Yonsei University will upgrade its on-premises IBM Quantum System One from a 127-qubit Eagle processor to IBM's Nighthawk QPU, with installation scheduled for November 2026 at the university's Songdo Campus in Incheon, South Korea. The upgrade makes Yonsei the second facility in the world to host a Nighthawk-based system, after IBM's own site in Miami.
Fewer qubits, different wiring
Nighthawk has 120 qubits, slightly fewer than the Eagle system it's replacing, and that's easy to misread as a downgrade. The real change is connectivity. Eagle uses a heavy-hex topology, where each qubit connects to an average of about 2.5 neighbors. Nighthawk uses a square lattice with 218 tunable couplers, giving each qubit 4 direct neighbor connections. Our ranking of the top quantum computing companies already covers Nighthawk as IBM's current fielded processor. This upgrade is a real-world example of what that connectivity change is for.
More connections per qubit means fewer SWAP gates are needed to move a two-qubit operation between qubits that aren't directly wired together, since SWAP gates exist specifically to work around limited connectivity by physically relocating quantum information through the chip. IBM and Yonsei describe the result as roughly a 40% computational capacity gain from reduced SWAP overhead and lower circuit depth, not from a larger qubit count. That's a useful data point for anyone assuming "more qubits" is the only axis that matters when comparing QPU generations.
Why a university runs its own on-premises system at all
Most researchers reach IBM hardware through the cloud, including through IBM Quantum's free tier. An on-premises System One is a different, much larger commitment: dedicated hardware physically located at the institution rather than shared cloud queue time. Yonsei's system supports a specific research portfolio that benefits from that dedicated access: a collaboration with RIKEN's Fugaku supercomputer studying treatment mechanisms for Leigh syndrome, a research node with Cambridge's Milner Therapeutics Institute on AI-driven drug discovery, and a new software platform called Q-Bridge aimed at enterprise integration. Owning the hardware outright, rather than queuing for cloud time, matters more when a research program is running sustained, iterative workloads rather than occasional exploratory circuits.
What's confirmed and what isn't
Confirmed: the upgrade itself, the November 2026 installation date, Nighthawk's published specifications, and Yonsei's status as the second Nighthawk site globally. Not detailed in the announcement: what specific quantum circuits or benchmarks Yonsei will run first on the new hardware, or whether the claimed 40% capacity gain has been measured on Yonsei's own workloads versus cited from IBM's general Nighthawk specifications.
What to watch next
Whether Yonsei publishes results from its Leigh syndrome or drug-discovery research using the new hardware, which would be a concrete test of whether Nighthawk's connectivity advantage translates into faster or more accurate results on a real biomedical workload, rather than staying a chip-architecture specification on paper.