PALO ALTO, Calif., May 26, 2026 -- D-Wave Quantum Inc., (“D-Wave” or the “Company”), the only dual-platform quantum computing company providing both annealing and gate-model systems, software and services, today issued the following response to recent claims that newly published classical simulation work has “overturned” D-Wave’s demonstration of quantum computational supremacy in quantum simulation.

The claim that D-Wave’s achievement has been overturned is inaccurate and not supported by the scientific record.

D-Wave welcomes advances in classical algorithms, including recent tensor-network work from researchers at the Flatiron Institute and collaborators. Scientific progress in quantum computing depends on rigorous comparison between quantum and classical methods. However, the recently published work does not reproduce the full scope of D-Wave’s peer-reviewed Science result, nor does it solve the hardest problem instances and measurements reported in that work. The only major change in the Flatiron work that has now been published in Science is the addition of large diamond lattice data. This supports the claim that they are using BP-TNS to extract a Kibble-Zurek exponent in a 3D system, but the same criticisms apply as a year ago: they don’t compute the same observables, nor all the geometries, nor the largest size geometries, nor all the couplings computed by D-Wave and its collaborators.

In the peer-reviewed Science paper, “Beyond-classical computation in quantum simulation,” D-Wave researchers and collaborators demonstrated beyond-classical computation in the quantum simulation of nonequilibrium magnetic spin dynamics using D-Wave annealing quantum computers. The work studied square, cubic, diamond and biclique topologies and showed that D-Wave quantum processing units produced samples consistent with quantum theory at scales where direct classical computation becomes impractical. For the largest problems studied, the paper reported that matching the D-Wave quantum processor’s simulation quality with matrix product state methods would require nearly a million years on the Frontier supercomputer, with memory and energy requirements exceeding practical limits.

“D-Wave’s demonstration of beyond-classical computation continues to hold up under careful scientific scrutiny,” said Dr. Alan Baratz, CEO of D-Wave. “We welcome advances in classical methods, including recent work from the Flatiron Institute, but claims that these advances overturn D-Wave’s result are inaccurate. A claim that strong requires reproducing the full scope of our demonstration, including the hardest cases and the full set of measurements. That has not happened.”

The Flatiron Institute’s BP-TNS algorithm is a meaningful contribution to the classical state of the art, and it is effective in some regimes. But it is not effective across the full range of problem classes studied in D-Wave’s Science paper. In a March 2025 response, D-Wave researchers and collaborators noted that the Tindall et al. work did not attempt the most complex lattice geometry, did not reproduce the largest simulations in 3D lattices, did not simulate the low-precision ensembles in which correlations grow fastest, and did not produce the full-state and fourth-order observables reported in D-Wave’s Science paper.

In the arXiv paper “Evaluating Classical Simulations with a Quantum Processor”, D-Wave researchers and collaborators further evaluated the limits of classical tensor-network simulations using a quantum processor as a reference. That work showed that BP-TNS fails for strongly coupled three-dimensional spin glasses on cubic and diamond lattices, and that loop-corrected BP-TNS is ineffective for higher-dimensional biclique problems. These are important quantum simulation regimes included in D-Wave’s original demonstration, not peripheral examples.

“The BP-TNS algorithm is effective in some regimes and ineffective in others,” said Dr. Trevor Lanting, chief development officer at D-Wave. “Our analysis showed that it fails for strongly coupled three-dimensional spin glasses on cubic and diamond lattices, and that loop-corrected BP-TNS is ineffective for higher-dimensional biclique problems. These are important cases from our Science paper, and they remain beyond the reach of current classical methods.”

D-Wave encourages continued work by the Flatiron Institute and others to advance classical simulation methods. Such progress is valuable and can help sharpen the boundary between classical and quantum capabilities. But scientific communication should distinguish between a meaningful advance in classical simulation and overturning a quantum supremacy result. The former is supported by the evidence. The latter is not.

“We should all hold ourselves to a higher standard when communicating scientific results,” Baratz added.