Skip to content
Back to the Guavy Wire
Stocks

IBM to build 1,040-qubit quantum system by merging two processors

Instruments
IBM
Share

IBM is set to merge two of its next-generation quantum processors into a single system with 1,040 qubits in 2027. The new system will leverage the company’s upcoming Nighthawk r3 architecture, with each processor supporting 520 programmable qubits and executing up to 10,000 quantum operations. This move underscores IBM’s ambition to build larger quantum systems capable of tackling increasingly complex computational tasks.

The announcement follows significant performance gains from IBM’s existing Nighthawk r2 processor, introduced in September 2026. The system delivers up to 100,000 samples per second and supports computations involving over 7,500 quantum operations across 120 qubits. Industry partners, including German energy company E.ON, Algorithmiq, and Q-CTRL, have reported speed improvements of up to 19 times. E.ON achieved a 15-fold speedup in quantum machine learning for electricity grid optimization, while Oak Ridge National Laboratory saw a 12-fold acceleration in magnetic materials simulations.

Nighthawk r3 will introduce new hardware capabilities, including 2,034 physical qubits and 10 chip-to-chip connections called l-couplers. These components enable separate quantum processors to exchange quantum states, potentially expanding computing capacity without relying on a single large chip. IBM has already demonstrated this technology with two Nighthawk r1 processors, achieving 99.3% quantum state transfer fidelity, with a target of 99.9%.

IBM is also enhancing its Quantum Platform to integrate quantum processors with classical high-performance computing resources. This integration helps handle error correction, noise analysis, and post-quantum measurement processing. Early results show improved research efficiency, with IBM and RIKEN scientists completing a quantum diagonalization workflow five times faster and processing 100 times more computations than their original study. Cleveland Clinic and RIKEN used the platform to process 608 million samples for modeling biological structures, completing the task 33 times faster.

More on Stocks