AI Error Control Meets IBM’s Next Quantum Leap

Quantum computers face a stubborn enemy: noise. LTM is taking aim at that problem with an AI-powered error mitigation capability now available through IBM’s Quantum Functions Catalog, giving eligible users a new way to improve quantum results without adding quantum processing time or qubits.

The capability, called AI-QEM, works inside Qiskit workflows and uses classical post-processing to learn the error patterns of individual quantum devices. Could this approach help researchers run useful experiments with fewer resources while the industry works toward full error correction?

AI-QEM Brings Error Mitigation Into Qiskit

LTM has integrated AI-QEM into IBM’s Quantum Functions Catalog, an ecosystem of partner-built quantum software services designed to simplify workflows for researchers and enterprises. As a circuit function, AI-QEM lets users with eligible IBM Quantum access plans apply the technology directly within their Qiskit workflows.

IBM’s listing says the function examines circuit, backend and observable characteristics to learn correction factors. It then uses classical inference to mitigate noisy quantum expectation values, targeting better results without additional quantum processing.

That approach separates the correction work from the quantum processor itself. AI models learn device-specific error patterns and estimate corrections through classical post-processing, so AI-QEM does not require extra QPU time, repeated circuit executions or extra qubits.

Those requirements can add cost and complexity to conventional error mitigation techniques. LTM says AI-QEM improves result accuracy without those added resources, creating a direct path for users who want to test algorithms and applications on available quantum systems.

LTM tested the capability on IBM Quantum computers across circuits with different sizes and levels of complexity, including experiments involving more than 100 qubits. The company did not disclose the accuracy improvements achieved or provide comparisons with other error mitigation methods.

LTM also said it became the first global systems integrator, or GSI, to publish a function in IBM’s catalog. Other providers in the catalog offer related tools, including Q-CTRL’s AI-driven error suppression, Qedma’s QESEM error mitigation function and Algorithmiq’s tensor-network-based approach.

A New Tool Arrives as IBM Reworks Its Quantum Roadmap

The AI-QEM launch comes as IBM’s Quantum Developer’s Conference provides an update on the company’s hardware roadmap. IBM aims to make error-corrected quantum computers before the decade is out, while running a parallel public-facing track for users who want to test algorithms on quantum systems today.

One track focuses on the technical development needed for error-corrected hardware. The other delivers high-performance general-use processors through revisions to IBM’s Nighthawk family, which began with 120 hardware qubits and an expectation of about 5,000 gate operations before errors become too likely to trust the output.

IBM’s previous plan expected a 2027 processor to reach 10,000 gate operations while keeping 120 hardware qubits. It also called for linking three Nighthawk processors in 2026 to create a 360-qubit machine.

The latest roadmap changes that path. IBM plans to add a version with over 500 programmable qubits to its next-generation Nighthawk processors and reduce the intended number of linked systems from nine to two. The total number of qubits planned for next year will remain the same.

Nighthawk r2, released in September, is offered as a single chip. IBM says hardware-based qubit resets give it a major performance boost compared with r1, with up to a 19-fold acceleration for some operations.

IBM says the chip contains 458 physical qubits, but only 120 are programmable. The remaining hardware supports resets that return programmable qubits to the ground state. Oliver Dial, a VP of Quantum Systems, said, “That capability lets our users run circuits on this chip faster than was ever before possible.”

The plan for Nighthawk r3 in 2027 targets 520 programmable qubits, while retaining the performance technology used in the current revision and adding other hardware improvements. IBM’s public-facing roadmap continues through 2028 with revisions to the current Nighthawk generation.

Fewer Connections, Bigger Systems

IBM’s existing roadmap promised users access to a single system with 1080 qubits. The updated plan targets a 1040-qubit machine built by linking two chips, instead of reaching the goal through nine linked systems.

IBM has already reported progress connecting processors. Dial said the company had been “showing four L-couplers joining two Nighthawk R1 processors in the same refrigerator, reaching state transfer fidelities of up to 99.4 percent, very close to our long-term target of 99.9 percent.”

Jerry Chow, IBM’s CTO of Quantum-Centric Supercomputing, said the speed of those connections is comparable to the qubit interconnections on the chip itself. That connection technology will matter as IBM works toward larger systems and error-corrected computing.

IBM’s long-term plan calls for roughly 300 hardware qubits to host a block of 12 logical, error-corrected qubits. Reaching 100+ logical qubits would require a larger chip than Nighthawk r2 and five or more connected chips, with IBM needing to achieve both by the end of next year to stay on track.

Against that backdrop, AI-QEM offers a practical bridge between today’s noisy quantum processors and tomorrow’s error-corrected systems. LTM continues to work on quantum optimization for a leading global insurer, hybrid quantum machine learning for critical infrastructure operations and quantum-safe security programs for financial services and insurance organizations.

IBM’s hardware roadmap is still moving toward larger and more capable systems, but tools such as AI-QEM could shape what users accomplish before full error correction arrives. The next phase of quantum computing may depend not only on adding qubits, but on learning how to get more trustworthy answers from the qubits already available.

Stay connected via Google News

Follow us for the latest AI News