Soon quantum computing power will be available in the cloud as a service. Source: Shutterstock

Quantum-as-a-Service not as crazy as it sounds

WITHIN most of our lifetimes, we will be able to use quantum computing to help us solve problems considered too complex for the computers we use now.

The nature of quantum computers’ physical needs at present (and for the foreseeable future) means that we won’t be carrying around quantum devices in our pockets or even have them sat in our present-day data centers.

Quantum devices need to be supercooled to a fraction of a degree above absolute zero for the superconductors in them to function.

It’s likely therefore that we will access quantum computers’ power in the same way we currently use the cloud: by subscription, lease, or rental of a service.

The machines themselves will need to be held in specialist facilities: the similarities between the early mainframe deployment (and appearance) is obvious:


The mainframes of old pictures look remarkably similar to today’s cutting-edge quantum machines. Source: Bosch/IBM

Quantum computers are already available in this manner. IBM’s Quantum Experience provides everyone with access to a limited-qubit computer. For the moment, its uses are educational and experimental, but the cloud-based model could easily be scaled up to help us solve real-world problems.

So, using quantum power alongside ‘classical’ computing methods (as used by the machines we all work on every day) is, coincidentally, the way which we once used mainframes’ resources.

Soon, we’ll begin to utilize the astounding power of the new generation of computing similarly. But what is quantum computing, and how will it prove to be useful to us?

Many of our everyday problems are of exponential complexity. Put simply, as events take place and items interact with one another, the levels of possibility multiply, exponentially.

Here’s a simple example. In a logistics chain, if a truck carrying goods fails to arrive on time at a designated drop-off point, it might prove useful for the load to be split into two and distributed separately; some items might be ordered via express service, while others are less time-dependent, for instance.


A single possibility multiplies exponentially. Source: Shutterstock

However, the same possibilities and foibles of happenstance which might have affected the first truck are now at play on two trucks. So, the possibilities (already at least in the dozens) have doubled.

At the next phases(s) in the supply chain, similar possibilities and alterations are possible, so the number of ways in which a consignment may reach its various destinations multiply at a massive rate.

Asking a classical computer to plot its way through all the exponential pathways to find the most cost-effective logistical solution is not going to yield an answer quickly.

In fact, for complex situations, running through all the possibilities would take years of classical processing.

Instead, we rely currently on algorithms which make a best-guess at what would be most appropriate. Classically-computed “guesses” are based on historical data, and we allow alterations to plans as events take their course.

Even machine learning (ML) techniques, which allow changes to be made to the determining algorithmic processes themselves will still only be producing an approximation of the most effective outcome.

The problem of exponential multiplication troubles many areas of industry – the example of a logistics chain was chosen here for the sake of simplicity. Another, more complex use for quantum computing will be, for example, in the pharmaceutical research field.

The chemical formulae of modern drugs are highly complex and comprise of molecules which form when electron orbitals overlap. One single iron sulfide molecule, for instance, requires the calculation of 76 electron orbitals, which has 2^76 possible outcomes (75,557,864,000,000,000,000,000 different versions, more or less precisely). Considering all the permutations of a researched treatment at a molecular level is the type of routine at which quantum computers will excel.

Mapping out new drug formulations and predicting weather systems are just two headline-grabbing quantum computing uses. On a more mundane (but practical) level, what about planning optimum routes for several thousand different items and people moving separately around the globe attending events and exhibitions, but which all need to be kept organized to keep costs as low as possible?

The complexity of these issues can be solved by combining classical and quantum computing. The machines on our desks (or in our pockets) will hold models and gather the required data, plus provide an approachable interface. But the heavy lifting, the quantum possibilities, would best be processed remotely by quantum devices and sent back to us over the Internet – in much the same way we use any cloud service today.

The math of the extent of requirements we’ll need from our quantum computers is theoretically simple to determine. If a fairly complex molecule has 80 or 90 bonds between atoms, a quantum computer will need the same number of qubits (as a start) to determine the possibilities for molecular interactions. Currently, IBM has produced a prototype machine capable of calculating to 17 qubits, and Canadian company D-Wave is selling commercial quantum computers (of limited power) for a cool US$15 million each.

The actual fiendish complexities of quantum computing are, of course, beyond most but the most gifted scientists. A fully-tolerant quantum computer, complete with in-built error checking is a good few years off.

But given the speed at which we have progressed from the days when computers filled huge rooms, and “bugs” were actual bugs in the machinery, it may only be a dozen years until we can access the new generation of computing power. This will help us to solve those exponentially-based problems which are an intrinsic part of the natural world.

We already have a decent infrastructure in the form of the Internet, and classical computing power (including breakthroughs in machine learning techniques) is robust enough to act as the interface between the frankly strange quantum universe and the way we all live and do business today.

As theoretical physicist Richard Feynman wrote:

“Nature isn’t classical, dammit, and if you want to make a simulation of nature, you’d better make it quantum mechanical, and by golly it’s a wonderful problem, because it doesn’t look so easy.”