Three years from now, if all goes to plan, an industrial estate near Brisbane Airport will house an enormous quantum computer: one of the most complex machines ever built.
From the outside, it’ll be a hangar topped with a plume of white steam from its cryogenic freezers.
Inside, it’ll be like nothing else on Earth: racks of cabinets cooled to the temperature of outer space, holding custom-built silicon wafers that can detect individual photons, particles that have no mass and are also (confusingly) waves.
The Californian start-up behind the machine, PsiQuantum, says it will be the world’s first useful quantum computer, able to solve problems conventional computers cannot.
All up, it will cost more than $1 billion. Most of that will come from taxpayers, thanks to a massive investment from the Queensland and federal governments.
A rendering of PsiQuantum’s Brisbane quantum computer, offices and research labs. (Supplied: PsiQuantum)
So what will it do? How will it be useful?
Given the scale of investment, surprisingly few agree on these points.
Partly due to the inherent uncertainties of quantum mechanics, even basic specifications like “size” are hard to nail down.
“There is a huge gap between what is required for commercial utility and the very unclear ‘specs’ of what PsiQuantum is willing to say publicly about the machine,” says Simon Devitt, a theoretical quantum physicist at the University of Technology Sydney.
There are also questions over who gets to use the machine.
Politicians like to call the computer “Australia’s Moon landing”, and many in the Australian quantum community hope public funding means they will get priority access. If the machine lives up to expectations, there could be a long queue.
But PsiQuantum says the funding agreement includes no detailed formal arrangement around access.
“We’re a for-profit company,” PsiQuantum’s chief scientific officer Pete Shadbolt says.
“We will sell time to people on a commercial basis.”
How powerful will the computer be?
To understand the power of the computer you have to know something about quantum mechanics, which predicts effects that are counter-intuitive and difficult to reconcile with our daily experience of how the physical world works.
Dr Shadbolt is a co-founder of PsiQuantum. (Supplied: PsiQuantum)
In classical computers, like the one you’re using now, information is stored as numbers in a binary form: 0 and 1. These are known as “bits”, short for “binary digits”. Everything, from text to pictures or sound, is stored as long strings of these bits.
Each bit of information is stored in a tiny circuit etched onto the surface of a silicon chip, using processes based on classical physics. A bit is always in one of two physical states, similar to an on/off light switch.
But at very small scales, near the size of an atom, the familiar laws of classical physics break down. In their place are the quirky rules that govern quantum mechanics.
Particles can be in two places at once, or they can be invisibly and mysteriously linked or “entangled” across distance.
Instead of binary bits, a quantum computer uses qubits, which may be atoms, photons, or even tiny loops of superconducting metal.
Quantum computers exploit the quantum properties of qubits to solve certain computational tasks exponentially faster than a classical computer.
PsiQuantum houses its components in cryogenic cabinet freezers cooled to near absolute zero (-270 Celsius). (Supplied: PsiQuantum)
The world’s current largest quantum computers have just over 1,000 qubits. PsiQuantum says its computer will have a million “qubits”, but exactly what that means is complicated.
Determining quantum computing power isn’t a simple matter of counting qubits.
Quantum computers are plagued by noise and errors that can destroy reliable calculations; to really work, they need systematic error correction. This is known as “fault tolerance”.
Depending on the type of hardware and the strength of error correction, hundreds or thousands of physical qubits are generally required to make one “logical qubit” which can be used reliably for computation.
Each custom-built silicon wafer contains tens of thousands of single-photon detectors. (Supplied: PsiQuantum)
Computing experts offer wildly varying estimates of how many logical qubits PsiQuantum’s computer will have.
Howard Wiseman, director of the centre for quantum physics at Griffith University, which has partnered with PsiQuantum, says the computer may have “1,000 to 10,000”.
Stephen Bartlett, a quantum physicist at the University of Sydney and the director of Quantum Australia, a new quantum industry hub, gave an even higher figure.
Simon Devitt from UTS suggests the number could be a “few dozen”.
PsiQuantum hasn’t publicly quoted a figure — until now.
“We are planning for our first machine to have about 200 logical qubits,” Dr Shadbolt told the ABC.
At that scale, it’ll have more logical qubits than any current quantum computer. But whether it will meet the goal of “useful” quantum computing isn’t yet clear.
What could the proposed computer do?
The word “useful” is being used a lot in quantum computing these days. After years of experiments and prototypes, the esoteric field may soon leap from the research space into the commercial one.
Several companies are vying to build what they variously describe as the world’s first “useful”, “utility-scale”, “commercial-scale” or “industrial-scale” quantum computer.
All these terms mean a quantum computer that can solve real-world problems classical computers cannot, or that can at least solve them more cheaply.
The hype says quantum computers will one day break state-of-the-art encryption and unpick intractable logistics problems like trucking routes or organising a port. They may accurately model the countless variables of the world’s climate or give humanoid robots capable artificial brains.
PsiQuantum’s Brisbane computer won’t be doing any of that, Dr Shadbolt says.
“That first machine is the starting line as opposed to the finish line on quantum computing.”
It won’t be breaking codes.
“We don’t expect meaningful code-breaking on that first-generation machine.”
