IBM’s quantum advance: company machines outperform classical computers in tests
IBM researchers announced on August 3, 2026 that they had obtained a quantum advantage in three different experiments. The achievement means that the company’s quantum computers were able to outperform classical algorithms in specific tasks. The universe, by its nature, is quantum, which makes these computers promising for modeling complex phenomena that classical machines can only approximate.
Traditionally, computational chemistry methods such as Density Functional Theory (DFT) need to convert complex quantum behaviors into equations that classical hardware can process. This conversion, as systems become more complex, requires trade-offs in accuracy, speed or cost. The quantum advantage eliminates the need for these adaptations.
For Netanel Lindner, co-founder and chief technology officer at Qedma, “chemistry is one of the most important long-term applications of quantum computing.”
The expectation is that quantum computers will not only simulate systems with greater accuracy than classical ones, but also explore scenarios that are beyond the capabilities of traditional computing. Dominik Hangleiter, a computer scientist at the Swiss Federal Institute of Technology (ETH) in Zurich who was not involved in the studies, cited the simulation of quantum materials and the breaking of certain cryptographic systems as examples.
Despite their potential, current quantum machines are still susceptible to noise, errors and loss of information, and are often outperformed by classical computers in modeling. Thus, demonstrating the superiority of quantum computers remains an essential goal for the field.
Papers detailing the three new demonstrations, performed on IBM’s Quantum Heron R3 quantum processors, have been published as preprints and are awaiting peer review. In a conference call with journalists on July 28, IBM researchers and their collaborators presented the results, highlighting how quantum computers, supported by classical processors, can be superior in specific problems.
First demonstration: creating patterns with qubits
Quantum computers operate with qubits — their basic units — which are controlled by quantum gates, analogous to the logical bit gates in conventional computers. The set of available ports determines the capabilities of each quantum computer.
Clifford gates are widely used in quantum computers. They are easy to simulate, but have restricted functionality, and models using only these ports can be reproduced efficiently by classical machines.
The inclusion of special gates next to the Clifford gates allows a quantum computer to become universal, capable of performing any type of computation.
Simulating a universal quantum computer in a classical system is considered impractical, as it would take so long that the result would lose its purpose. Some calculations could take millions of years, surpassing even the age of the universe for the most advanced supercomputers. It is precisely in this scenario that quantum computers are expected to demonstrate their value.
On the same topic: Helios quantum computer achieves unprecedented precision and accelerates technology evolution
IBM, in partnership with researchers at the University of Chicago, used the universal quantum computer to generate samples composed of binary patterns (zeros and ones). The team reported that no classical computer could replicate the same distribution of patterns in a reasonable time.
IBM scientist Ali Javadi-Abhari said the team “mathematically proved that no classical computer can generate this pattern.”
Hangleiter clarifies that this mathematical proof indicates that, at a minimum, a set of samples from this quantum algorithm cannot be simulated classically. He muses: “That doesn’t mean all of these samples are difficult to simulate.”
It is not yet possible to completely rule out the classical reproduction of samples obtained by researchers, in case some unknown method emerges to replicate the distribution of results. Hangleiter stated, “All you need is someone with a genius idea.”
Given this uncertainty about classical reproduction, how can researchers confirm that the quantum computer is operating correctly?
To validate the results, teams from IBM and the University of Chicago carried out tests with circuits of different levels of complexity.
In the model developed, the researchers added special gates, called T gates, to the Clifford gates. However, the inclusion of T ports physically introduces errors: the greater the number of T ports applied, the more powerful the computer becomes, but it also increases the accumulated noise, which can compromise the accuracy of the output.
With just five T-gates, the classical simulation was still viable, and the results obtained on quantum and classical computers were compatible.
The increase in the number of T gates made the classical simulation considerably more difficult and reduced the accuracy. When the total T gates reached 468, no known classical algorithm was able to simulate the model.
Javadi-Abhari’s group altered the strategy: they operated the quantum computer without the T gates so that a classical machine could predict the outcome.
By comparing the predictions with the quantum results, the researchers estimated a circuit fidelity of 32%. They then calculated the impact of T-gates on reducing fidelity. Because the ports were positioned to minimize noise, the team found that the quantum computer’s fidelity could decrease, at most, to 28.4%.
