Amazon claims to cut quantum computing timeline by 5 years with new Ocelot chip

Amazon Web Services has unveiled a new quantum computing chip, Ocelot, that the company claims could reduce error correction costs by up to 90% compared to traditional approaches.

“We believe this will accelerate our timeline to a practical quantum computer by up to five years,” says Oskar Painter, AWS director of Quantum Hardware, in a blog post released today.

Ocelot uses so-called cat qubits that are inherently resistant to bit-flip errors and, in theory, are more fault-tolerant than some other approaches.

But there’s still a lot of work to be done.

“Next milestones for Ocelot are to push the error-corrected logical qubit error rates even further down, by multiple orders of magnitude,” Painter told Network World.

It’s not yet at the stage where it can be made available to the public.

“As the Ocelot architecture is advanced to include more physical qubits, and multiple error-corrected logical qubits that can perform logical computations at the logical qubit level, there is an opportunity for us, even if early, to provide access to customers to this early-stage hardware,” Painter said. “We haven’t decided when that date will be, and we will continue to evaluate when that right time might be, but now is too early.”

It will eventually be available on the AWS Braket quantum cloud service, Painter said.

“It will likely be beside other quantum hardware offerings, much like in EC2 where AWS’s Graviton chips are offered alongside NVIDIA and other instances,” he said.

The first industries to use it will probably be those developing new materials, new chemical processes, or in molecular biology, Painter said.

“AWS is showing compelling results,” says Rob Schoelkopf, chief scientist and co-founder at Quantum Circuits, a quantum computing company that is also working on fault-tolerant qubits. Schoelkopf is known as one of the founders of the superconducting quantum computing field. “I look forward to seeing where it goes.”

The AWS announcement comes just one week after Microsoft revealed its own quantum breakthrough, the Majorana 1 chip, intensifying the race among tech giants to deliver quantum computing at scale.

Both companies claim that their two approaches, both still in the very early stages, have the potential to scale better than their competitors.

But they’ll have a lot of catching up to do.

“In terms of quantum capabilities, both chips are far behind companies such as IBM, Quantinum, IonQ, Atom Computing, etc.,” says Paul Smith-Goodson, quantum computing analyst at Moor Insights & Strategy.

The Ocelot chip is still only experimental, he says. “Production use could be decades away.”

Meanwhile, the other companies in this arena have been making a lot of advances over the past five years.

“We are at the point where we can now do useful work,” said Smith-Goodson, though he puts “useful” in quotation marks. “Fault-tolerance is still years away.”

“If the Ocelot is able to refine and advance the technology so it provides more efficient bulk error correction than the other quantum companies, it is possible, but not likely, it could pass them,” he said. “Although Amazon’s work is promising, it will be hard to tell if it has compressed quantum development time without more evidence.”

According to Amazon’s Painter, practical use of the Ocelot chip will require around 100 logical qubits, and logical qubit error rates of 0.000001%.

“That would likely be the very earliest,” he says, “with most applications requiring several orders of magnitude improvement in error rates.”

Another challenge is the development of quantum algorithms suitable and optimized for a quantum computer,  Painter added.

Cloud platforms take the lead

The Amazon announcement is a “moderately big” deal, says John Preskill, director of the Institute for Quantum Information and Matter at Caltech and a professor of theoretical physics.

Right now, he says, Google seems to be in the lead, having made impressive progress toward scaling up quantum technology over the past year.

“But because such big challenges remain we can’t say at this stage whether the Google or AWS approach has the better long term prospects,” he said.

Still, the pace of progress over the past few years has exceeded his expectations, Preskill said.

We’re now in an era where the cloud companies are making their own quantum chips, said Holger Mueller, an analyst at Constellation Research.

“It’s an inflection point,” Mueller said. “The cloud guys don’t want to have a repetition of Nvidia.”

Microsoft, Google, AWS, and IBM are all betting heavily on the superconducting approach to qubits, however, just using different materials. Microsoft’s topological qubit-based Majorana 1 chip is also a variant of the superconducting approach.

“They’re all using different materials, with different qualities,” Mueller said. “But it looks like superconducting has won right now.”

The key AWS innovation

Ocelot’s key innovation is in using Tantalum, a superconducting material, to boost the performance of the oscillators used to store quantum states.

Amazon’s Ocelot chip also stands out because it contains different type of qubits on the same chip — five cat data qubits and four qubits for detecting errors on the data qubits, plus five buffer circuits for stabilizing the data qubits, for a total of 14 different components.

“It shows that it is possible to reduce the hardware complexity of a reliable quantum processor drastically,” said Mazyar Mirrahimi, director of research at France’s National Institute for Research in Digital Science and Technology.

The choice of the cat qubit provides a bottom-level protection for error correction, Mirrahimi said. “The road is still long, even though perhaps shorter than other approaches based on conventional qubits which do not benefit from this bottom-level protection.”

Other companies are trying the brute-force approach to scaling quantum computers, adding more and more qubits in order to reduce errors through redundancy.

“Enterprises will inevitably benefit from a more efficient approach to superconducting quantum computing with error correction,” said Ray Smets, CEO at Quantum Circuits, which is also building quantum hardware with a focus on reducing errors from the start, rather than correcting them later.

“The ability to overcome error correction hurdles most efficiently will determine who stands out in producing viable, commercial-ready quantum computing,” Smets  said.