It also won’t be solving optimisation problems, such as the example of ports or trucking routes.
“For optimisation-type applications, you want a bigger machine later in the game.”
It definitely won’t be used for machine learning.
“The depth and the maturity is just not there with those algorithms.”
Dr Shadbolt expects the Brisbane computer will be mostly used for computational chemistry, modelling how atoms and molecules interact with their surroundings.
It could help develop new kinds of drugs or battery chemicals, or continue the search for a room-temperature superconductor.
PsiQuantum is already working with pharmaceutical company Boehringer Ingelheim on drug discovery, Mitsubishi Chemicals on photovoltaics and data storage, and Mercedes-Benz on batteries.
“As far as the problems we expect to solve … it’s basically condensed matter systems and small molecule chemistry.”
He says PsiQuantum in Brisbane will lead a “dedicated climate research” program investigating new kinds of fuels, solar panels, and catalysts for carbon sequestration.
“It’ll be spearheaded by PsiQuantum as part of this transaction [with the federal and Queensland governments].”
How far are quantum computers from other applications?
Dr Devitt says PsiQuantum’s proposed computer will be too small to run most quantum applications.
“It doesn’t seem those 200 logical qubits are going to get you very far.”
A recent study found modelling a theoretical type of ozone molecule encased in a cage of carbon atoms will require millions of physical qubits and take many years.
“A lot of these chemistry questions are of the same order [of qubits], if not larger,” he says.
Other applications, such as factoring for code-breaking, require at least “20 million physical qubits”.
A PsiQuantum silicon wafer undergoing testing. (Supplied: PsiQuantum)
Stephen Bartlett from Quantum Australia said “utility-scale” computing meant “many thousands … maybe up towards a million” of logical qubits.
“That’s what people mean by utility scale. We have a high bar to jump.”
Australia’s Chief Scientist Catherine Foley, who provided technical advice to the government during the evaluation of the PsiQuantum proposal, says quantum computing will “touch the lives of all Australians” at some point.
“There is not yet consensus around what the first quantum computing use case will be or when it will arrive,” she says.
She says companies around the world are looking into ways to use quantum computers for drug design, financial services, transport and logistics, and climate modelling.
“These are just the applications that are anticipated now. Chances are that applications that will prove the most useful have not been imagined yet.”
No detailed arrangement to give Australian researchers priority access
Demand to access PsiQuantum’s computer could be very high.
“We want to make sure that that the machine … is absolutely saturated with demand,” Dr Shadbolt says.
“The ambition is that it’s not idling for a single second and there’s people queuing out of the door to use it.”
So which users get priority?
Dr Devitt, who’s setting up a network for quantum software developers, hopes Australian quantum researchers will get some priority, given the large public investment.
“Does the government investing in PsiQuantum mean that Australian entities will have a subsidy or be able to use the machine freely? I don’t know. We don’t have any details about any of that.”
Another view of the cryogenic cabinet developed by PsiQuantum. (Supplied: PsiQuantum)
Professor Bartlett suggests an arrangement like the publicly funded supercomputing infrastructure which Australian academics currently access at a subsidised rate.
PsiQuantum says paying customers will have priority, including some of the “Fortune 500” companies in the pharmaceutical, automotive and energy industries that have been working with PsiQuantum for years. They are at the front of the line.
“They’re paying money today to put skin in the game on on quantum computing, and that’s because they hope that it will give them a massive competitive advantage when the machine lands.”
He adds that PsiQuantum has a responsibility to the Queensland and federal governments, which own an undisclosed number of shares in PsiQuantum, to “get really rich”.
“The plan is to make ourselves a very, very successful wealthy company.
“We’re not just selling to the highest bidder, of course. It is in our interest that our customers are both paying for the service, but also tackling problems that we think … showcase the capacity of the machine.
“We will be very thoughtful — when we have a choice — about what choice we will make on our applications.”
What’s next?
In July, a few months after the Brisbane announcement, PsiQuantum revealed plans to build a second quantum computer in Chicago following a $US1 billion investment plan from the state of Illinois, Cook County and Chicago.
Dr Shadbolt says the US computer will be similar to Australia’s.
“We’re doing the same stuff … in both locations.”
He added PsiQuantum will build other, more powerful computers at the Brisbane site.
“We hope we’ll look back on those early systems and say they were pretty wimpy.”
A rendering of the Illinois quantum campus, which will house PsiQuantum’s second quantum computer. (Supplied: PsiQuantum)
Meanwhile, plenty of other companies are also pursuing useful quantum computing.
Diraq, a startup led by Andrew Dzurak from the University of New South Wales, aims to build a “commercially relevant” quantum computer in Australia within five years.
Silicon Quantum Computing, founded by Michelle Simmons, also with the University of New South Wales, recently raised $50 million to manufacture the world’s first scalable, error-corrected quantum computer.
Google, IBM, Microsoft and other large tech companies are all building quantum computers.
A team at the University of Sydney is looking at ways to use future quantum computers to develop new molecules that can treat skin cancers or improve sunscreens. It’s part of an international program developing applications for the quantum computers “expected to emerge in the next 3–5 years“.
The future looks bright, but in the minuscule and complex arena of quantum mechanics, nothing is ever quite as straightforward as it seems.