Amid a flood of announcements from tech giants and specialist companies, Amazon has become the latest company to unveil a new quantum computing chip.
Amazon Web Services (AWS) says its prototype Ocelot chip could create a usable quantum computer with one-tenth of the resources of its competitors.
“We really now have a way of combating the major hurdle to a practical quantum computer, which is the errors,” said Oskar Painter, AWS Head of Quantum Hardware.
The Ocelot chip boasts a new technique for making sense out of qubits — the quantum computer equivalent to the bits of information most technology runs on.
According to AWS, Ocelot corrects errors in a more efficient way than quantum chips being developed by other companies, like Google and IBM.
While these companies have been racing ahead with high numbers of qubits in their early quantum computers, they need even higher numbers of qubits to manage the errors they carry.
“[Ocelot’s] right at this point where you’re starting to win with the size of the code, the size of the error,”
Professor Painter said
Details of the new chip were reported earlier today in the journal Nature.
The new chip is the latest in a long line of announcements from tech companies on their quantum computing research, after Microsoft announced it had taken a new step last week.
Brisbane-based company PsiQuantum also published details on a new chip called Omega earlier today in Nature.
How does the chip work?
Quantum computers promise to be significantly more powerful than traditional computers, but there are a number of problems that quantum physicists need to solve before they can be useful.
While traditional computers run on 0s and 1s, scientists can manipulate qubits so they can be both 1 and 0 at the same time. This is what should make them much faster at processing information.
Various types of qubit have been made, but they’re all very susceptible to the tiniest disruptions, meaning they produce huge numbers of errors as they compute.
Amazon’s technique uses a specific type of qubit called a “cat qubit”, after the Schrödinger’s Cat thought experiment, which concerns a metaphorical cat that’s both alive and dead at the same time.
A superconducting-qubit quantum chip being wire-bonded to a circuit board at the AWS Center for Quantum Computing in Pasadena, California. (Supplied: AWS)
Most qubits are vulnerable to two types of errors, but cat qubits are more resistant to one of these types.
Until now, that has made them more vulnerable to the other type of error, according to Andrea Morello, a quantum physicist at the University of New South Wales who wasn’t involved with the research.
“You suppress one of them very strongly at the price of making the other one a little bit worse,” Professor Morello said.
But the new study has shown qubits in the Ocelot chip can also correct both errors.
How does it compare to other technologies?
There are other methods of error correcting — a particularly popular one is to add more qubits to share information across.
Google currently has the most success with this technique, with a chip they announced late last year.
“That’s nice in principle, but once you get out and try and do it, it’s extremely hard because the overheads to make those codes actually work properly are really eye-watering,” Professor Morello said.
At the moment, it’s estimated that this technique would need about 10,000 physical qubits for every working qubit in a computer.
AWS’s technique, meanwhile, should theoretically require fewer qubits to correct enough errors for a useful computer.
The Ocelot prototype chip has nine qubits, while other companies like IBM have loaded hundreds of qubits into single chips.
Professor Painter said their calculations suggested that, if all technologies scaled up in a similar way, their cat qubit technique would need between a fifth and a tenth of the resources that their competitors required.
“There’s reasons to believe that the cat qubit architecture might improve even faster,” Professor Painter said.
But a usable quantum computer using this technology is still distant, with many scaling problems to solve first.
“They’re more of an engineering challenge, but they’re also not trivial,” Professor Painter said.
Professor Morello, who is also working on cat qubits, said that the AWS development was exciting.
“AWS has been in the game for less long than Google and IBM, and it’s really nice to see them really coming to speed,” he said.
Other companies like Microsoft are working on more nascent technologies that will — theoretically — make qubits that produce fewer errors.
Australia has ‘more scalable’ cat qubits
Professor Morello’s team at UNSW announced they’d successfully made cat qubits with atoms earlier this year.
They haven’t yet achieved the same error correction that AWS has managed, but he said their choice of qubit had more potential for scaling up.
Amazon has made its qubits out of metal materials called superconductors, while the UNSW team is using individual atoms to make its qubits.
An AWS engineer installs magnetic shielding around a quantum processor prior to testing at the AWS Center for Quantum Computing in Pasadena, California. (Supplied: AWS)
Once the UNSW cat qubits are embedded in code like AWS’s, they’re expected to be 100-500 nanometres in size – hundreds of thousands of times smaller than the 1 centimetre chip Amazon has developed.
These chips will likely be required in the millions, so the smaller size will be a big advantage when scaling up.
“For me to realise this will take several more years and a lot of blood, sweat and tears,” Professor Morello said.
