IBM and partners claim quantum advantage with built-in accuracy checks

Why it matters

Showing that a quantum computer beats classical ones on a task whose answers can be checked is a key test on the road to useful machines. IBM opened the results to public challenge, and one was quickly matched classically.

Sources (5)

• IBM Newsroom IBM and The University of Chicago Demonstrate Quantum Advantage, Establishing Trusted Quantum Computation on Logical Circuits
  Primary sourceJul 30, 2026

  “IBM (NYSE: IBM) and researchers from the University of Chicago today announced a demonstration in quantum computing that achieves the fundamental criteria for quantum advantage: performing computations beyond the reach of leading classical simulation methods while providing trust that the computation returned accurate results.”

• IBM Quantum Blog Quantum advantage through trusted quantum computation
  Primary sourceJul 30, 2026

  “Three papers demonstrate quantum advantage with built-in validation, enabling trustworthy quantum computations beyond exact classical verification.”

• IEEE Spectrum IBM Claims Quantum Advantage With New Validation Techniques
  Independent coverageJul 31, 2026

  “In a series of collaborations with researchers at the University of Chicago and quantum software startups Algorithmiq and Qedma, IBM claims to have now demonstrated three separate instances of quantum advantage alongside methods that validate the results are correct despite not being classically simulatable.”

• arXiv (Manabe, Gu and Pan) Classical Simulation and Design Frontiers for IBM's Doped Clifford Sampling Experiment
  Challenges the claimAug 13, 2026

  “Using 32 nodes, with eight NVIDIA H100 GPUs per node, we completed all 2051 amplitude batches corresponding to IBM's published output bitstrings in 37.3 minutes.”

• arXiv (Martiel et al., IBM and collaborators) Sampling hard circuits with verifiably high fidelity (version 3)
  Challenges the claimSep 2, 2026

  “Subsequent work exploited this geometry and computed the exact ideal probabilities of all 2,051 reported output samples using a tensor-network simulation requiring 256 H100 GPUs for 37.3 minutes, or approximately 159 H100-GPU-hours”