Quantum computing firm IonQ has sold a device that it claims will be able to outperform any classical supercomputer or other quantum computer in the world 鈥� but the machine has yet to be built and experts are sceptical about its performance. So who currently claims the quantum crown?
Part of the problem is that comparing quantum computers made using a variety of technologies by different companies is far from an exact science. One measure is the number of qubits, or quantum bits, a computer has, with more qubits bringing theoretically higher performance.
In truth, this is a crude measure of power that doesn鈥檛 take into account how fast the machine can carry out operations, how interconnected the qubits are and how reliable their calculations can be.
Advertisement
Because of this, IonQ has created a metric called algorithmic qubits (AQ), which the firm claims is the most holistic measure yet, taking into account both the number of qubits and how interconnected they are. In May, it claimed that its machine Aria had reached a 听鈥� making it 鈥渢he most powerful quantum computer in the industry鈥�.
Since then, IonQ听says it has created a computer with an AQ of 29, and is now developing a machine with an AQ of 35 and then 64 for a paying customer, QuantumBasel, in Switzerland. QuantumBasel says it will use the machines to solve problems in finance, pharmaceuticals and chemistry 鈥� but IonQ says the machines will only be shipped in the next 鈥渃ouple of years鈥�.
IonQ claims that the AQ 35 machine will be on the verge of exceeding the capabilities of quantum hardware being simulated on standard computers, while the AQ 64 one will be capable of calculations that even the best classical supercomputers cannot simulate today. 鈥淚f an AQ 64 universal quantum computer were available today, it would certainly be the most powerful quantum computer out there,鈥� says IonQ co-founder .
How do these yet-to-be-built machines measure up against competitors? Monroe says that the AQ 35 machine is likely to have 35 qubits, but that the AQ 64 one could have anywhere from 96 to 1024. 鈥淭he chase for ultra-high qubit numbers, as can be observed with some of our competitors, is mostly recognition that these qubits have low fidelity and they need high number of qubits for aggressive error correction,鈥� says Monroe.
The current qubit record holder is IBM鈥檚 Osprey computer, with 433,听but IBM hasn鈥檛 yet released details about how it performs. Its previous device, the 127-qubit Eagle, has been put head-to-head with a supercomputer at the Lawrence Berkeley National Laboratory in California and won. Google has also claimed for several years that its machines can best classical supercomputers in some scenarios.
On AQ, it isn鈥檛 possible to make a comparison with devices from other companies, because they don鈥檛 publicly release AQ figures for their own computers. IBM wasn鈥檛 able to provide 快猫短视频 with AQ measurements before publication, while Intel and Google didn鈥檛 respond when asked.
But is AQ even a meaningful measure for performance comparisons? at Orca Computing says that AQ as a metric makes sense, and that IonQ鈥檚 approach to building a quantum computer 鈥� creating a string of qubits made of trapped ions that can each talk to any other qubit 鈥� is particularly suited to doing well on it.
鈥淭hey love it, and that鈥檚 why they propose it. And I think other people are kind of a little bit sceptical of it, just because it鈥檚 being proposed by a company who then can do well in it,鈥� he says.
鈥淚onQ is highly focused on applications that will derive value for users and customers, and AQ nicely captures the structure of the algorithms that we are investigating with customers,鈥� says Monroe.
at the University of Texas at Austin says that ultimately inventing new metrics is unhelpful.听鈥淭he suspicion naturally arises that each company will invent and advocate a benchmark on which its devices are the best,鈥� he says. 鈥淭he only metric that can鈥檛 be gamed is what problems you can actually solve.鈥�
