IBM has unveiled two new quantum processing units (QPUs) designed to accelerate the path toward practical quantum computing. These developments include the 120-qubit Nighthawk processor, which offers a 30% performance increase over previous models, and the 112-qubit Loon processor, engineered as a blueprint for fully fault-tolerant quantum computation.
Scaling Performance with Nighthawk
The Nighthawk processor enhances qubit connectivity through improved tunable couplers, enabling each of its 120 qubits to connect with four neighbors. This architecture supports quantum calculations requiring up to 5,000 two-qubit gates—fundamental operations in quantum computing. IBM aims to scale Nighthawk to 7,500 and 10,000 gates by 2026 and 2027, respectively, with a long-term goal of 15,000 gates on a 1,000-qubit system by 2028.
The Pursuit of Fault Tolerance with Loon
While qubit count is important, the real challenge is mitigating errors. The Loon processor focuses on this by integrating all hardware components necessary for fault-tolerant quantum computing. This means the processor is designed to self-detect and correct errors in real time—a crucial step towards reliable quantum computation.
Why Error Correction Matters
Quantum computers are inherently unstable, with qubits prone to errors. The Loon processor addresses this by incorporating quantum error correction (QEC) technologies. QEC is not about making bigger processors; it’s about making more reliable processors. IBM’s 1,000-qubit Condor chip, while large, was less promising than its 127-qubit Eagle counterpart due to the latter’s lower error rate.
New Technologies in the Processors
IBM’s CTO, Oliver Dial, highlighted several new features in the processors: six-way qubit connections (allowing each qubit to connect with up to six neighbors), increased routing layers, longer couplers, and “reset gadgets” to return qubits to their ground state. These technologies are being tested together for the first time on the 112-qubit Loon processor.
Modular Design and the Kookaburra Processor
IBM is also developing the Kookaburra processor, expected in 2026. This will be the first modular-designed QPU, combining logic operations with memory storage. The modular design allows for more scalable and reliable quantum systems.
Tracking Quantum Advantage
IBM has established a quantum advantage tracker to measure when quantum computers can solve problems beyond the capabilities of classical supercomputers. The tracker includes three initial challenges: observable estimations, variational problems, and classically verifiable problems.
Advancements in Wafer Fabrication
IBM is also transitioning to 300mm (12-inch) wafer fabrication. This new format halves processor build time and increases chip complexity by a factor of ten. The process involves slicing silicon cylinders into thin disks, designing circuits with software, etching circuits, depositing metals, treating wafers, and layering/connecting chips.
Conclusion: IBM’s latest quantum processors, Nighthawk and Loon, represent significant steps toward practical quantum computing. The focus on both performance scaling and error correction, combined with advancements in wafer fabrication, positions IBM as a leader in the race to achieve quantum advantage and build fault-tolerant quantum computers by 2029