Javadi-Abhari explained that, “if this result were a photograph, fidelity would be more related to the similarity of the entire image to the original than to the fraction of pixels that are exactly correct.” He added: “Even if the pixels are slightly different colors or in the wrong positions, they can contribute to higher fidelity.”
For Hangleiter, “this number is really very high”, being a hundred times higher than previous pioneering works. He considers this demonstration unique and suggests the need for more similar research.
Second demonstration: the simulation of a quantum magnet
Researchers at Qedma, a quantum computing startup, employed an IBM quantum computer to simulate a quantum magnet. This material has magnetic behavior governed by quantum mechanics, which classical physics cannot fully explain.
Each atom has a particular magnetic moment, comparable to a small compass needle. In a conventional magnet, magnetism arises from the alignment of these moments in several atoms in the same direction.
In the context of a quantum magnet, the magnetic moment of an atom can manifest in multiple directions simultaneously. This phenomenon is known as quantum superposition, analogous to the idea of Schrödinger’s cat, which is alive and dead at the same time. Furthermore, the magnetic moments of nearby atoms can intertwine, interconnecting their behaviors.
More on this story: Expert again criticizes Microsoft’s advances in topological quantum computing
These entangled and overlapping magnetic moments, with their quantum potential, can act together in unusual ways, resulting in exotic states of matter. Understanding this behavior could help explain high-temperature superconductivity, with the potential to revolutionize energy transmission and electronics.
At Qedma, the team led by Netanel Lindner simulated a quantum magnet being subjected to regular pulses, like “kicks”.
A “kick” or pulse represents a sudden shock to the state of the magnet, similar to hitting it with a beam of light. In the quantum computer, each “kick” was created by applying a sequence of logic gates to all the qubits.
When a system is repeatedly disturbed, it does not reach thermal equilibrium instantly. Instead, it remains in an intermediate phase, called the pre-thermal state, where it maintains its structure before eventually relaxing and reaching equilibrium. This can be illustrated by shaking a snow globe: for a certain time, the snow is suspended in a “pre-thermal state” before settling.
See Also
Electrical discharge knocks man off roof in Jaguaruana after contact with power lines
Global auction raises US$11.9 million and Brazil’s historic currency reaches record value
Former fighter Prichard Colón dies at age 33 after decade in a vegetative state
It is in this pre-thermal period that the magnet’s most intriguing quantum effects manifest themselves. Magnetization during this phase is particularly useful for understanding its behavior. These same effects also make it difficult for classical computers to accurately simulate the magnet.
Lindner’s team simulated the impact on quantum magnets of different dimensions and compared the quantum results with those obtained in classical simulations for the same sizes.
For 28-atom magnets, classical computer data was consistent with both physical intuition and quantum computer data, demonstrating the expected functioning of the quantum machine.
However, for magnets with 51 and 74 atoms, three classic models failed: some ran out of memory, others lost accuracy with increasing complexity, or became excessively slow. The quantum processor, in turn, continued operating, presenting magnetization values that were in line with theoretical predictions.
The limitations of classical computers complicate the verification of quantum results for larger magnets, as there are no exact magnetization values for comparison. The researchers cannot compare the simulation with a real material, as the magnets were a theoretical system created specifically for the IBM processor. Hangleiter comments: “It’s like choosing a problem that is very easy to simulate on a quantum computer rather than something you would find in nature.”
Still, Lindner’s group conducted their simulation on multiple quantum processors, each with its own noise patterns and error mitigation systems, including one that operated on different hardware, to check for potential failures. The results were coincident.
Obtaining the same magnetization behavior on such diverse machines reduces the likelihood that the results were the result of a specific hardware failure.
Third demonstration: evaluation of a realistic system
In the third demonstration, researchers from quantum software company Algorithmiq simulated a material on an IBM quantum computer. Like the Qedma magnet, this material exists only in theory.
To make the system more similar to real matter, Sergei Filippov and his team intentionally introduced clutter and analyzed how information moved through it.
Sabrina Maniscalco, co-founder and CEO of Algorithmiq, during the press conference, said that “catalysts, batteries and all the materials that will power the next generation of clean energy are complex, disordered and irregular, and this complexity is what makes them difficult to simulate.” She added: “We have created a model of the real disorder in matter.”
