IBM plans to link two Nighthawk chips into a 1,040-qubit system

IBM says it plans to connect two Nighthawk r3 processors into a single 1,040-qubit system in 2027. The company says each new processor is designed to run 10,000 quantum operations on 520 programmable qubits.

The new chip is larger than the one shipping now. IBM says Nighthawk r3 will keep the square-lattice layout and fast operations of the current chip, and will add 10 chip-to-chip links, called l-couplers. It says the processor will have 2,034 physical qubits in all, of which 520 are programmable. The company says the current Nighthawk, introduced in September, delivers up to 100,000 samples a second and supports accurate computations with more than 7,500 operations on 120 qubits.

IBM has already tested the links it plans to use. The company says it connected Nighthawk r1 processors this year with four l-couplers and reached 99.3 percent state-transfer fidelity, against a target of 99.9 percent. It also says it connected two cryogenic modules into one system for the first time, reached operating temperature in under five days, and cooled the system below 15 millikelvin.

Partners have already run the current chip at scale. IBM says E.ON, Algorithmiq, and Q-CTRL have executed 34 billion circuits on Nighthawk r2, with reported speedups of up to 19 times versus earlier IBM quantum processors. E.ON reported a 15-times speedup in a quantum machine-learning job for grid optimization, measured against earlier work on IBM's Heron processor. IBM says an Oak Ridge National Laboratory materials experiment ran 12 times faster on r2 than on Heron, and an Algorithmiq materials simulation ran 8 times faster than the original Heron experiment.

The platform now pairs quantum jobs with classical computers. IBM says users can connect a quantum workflow to a laptop, a supercomputer, or classical resources on the IBM Quantum Platform. It says an IBM and RIKEN workflow ran about 5 times faster and performed 100 times more computations than the original study. Cleveland Clinic and RIKEN processed 608 million samples, which IBM says was 33 times faster than without the workflow automation.

"Useful quantum computing hinges on whether results can be delivered with accuracy and speed, both are fundamental," said Jay Gambetta, director of IBM Research and an IBM fellow.