On a large table in front of me is some of the world’s most advanced photonic hardware, which may soon drive one of the great technological revolutions of our time. I can see tiny microchips that look like precious jewels, with nanoscale patterns that make them glow like rainbows, along with detectors, filters and switches connected by fibre-optic cables, laid out on smartphone-sized circuit boards. These are the fundamental components of a massive photonic quantum computer that could be up and running in just a few years – though to me, it looks more like the shamble of cables and boxes that makes up my home Wi-Fi gear.
I have come to the Australian head office, in Brisbane, of PsiQuantum, a quantum computing startup that, until last year, was flying under the radar. That all changed when the Australian federal government and the state government of Queensland announced a AU$1 billion investment in the firm – the largest bet by any government, anywhere, on a private quantum computing firm. So what will Australia get for that money? Geoff Pryde, PsiQuantum Australia’s chief technical director, is clear. “We think once we turn this thing on, the quantum computing era begins.”
That is a big claim for a company only just beginning construction of a facility to house its quantum computer, at a 13-hectare site near Brisbane airport. But to understand why PsiQuantum thinks it can beat Google, IBM and other big players in the quantum computing industry, it helps to look at the route it is taking.
Advertisement
“It’s not the same trajectory,” says Terry Rudolph, one of the firm’s four co-founders, who also happens to be the grandson of legendary quantum physicist Erwin Schrödinger. “If your goal is to reach the top of a skyscraper, then building a really big ladder might make sense. However, once you realise the goal is the moon, then it doesn’t matter how many ladders you stack up on top of each other. If you look at what we have been building, we are developing a rocket, and the moon has been our goal since we started the company.”
Put more technically, PsiQuantum has excused itself from participating in what the industry calls the noisy intermediate-scale quantum (NISQ) computing era, during which key players have built machines with up to 1000 qubits, or quantum bits, which are prone to error and have limited practical application. It is widely held that these NISQ-era computers are marvels of technology but have little practical use.
“In the past few years, it has become quite clear that the value proposition of NISQ is far less rosy than originally thought,” says at the University of New South Wales in Sydney. “There really isn’t much useful computation that a NISQ device can perform which cannot be done on a classical machine.”
Instead, PsiQuantum plans to scale up, fast, to a one-million-qubit machine by the end of 2027 – far larger than any that exists today. It all begins with the mess of hardware in front of me. PsiQuantum has thrown itself wholly into photonic quantum computing, which means that it has to manipulate particles of light into quantum states that will serve as the qubits in its computer, the rough equivalent of the transistors found in conventional computers.
That comparison is apt, because PsiQuantum is using the same semiconductor fabrication techniques that are used to make the huge number of tiny transistors found on regular computer chips. “PsiQuantum worked out how to miniaturise to the nanoscale all requisite photonic components and manufacture them using the same technology used to build laptops and cell phones,” says Rudolph. “Such semiconductor engineering is the only approach to building billion-component machines that humans have invented thus far.”
PsiQuantum has partnered with US-based chip manufacturer GlobalFoundries to produce its photonic chips. Some of those components are sitting on the table in front of me, including a small black case that looks like it is housing a precious gem. In fact, it is an example of the microchips that have been developed to drive the control and filtering of photons in PsiQuantum’s device.
This approach aims to benefit from decades of experience both in chip fabrication and manipulating quantum states of light. There are a number of different technologies that can be the basis of a quantum computer, including superconducting materials and trapped ions, but photonics can be more robust than these. “This combination of existing large-scale, efficient, high-tech industry capability and light’s noise resistance makes photonics a strong candidate for the first useful quantum computer,” says Rudolph.
That isn’t to say the path has been easy. PsiQuantum’s Dylan Saunders says the firm had to overcome the challenge of building microchips that are room temperature on one side and the temperature of deep space on the other. Another crucial breakthrough was successfully manufacturing filters that can block out unwanted light from the photon source but let the qubit photons go through.
“We need to block that unwanted light by 12 orders of magnitude, which is a lot,” says Saunders. “An analogy is it’s kind of like letting off a nuclear explosion at one end of a lecture theatre and blocking the explosion so much that you could hear me talking to you if you were standing there. That’s how much we need to suppress it.”
Many challenges still remain, particularly in the integration of so many millions of components and the optimisation of the photonic qubits. These small parts in front of me will need to be replicated on a massive scale: PsiQuantum must manufacture tens of thousands of photonic chips and connect them with over 1000 kilometres of optical fibre. Once complete, the photonic quantum computer will cover an area of 100,000 square metres, which includes the cryoplant needed to keep everything cool.
Despite the task ahead, Pryde says there are no insurmountable gaps between where the team is now and the big switch-on in 2027. “There’s no point in this hardware picture where some miracle needs to occur,” he says.
In fact, PsiQuantum is so confident of its progress that it already has a team of quantum software programmers working with numerous industry clients to develop algorithms for extracting useful information from the quantum computer when it is up and running. These clients include pharmaceutical giant Boehringer Ingelheim, which wants to study the properties of enzymes that are vital for drug absorption, and Mercedes-Benz, which hopes to use the quantum computer to improve battery design for electric vehicles.
“Once the hardware is installed and the system is validated, we want to have the algorithms that we’ve prepared ready for that generation-one machine running ASAP,” says Pryde.
So, will the big bet of skipping the NISQ era pay off? at the University of Sydney says there are clear benefits to not building smaller devices. “It may be important as part of technological development, and realising those things and showing them off may be exciting, but you could easily argue they are a distraction from the main game. PsiQuantum want a million-qubit device. They have a robust and serious plan and are in a position not to have to show off baby steps and instead go for the main game.”
Morello says there is an argument that the NISQ era was “useful and necessary as an intermediate step” for researchers to get experience in engineering and programming quantum computers, but ultimately PsiQuantum may prove this was unnecessary. “They said ‘we’re just going all the way for the big thing’,” he says. “And by now, I think it’s fair to say that they were right.”
But exactly what a million-qubit device will be able to do – and whether it will truly herald the start of the quantum computing era, as Pryde claims – remains to be seen. “We are all waiting for the day when a quantum computer is able to give a convincing demonstration that it can do something useful that cannot be done by a classical supercomputer,” says Bartlett. And while PsiQuantum has a clear plan to scale up, it may not be the first to get there.
“Nobody yet knows what technology is going to enable large-scale, useful quantum computing,” he says. “It could be photonics, superconductors or ion spins in semiconductors, and there’s more. We don’t know yet who is going to win.”
