IBM has announced Quantum Nighthawk r3, its next superconducting processor design, MarkTechPost reports. The plan is a chip that can run 10,000 quantum operations accurately across 520 programmable qubits, and, in 2027, to link two r3 chips into a single 1,040-qubit system. That’s a roadmap, not a box you can use today.
What’s actually in the cloud now is Nighthawk r2, as ibm_phoenix on the IBM Quantum Platform: 120 programmable qubits and 218 couplers. IBM says r2 already runs more than 100,000 circuits per second, about 25 times the earlier Heron generation, with Heron-class gate fidelity, and up to 19 times faster than earlier IBM processors on some partner workloads. The r3 design calls for 520 programmable qubits, 2,034 physical qubits and 10 l-couplers. Chip-to-chip links are currently averaging 99% state-transfer fidelity, short of IBM’s 99.9% target.
The important distinction in the coverage is easy to miss in a headline: r3 is a design announcement. The 1,040-qubit system is a 2027 plan. The speed numbers you can cash today belong to r2.
My take
I’ll take the throughput claims on the live r2 chip seriously, and I’ll treat r3 and the 1,040-qubit pairing as a schedule until there’s hardware. Roadmaps in this field slip. A faster chip you can actually queue a job on is the news that matters this week.
God Bless, and Happy hacking!
