Quantum computers may have achieved a “quantum advantage” by performing tasks beyond the capabilities of the most powerful supercomputers. Experts estimate that replicating the calculations made by classical machines could take an incomprehensible amount of time, equivalent to trillions of times the age of the universe. What implications does this development hold for creating truly functional quantum computers?

The latest record holder in this domain is a quantum computer known as Jiuzhang 4.0, which utilizes particles of light, or photons, to execute computations. Chao-Yang Lu and his team at the University of Science and Technology of China utilized it for Gauss Boson Sampling (GBS). This involves measuring a sample of photons after they navigate a sophisticated arrangement of mirrors and beamsplitters connected to computers.

In earlier attempts to perform this task, the number of utilized photons never exceeded 300. In contrast, Jiuzhang employed 3,090 particles, representing a tenfold improvement in computational strength. Lu and his colleagues estimate that contemporary algorithms on the most powerful supercomputers would require a staggering 10⁴² years to replicate what Jiuzhang accomplished in just 25.6 microseconds.

“These results are certainly an impressive technical achievement,” said Jonathan Lavoy of the Canadian quantum computing startup Xanadu, which previously held the GBS record with 219 photons. Chris Langer of Quantinuum noted that while their systems have previously demonstrated quantum advantages in various forms of quantum computing, this advancement is significant. “It’s essential to establish that quantum systems cannot be simulated by classical means,” he asserts.

However, Jiuzhang’s previous versions have been used successfully in conducting GBS with a considerable number of photons, but each time a classical computer eventually replicated the results, sometimes within an hour.

Bill Fefferman from the University of Chicago mentions that he is working on a classical algorithm to achieve victory over quantum systems but notes that significant challenges exist for photonic devices. Many photons are lost during the operation of quantum computers, and the systems tend to be noisy. “Currently, we’ve managed to reduce noise while simultaneously ramping up experimentation. However, our algorithm has yet to find a breakthrough,” states Fefferman.

Lu points out that addressing photon loss is the primary hurdle his team faced in the latest experiment. Nevertheless, Jiuzhang remains free of noise, suggesting potential for new classical simulation strategies to take on the title of superiority.

“In my view, they haven’t achieved full power yet, but they are certainly in a position to prove that such classical strategies may not be feasible,” remarks Gelmarenema from the University of Twente, Netherlands.

This presents a “noble cycle” where the competition between classical algorithms and quantum devices enables a better understanding of the blurry lines separating classical and quantum realms, according to Fefferman. From a fundamental science view, this signifies a triumph for all; however, whether quantum computing can be effectively harnessed in more powerful machines remains an open question.

Langer describes GBS as an “entry-level benchmark” that highlights the distinction between quantum and classical computers, but the results do not necessarily indicate the practical utility of such machines. From a rigorous mathematical perspective, evaluating GBS as concrete evidence of quantum advantage is challenging, as Nicolas Quesada at Polytechnic Montreal, Canada, points out. Identifying a clear pathway to developing a superior machine using GBS remains elusive.

This is primarily because Jiuzhang’s hardware is highly specialized, and programming quantum computers for a variety of calculations remains unachieved. “It might demonstrate computational advantages for narrow tasks, but it fundamentally lacks the key components for practical quantum calculations that involve fault tolerance,” explains Lavoy. Fault tolerance refers to a quantum computer’s ability to recognize and correct its own errors—an essential capability that has yet to be realized in contemporary quantum systems.

Meanwhile, Lu and his team advocate for various applications stemming from Jiuzhang’s remarkable capabilities in GBS. This approach could revolutionize computations tied to image recognition, chemistry, and specific mathematical challenges associated with machine learning. Fabio Sciarrino from the University of Sapienza in Rome suggests that though this quantum computing paradigm is still nascent, its realization could lead to groundbreaking changes.

Specifically, advancements like Jiuzhang’s device could pave the way for the creation of extraordinary light-based quantum computers, asserts Sciarrino. These computers would be programmed in entirely innovative manners and excel in machine learning-related tasks.