What if you could test software for a computer that doesn't fully exist yet? That is what quantum circuit simulation does: it uses today's supercomputers to mimic how a quantum computer would behave. The catch is that every additional qubit doubles the computing power required, and for quantum chemistry problems, 40 qubits had been the practical ceiling. A Japanese research team just broke through it.

1,024 GPUs and a New World Record

On March 12, 2026, a research group led by Professor Wataru Mizukami of Osaka University's Center for Quantum Information and Quantum Biology (QIQB), with QIQB technical staff Shoma Hiraoka and Sho Nishida and Yusuke Teranishi of Fixstars Corporation, announced they had achieved the world's largest quantum chemistry circuit simulation.

The team harnessed 1,024 NVIDIA H100 GPUs on System H of ABCI-Q, the quantum-classical hybrid computing platform operated by Japan's National Institute of Advanced Industrial Science and Technology (AIST). That is roughly half of the 2,020 H100 GPUs installed in that system. Running their custom-built simulator "chemqulacs-gpu," they set two world records:

  • Largest problem size: Simulated the water molecule (H₂O) with 42 spin orbitals (using qubit reduction techniques)
  • Largest circuit size: Computed a 41-qubit circuit for the iron-sulfur cluster molecule (Fe₂S₂), a pure circuit-scale benchmark

Both results surpass the previous limit of 40 qubits for state-vector quantum chemistry simulations, representing a milestone that no other research group has achieved.

Why Simulating Quantum Computers Matters

When people think about the quantum computing race, the first names that come to mind are typically Google, IBM, and Microsoft, all competing to build quantum hardware. But hardware alone is not enough. The algorithms that will actually run on quantum computers need to be developed and validated in advance, and this is where simulation becomes indispensable.

Think of it this way: imagine building a new kind of airplane, but with no wind tunnel to test your wing designs before the first flight. Quantum circuit simulation serves as that wind tunnel, a virtual test environment where researchers can verify whether their quantum algorithms work correctly and efficiently before actual quantum hardware becomes widely available.

The fundamental challenge is scale. Simulating a quantum system on a classical computer requires memory and processing power that grow exponentially with each qubit. At 40 qubits, you need roughly 18 terabytes of memory, a staggering amount that had marked the practical ceiling for quantum chemistry simulations using quantum phase estimation.

The IQPE Algorithm: A Smarter Approach

The key to this breakthrough lies in the team's choice of algorithm. Rather than implementing standard Quantum Phase Estimation (QPE), which requires many auxiliary qubits and produces deep, complex circuits, the team focused on Iterative QPE (IQPE).

QPE is a foundational quantum algorithm used across many applications, particularly valued in quantum chemistry for its ability to precisely determine molecular energy levels. However, conventional QPE demands substantial quantum resources.

IQPE is an elegant variation that needs only a single auxiliary qubit and produces shallower circuits. This has two important implications: it makes simulation more tractable on today's classical computers, and it aligns well with "Early-FTQC" (early fault-tolerant quantum computers), the transitional machines that will offer partial error correction before fully fault-tolerant systems arrive.

In other words, this simulation validated at world-record scale the very algorithm that researchers expect to run on the first generation of practical quantum computers.

48 Hours on the Clock: A High-Stakes Sprint

The computation was carried out under AIST's "ABCI-Q Grand Challenge" program, which grants research groups exclusive access to 256 compute nodes (1,024 GPUs) for a maximum of 48 hours.

Professor Wataru Mizukami of QIQB reflected on the experience: the team faced repeated unexpected technical issues while trying to keep 1,024 GPUs working in unison. Within that 48-hour window, a small team led by two young researchers, Yusuke Teranishi of Fixstars and Shoma Hiraoka of Osaka University, persevered, with support from the ABCI-Q operations staff, to reach the record.

The findings were presented at NVIDIA GTC 2026 in San Jose, California, in March 2026.

Where This Fits in the Global Quantum Race

The quantum computing landscape is increasingly competitive, with major players pursuing different strategies.

IBM aims to demonstrate verified quantum advantage by the end of 2026 and to deliver Starling, a fault-tolerant machine running 100 million gates across 200 logical qubits, by 2029. Its Nighthawk processor, announced in November 2025, pairs 120 qubits with 218 tunable couplers and runs circuits about 30% more complex than its Heron predecessor.

Google continues to push toward error-corrected quantum computation, with long-term plans for a million-qubit system. Microsoft is betting on topological qubits that could offer inherently lower error rates.

Japan takes a different but complementary approach. While its hardware investment is smaller than that of the U.S. or China, Japan has built world-class infrastructure for quantum algorithm development. ABCI-Q is one of the world's largest quantum-computing-focused supercomputers, equipped with over 2,000 NVIDIA H100 GPUs and integrated with several types of quantum processors: superconducting (Fujitsu), neutral atom (QuEra), and photonic (OptQC).

The Osaka University breakthrough demonstrates Japan's strength in what might be called "quantum software readiness." When fault-tolerant quantum computers finally arrive, the countries and institutions that have already developed and validated their algorithms will have a decisive head start. Japan's combination of academic excellence at QIQB, GPU optimization expertise from Fixstars, and national-scale computing infrastructure at AIST positions it well in this often-overlooked dimension of the quantum race.

From Drug Discovery to Climate Solutions

What practical impact could quantum chemistry calculations have? The most anticipated application is drug discovery, precisely simulating how candidate molecules behave at the quantum level to dramatically accelerate and reduce the cost of pharmaceutical development.

New materials for combating climate change represent another frontier. Quantum chemistry could help researchers predict the performance of catalysts, battery materials, and other substances at the molecular level, streamlining the search for optimal compositions.

The Fe₂S₂ (iron-sulfur cluster) molecule used in this simulation is particularly noteworthy, it plays a crucial role in biological electron transfer and is of fundamental importance in biochemistry and enzyme research.

By pushing past the 40-qubit barrier, this work expands the range of molecules that can be studied through simulation, enabling the development and refinement of quantum algorithms for increasingly realistic and complex chemical systems.

The "real hardware vs. simulation" narrative often frames these as competing approaches, but in reality they are two sides of the same coin. As quantum hardware improves, the need for validated software grows, and the more prepared the software is, the faster hardware can be put to practical use. This result from Osaka University and Fixstars accelerates that virtuous cycle.

How is quantum computing research progressing in your country? Is it driven by government programs, universities, or private companies? We'd love to hear your perspective!

References