IBM has boosted its growing stable of quantum computers with a new 53-quantum bit (qubit) device, the most powerful ever offered for commercial use.
Google announced a more powerful 72-qubit ‘Bristlecone’ model last year, but that was for its internal techies only. IBM’s, by contrast, feels significant because it can be used by absolutely anyone who can find a use for such a computer.
The new and still-to-be-named computer will sit in the company’s Quantum Computation Center in Poughkeepsie, New York State, which has recently turned into a hotbed for commercial development.
The facility also houses an array of older quantum computers, including five with 20 qubits (including the first Q System One launched in January), four with 5 qubits, and one with 14 qubits.
The involvement of Poughkeepsie is no coincidence – this is the heritage site where IBM built many of the mainframes that made its name synonymous with business computing.
Might quantum computers be on course to be the mainframes of the 21st century?
Lab coats
Readers will doubtless know that the qubit is a rough measure of the amount of work a quantum computer can do (read our detailed backgrounder on how quantum computers work for more on this), which loosely parallels the number of bits in a classical computer.
It’s not a perfect analogy, but what matters is that the more qubits you have, the more work you can do (IBM favours a different measure called ‘quantum volume’ which takes into account things such as connectivity and ‘gate set’ performance, algorithm errors, and the efficiency of software and compilers).
But the real significance of IBM’s new system isn’t the number of qubits it has so much as the claimed ease with which customers can get access to them.
This issue has long been a problem for quantum computers, which even today rely on physicists to run them. Then there’s the need to keep the qubit hardware mounted on a data plane at an incredibly cold temperature of -273.15 degrees Celsius, only 0.02 degrees above absolute zero.
Don’t take our word for it – IBM has an image of the sort of cooling system that makes this possible on its website as well as video of the Q System One qubit device itself.
Cloud service
Here’s the interesting part – customers don’t need to get their hands dirty with any of this because IBM’s new quantum computer will be offered as a cloud service.
This makes sense. Quantum devices are complex, specialised bits of kit that it might take organisations years and huge sums of money to master. Accessing them as a service is a simple way to benefit from their theoretical advantages now without worrying about how they work.
Despite their exotic reputation, it seems that now is finally the right moment for quantum devices in a growing number of niches, including research and development, and intriguing if rarified areas of financial engineering such as options pricing.
At some point in the future, that list might also include tamper-proof cryptography and new ways to crack cryptographic algorithms. Quantum computers’ ability to perform calculations in parallel is a potential threat to many of the cryptographic algorithms we rely on, as Paul Ducklin explains in his article Post-Quantum Cryptography (and why we’re getting it):
…many people are worried that quantum computers, if they really work as claimed and can be scaled up to have a lot more processing power and qubit memory than they do today, could successfully take on problems that we currently regard as “computationally unfeasible” to solve.
The most obvious example is cracking encryption.
…
In other words, if reliable quantum computers with a reasonable amount of memory ever become a reality – we don’t know whether that’s actually likely, or even possible, but some experts think it is – then anything encrypted with today’s strongest algorithms might suddenly become easy to crack.
In truth, the current applications for quantum computing emphasise that large companies are still very much at the experimental phase, feeling their way to understanding which sorts of problems quantum computers might be good at, and which are better left to today’s computers.
But customers must start somewhere, just as IBM must if it is to find a way to start making money from quantum computing after decades of lab testing.
Might quantum be a new paradigm that supplants today’s computers?
For that to happen, they’d have to reach ‘quantum advantage’, a hypothesized ‘superpolynomial’ moment when quantum devices start turning out answers that would not be possible when using classical computers.
The catch – those answers would almost certainly be to problems nobody has yet found a way to ask. So, don’t throw away your microprocessor laptop just yet.
Gary
I’m almost afraid to put this out there, but why not?
“The answer is 42.”
“But what’s the question?”
“For that you will need a much larger and more powerful computer…”
McChuck
Quantum computing is still a boondoggle. A technically interesting one, but boondoggle nonetheless.
There is nothing a “quantum computer” does that can’t be replicated more easily and cheaply with a standard computer connected to a reliable random number generator, like a cosmic ray detector.
At best, a quantum computer is a variant of analog computing. We could make these things hydraulic for a fraction of the cost.
John Dunn
“At best, a quantum computer is a variant of analog computing”. McChuck, Whilst in a state of superposition, a qubit could be considered to be in an ‘analogue’ state but this is really stretching the description of ‘analogue’. To claim that a quantum ‘processor’ can do no more than a classical ‘binary’ digital computer, is false.
John L. Ries
As long as I’ve been working in the industry, I’ve never seen an analog computer. Perhaps you meant “digital computing”, or perhaps “standard” or “traditional” or even “conventional” digital computing. But standards and traditions change. We’ll see if happens here in due course.
Paul Ducklin
Analog computing devices were still widely used until the 1990s or so – many car ignition systems, for example, still used centrifugal and vacuum systems to control the spark timing. Most, but not all, driveshaft differentials are still mechanical. I have a pair of vernier calipers that I bought around 2000 – today they are almost all digital but back then the digital ones were way more expensive and the vernier versions were no less accurate and looked cooler.
True digital watches are a minority item these days – the movement may be digital but the user interface rarely is, and luxury watch brands pride themselves on being analog works of art-and-engineering, especially if they have a complication – say it with a French accent – such as phase of the moon, self-adjusting calendar, chronometer or tide indicator. If you’ve ever looked through the window of a high-end jeweller’s or clockmaker’s shop, you will have seen analog computers of some distinction…
You get fuzzy but infinitely-valued outputs across the range of inputs, which is why I presume the OP made that reference in respect of quantum computation.