In a way analogous to inferring the transparency or opacity of a material by projecting light at one end and observing it pass through, Filippov’s team perturbed a set of qubits — each representing an atom — and evaluated the effect on another set of qubits at the opposite end of the computer, after the perturbation propagated.
Due to the disorder present in the system, information was disseminated irregularly: in some areas, the passage was quick, while in others, it was delayed.
In classical physics, it is possible to measure a system before and after a disturbance or process to identify changes. However, in quantum mechanics, the measurement itself modifies the system. Checking a property, such as the position of an electron or the state of a qubit, destroys the essential quantum state, making it impossible to measure the same system twice and compare the results directly.
To overcome this issue, the researchers used a creative solution to measure how the flow of information changed the material: inverting the simulation. First, they applied a sequence of operations to simulate the natural evolution of the material over time. Then, they introduced a small disturbance and, finally, performed all the operations in reverse order, reversing the flow of time in the material.
Without the presence of a perturbation, inverting the simulation would cause the qubits to return to their initial state. However, the disturbance interrupts this perfect retrograde. As the system operated in reverse, only the effects associated with the disturbance remained.
Learn more: Historic milestone: IBM’s first sub-1 nanometer chip opens new era in computing and AI
Measuring these changes results in a value known as the Loschmidt echo. Just as echoes in a cave can provide data about its shape, this Loschmidt echo reveals crucial information about the material.
Filippov’s team ran the simulations on two different IBM quantum computers and employed two classical algorithms on a classical processor.
With different degrees of disorder, the two quantum computers generated similar echo signals, while the classical simulations showed divergences between themselves and in relation to the quantum results. This suggests that quantum simulations have maintained accuracy where classical methods failed. The agreement between the quantum processors, each with distinct noise and error profiles, also reduces the likelihood that the signal was caused by particularities of one of the machines.
However, both quantum processors use the same architecture, and there is no external comparison parameter to validate the results. The material was developed specifically for the IBM processor and does not correspond to any real substance that can be produced in a laboratory.
Therefore, verification remains internal: IBM quantum computers validate each other, while classical methods diverge. Once the simulated material has been built for the hardware, it is challenging to determine whether the corresponding results reflect real physics or just fit the machines they were designed for.
Hangleiter said, “I don’t know enough of the details to know if I can trust it.” He states that the researchers validate the result by characterizing the noise in the processor very precisely, which gives them good control over what happens and allows them to estimate the real calculation error. “That might be a reason to trust.”
Quantum advantage: significance for the field of chemistry
According to Hangleiter, it is not certain whether these demonstrations achieved the quantum advantage. He stated that “it is unlikely that we will ever be able to definitively prove the quantum advantage.”
In all three demonstrations, a recurring point is the difficulty in verifying the accuracy of the results of a quantum computer. Hangleiter said “the verification is better than we’ve seen in other demos so far, and they’re really trying hard.” However, there is no way to prove with one hundred percent certainty that the researchers are correct.
Dominik Hangleiter, a computer scientist at the Swiss Federal Institute of Technology (ETH) in Zurich, said that “it is unlikely that we will ever be able to definitively prove the quantum advantage.”
One way to validate the results of a quantum computer would be to simulate a real experiment and compare the results directly with observations of nature.
Jay Gambetta, director of research and fellow at IBM, stated at the press conference that “this is our goal and everyone’s goal here.” Gambetta also stated that the purpose of publishing these three articles was to strengthen confidence in quantum computing.
Although these demonstrations are extremely specific applications and do not yet have immediate commercial viability, Hangleiter believes that these new results increase confidence and reinforce the potential of quantum computing.
Lindner sees the recent demonstrations as “a move toward practical quantum computing that people can use today for scientific applications.”
Lindner suggests that today’s theoretical physicists should be using quantum computers. Over time, he hopes this practice will extend to chemistry and other scientific areas. He reiterated that “chemistry is one of the most important long-term applications of quantum computing.”
However, Lindner highlights that “the path to practical advantages in chemistry depends heavily on further advances in quantum hardware.” In the meantime, he encourages chemists “to get involved with quantum computing, strengthening the community’s understanding of what questions can be addressed efficiently by quantum computers.” Classical computational chemistry is not being replaced. He concluded that “this result is a proof of principle for an additional scientific instrument.”
