Quantum computing in the cloud - Xanadu discovered in Toronto
Chris Middleton speaks to the CEO of Canadian quantum cloud startup Xanadu about what it claims is the world’s first photonics-based quantum computing platform available in the cloud.
Moore's Law may not be sufficient when it comes to processing power. Some believe that computers are not getting faster fast enough for the emerging tasks that are expected of them, whether it's crunching vast data sets or looking for probabilistic connections within much smaller ones.
In areas such as drug discovery, particle physics, genomics, and materials research, quantum computing is increasingly proposed as a better model for the IT industry, alongside related quantum technologies such as timing, imaging, sensing, communications, and security. Yet all these are hard tech, which demands patient capital: long-term, speculative investment coupled with an appetite for risk and uncertain reward.
There are three key challenges facing all quantum innovators:
- The industry needs to focus on developing and building the hardware first, before new, non-binary applications can emerge.
- With subatomic technologies, system noise within that hardware is a serious problem - due to microscopic flaws in fabrication. Meanwhile, writing algorithms that can work in a noisy environment is tough. And,
- Building viable quantum hardware is not the same as building a viable commercial industry, in which qubits eventually sit alongside bits (classical computing) and neural networks (AI) in a trinity of complementary models.
Canada isn't the first country that springs to mind as a potential leader in this space alongside the likes of the US, China, and the UK, but Toronto-based 2016 startup Xanadu aims to change that. It has launched what it claims is the world's first photonics-based quantum computing platform available commercially in the cloud.
According to an announcement from the company, the Xanadu Quantum Cloud currently gives developers access to eight- and 12-qubit processors, and soon a 24-qubit machine. As a photon- (light) based system for quantum calculations - as opposed to superconductor or ion-trap systems - these processors can operate at room temperature and could potentially integrate more easily with fibre-optic based telecommunications.
To date, Xanadu has raised $45 million from investors, alongside grants from the Defense Advanced Research Projects Agency (DARPA). Developers can already access its open-source tools Strawberry Fields and Penny Lane on GitHub, revealing a Beatles fan in its leader. That may suggest global ambitions.
In search of a problem?
So does the company plan to become a huge private fiefdom, like the Xanadu in Citizen Kane? Has a new computing Kublai Khan emerged to found a quantum dynasty? I spoke to founder and CEO Christian Weedbrook to find out.
First, is Xanadu a true quantum system, or merely using quantum elements to speed up processing within classical systems? He says:
It's the world's first photonics-based quantum computing cloud platform and we are truly excited about it. Every other cloud platform essentially uses electronics. Instead of photonics they use electrons - matter - and they have a very different approach to scaling up to a fully functioning, fault-tolerant quantum computer.
We decided to take a unique, alternative approach using light. One of the biggest advantages is the extremely low amounts of power that are needed to run these devices. We're leveraging the weird properties of quantum physics to do things that traditional computers, classical computers, would never be able to do. Or they could do it but it would take them hundreds of years.
A familiar message in terms of the time-saving element. But when it comes to commercialising it, is quantum computing a solution in search of a problem, or has Xanadu identified specific opportunities? Weedbrook says:
The world can always use more computing power. One way to think about the quantum computing landscape at the moment is you can break it up into two lines. The first is what we can do in the next three to five years with qubits that are noisy and don't have that fault tolerance or error correction. And the second is what can we do after that with a fully functioning, fault-tolerant quantum computer. But that's a very difficult machine to build.
In the near term, what differentiates Xanadu using a photonics-based approach is the things that we can do now, which have business applications - network planning, logistics, things of that nature.
Our plan is to get early adopters that we can build an ecosystem around: government labs in the US and here in Canada, financial institutions, large corporations - pharma and materials design are classic examples - and logistics. All these industries have problems that get more and more difficult, that don't scale well, so you need a quantum computer to help solve them. We have paying customers already on the platform.
What's remarkable in the last couple of years is that large corporations and banks such as Goldman Sachs have been forming small quantum computing teams, because they don't want to get left behind. But it's a unique skill set that only a few people in the world currently understand.
Indeed, some have suggested that is a concern. When technology becomes too abstract for most people to comprehend, then transparency and auditing are difficult. Who could explain or justify decisions made by quantum neural nets, for example?
That's a valid criticism. We have to be careful, particularly with a technology that may be the missing link to a singularity event. I don't think ethics and quantum computing should just be presented to you. It's something that the whole community should be converging on thinking about. But building a quantum computer that's fault tolerant - the end game - is extremely hard. It's not going to be built tomorrow, which gives us some time.
Does he see quantum computing as a distinct alternative to the classical world, an evolution of it, or simply a component for solving specific problems?
First, it's a safe bet that it will complement traditional computing whatever happens, in the same way that GPUs complement CPUs, or special-purpose chips complement other types. A computer will naturally access the CPU or the quantum processor when each is needed.
This is a personal view, but more and more I'm leaning towards quantum computing replacing ‘old' computing. One reason is that, if you look at our photon-based approach, roughly speaking you can press a button and make this computer classical by making qubits act like normal bits of information. If quantum computing can infiltrate all computers, then you can just simulate a traditional computer, which means there is an opportunity to really take over all of the computing industry - up to a point.
The company's global ambitions seem evident. But until then, physics and history combine to pose some tricky problems. For example, putting a quantum computer in the cloud could put a brake on its power, given the slow speeds, latency, and poor reliability of some broadband infrastructures.
Yes, with quantum cloud quantum providers you do see that slowdown in retrieving the asset. I would say that there are some solutions to that, such as a hybrid system where you actually have a quantum chip on premise alongside AWS or Azure, for example.
But the ultimate goal is to have a quantum computer that can solve problems that a classical computer could never do, or would take thousands of years to solve. So slowing something down by an hour versus a thousand years is irrelevant. As you're scaling up and getting more powerful quantum computers through fault tolerance and error correction, any slowdown is going to be negligible compared to the size of the problem solving.
Are real applications emerging yet for such systems?
A lot of applications have been inspired by what's been done classically, and the thinking has been how do we make them quantum. But I think the true revolution will be in thinking about it from the quantum side of things. There's an analogy here with the PC revolution: people didn't really know what applications would be useful, there was no idea about the internet. I think there'll be something similar for quantum: we really don't know the full extent of what a quantum computer can do.
Making real change
With a team that now numbers 58 - many of whom have PhDs - what drives Weedbrook and his colleagues: the academic research glory of pushing back the boundaries of human knowledge? Or something more commercially focused - given those Beatles allusions and the name itself, Xanadu? Both speak of lofty ambition...
I think often you see with startups, it's a cliché to really want to solve huge, world-changing problems. Our mission statement has a hint of that, but to be honest it's too much of a cliché now to really believe it. So, our mission statement is to make quantum computers, make them useful, and available to people everywhere.
It's how do you start in helping customers solve problems - like in drug discovery. How do you find the best candidates significantly faster in order to synthesise them? That problem's not going to change. So how best to solve it?
This may be correlated with the fact that we're working on subatomic particles, but it's in very small systems that you make real change in the world. The common examples are helping with global warming, curing cancer by identifying drugs, these are really important problems for humanity.
But taking a different approach, how do you stay excited on an individual, personal level each day? That gives you the best chance of solving truly important problems.
According to the 80s song, Xanadu is the place where "a million lights are dancing". That's photons for you. And it turns out it's not in China, but Toronto.