Of all the technologies that have applications and implications for financial services, quantum computing is the strangest and the most distant – but also the most profound.
Commonwealth Bank of Australia, Goldman Sachs and UBS are among those financial institutions with quantum computing research projects. Several startups are developing quantum-related businesses. So what is it, and how can it be used?
A recent report* by a group of scientists from Massachusetts Institute of Technology, funded by the U.S. Department of Defense, offers guidance into what to expect, and when.
Quantum information processing studies how information is gathered, transformed and transmitted at the level of atoms, ions, photons and elementary particles. In practical terms, this allows computing to be done at the ultimate limits imposed by the laws of physics. It’s like taking Moore’s Law – which says that our technological advances double a transistor’s computing power every two years – to its ultimate destination.
Such an outcome is based on the weird, counterintuitive properties of quantum physics, which Albert Einstein characterized as “spooky at a distance”.
This article is about quantum’s practical outcomes, rather than its physics. But the essential thing to know is that whereas traditional computing is ‘digital’ (i.e. bits of information are either stated as 0 or 1), quantum computing allows for a third possibility (expressed as a ‘qubit’), which is a property that is simultaneously 1 and 0. By using qubits, quantum computing offers power at levels that are exponentially greater than any traditional computer's. (And when it comes, DigFin might need to change its name!)
The science goes back to 1900, when Max Planck discovered that light is comprised of discrete chunks, or ‘quanta’, of elementary particles. Nature’s building blocks are digital: there are pieces of it, or not, in a given space. But these building blocks cohere in uninterrupted waves. So the quanta are both specific particles (0s or 1s) and they are waves (0 and 1), and this duality leads to mind-bending properties when studying the universe at the very smallest level.
Quantum science isn’t new, therefore: quantum mechanics have been used to develop semiconductors, lasers and atomic clocks. But only since the 1990s have researchers decided that the weird goings-on in the quantum world aren’t just about odd side effects, but are features that can be harnessed to create new benefits.
What’s it good for?
In particular, quantum computers can accomplish tasks that no classical algorithm could handle (think of waaaay-bigger data analytics). They can communicate securely in a way that can’t be hacked. And they can find or generate patterns invisible to classical computers (machine learning out of science fiction).
But it's hard to be sure, because traditional computers can’t make accurate predictions about what these outcomes will look like. Today, to the frustration of quantum computing's researchers, there aren’t any applications for quantum computing, aside from breaking public-key encryption.
The industry is therefore now building basic quantum computers, which in turn will be tasked with developing this new generation of machines that encrypt, communicate, measure and analyze at levels we can’t really understand today – and not just as single machines, but in a network: the quantum internet.
MIT’s researchers believe, however, we are now at a threshold. They say a 50-qubit quantum computer will achieve the scale necessary to outclass the most powerful classic computers. That means we have no idea what a quantum computer of that size will be able to calculate, because it exceeds any model we can develop with today’s computers. This milestone is called quantum supremacy.
That is also the point at which quantum computers will begin to generate simulations of other quantum systems, putting us on the path to realizing computers with literally unimaginable capacity.
Sounds speculative? Well, consider this: in March 2017, researchers at the University of Science and Technology of China and Zhejiang University unveiled what they say is the world’s first quantum computer, with 10 qubits. They hope to have a 20-qubit model by the end of this year.
So assuming that claim stacks up, a 50-qubit computer is now on the horizon, well within the MIT study’s prediction of realizing this within five years.
And it won’t take much after that to develop quantum computers operating with 100 to 1,000 qubits, which may be enough to break encryption based on a public and a private key: in other words, blockchain and any system relying on two-step authentication – such as your online bank account.
So, unless you’re a government’s spy service or a master criminal, what good is a technology that simply renders all of our cyber protections irrelevant?
Quantum computing has three areas in which applications are likely to deliver major benefits.
First is communication. Transmitting messages will be much faster and more reliable, and new means of encryption will arise. “Quantum secret sharing and quantum data locking’s…security is guaranteed by the laws of physics,” says the MIT report. (Today, companies and governments are already using quantum cryptography and transmission to communicate with satellites.)
Together these properties will fuel the rise of a quantum internet. Like our internet today, its emergence will enable a new world of applications, in an environment that should be far more secure.
The second area for apps is sensing and measuring. In the research world, this means clocks, maps and perhaps ways to monitor, say, seismic activity. For businesses, this translates into detection and imaging. Although researchers are now focused on creating a far more accurate GPS, the possibilities for KYC, identity, and insurance are also huge.
The third app is machine learning to develop algorithms. Once we achieve quantum supremacy, the ability to source, crunch and analyze data will be many times greater than what is now possible. Imagine putting quantum computers to the task of finding patterns in data: not only to detect patterns that elude classical computing, but to generate them as well. Think about the alpha in that technique.
…And three challenges
It’s not going to be a straight line from China’s new 10-qubit machine to quantum supremacy. There are broadly three challenges.
Challenge number one: our inability to model apps using classical computers. We only have a vague roadmap. A quantum ‘internet of things’, a quantum cloud, and secure multi-party computation (which could replace blockchain) are all possibilities. We imagine these because we have primitive versions today. We do not know what the ultimate limits of physical laws will allow. We won’t know until we build bigger quantum computers.
Challenge number two: building 50-qubit or larger quantum computers that work. Researchers must learn how to scale their creations, while maintaining fidelity (ensuring consistent application of quantum logic to the utmost limit).
Not all quantum computers are the same, so it’s unclear how exactly the problems of scale and fidelity will play out. The MIT report outlines versions such as ion traps, atom-optical systems, and different quantum versions of superconductors and semiconductors, all of which is well above DigFin’s own processing power. For our purposes, it’s enough to know that the quantum world is bizarre, and it’s easy to generate wrong signals. Researchers will be experimenting with a range of theories and tools to develop quantum computers, and what that implies for applications is unknown.
And challenge number three: persistence. Governments, universities, tech companies and potential users such as financial institutions must collaborate over a five- to ten-year period, if not longer. Quantum computing is becoming a reality but it requires a lot of investment into basic research as well as into enterprise-ready apps. We need to improve computers to the point that they can carry out computations in a mundane setting – a solid-state, room temperature environment, as opposed to the giant, deep freeze labs where the first generation of machines is being born. Getting a bank’s C-suite to sign off on these projects, and stick with them, won’t be easy.
But the MIT report says, “At this moment, there is a clear path to developing mid-scale quantum computers that can solve problems that are difficult or impossible for classical computers to solve.”
Quantum cryptography is already a reality. For anyone in financial technology, this is a field worth keeping an eye on.
* “Future Directions of Quantum Information Processing: A workshop on the emerging science and technology of quantum computation, communication, and measurement”, by Seth Lloyd and Dirk Englund of Massachusetts Institute of Technology. The report is the product of a workshop held in August 2016.