software development for practical quantum computing applications. The company was involved in developing Qulacs, an open-source quantum simulator originally created at Osaka University, which has been adopted by Fujitsu for its quantum simulation platform.

Fixstars Amplify provides a cloud-based optimization service using quantum annealing and GPU technology. By December 2025, the platform had surpassed 1,000 registered organizations and 100 million total computation runs, handling 150,000 to 250,000 optimization problems daily. The company has become a de facto standard for combinatorial optimization in Japan, with applications ranging from manufacturing scheduling to logistics planning. In 2024, Fixstars Amplify also participated in RIKEN, the University of Tokyo, and NTT's joint development of the world's first universal optical quantum computer.

The Japanese government has positioned quantum technology as a national strategic priority, promoting industry-academia-government collaboration through Q-STAR (Quantum STrategy for ARchitecturing the future). The University of Tokyo's SQAI (Sustainable Quantum AI) research hub, funded by the Japan Science and Technology Agency, counts Quemix among its participating companies.

How active is the quantum computing startup scene in your country? Does the research culture encourage giving scientists freedom to pursue their own ideas? Share your thoughts!

References