(Bloomberg) -- Four decades ago, physicists were theorizing that the mind-bending mechanics of quantum physics could be harnessed to make a new kind of computer that’s exponentially more powerful than conventional machines. A series of breakthroughs has now brought “quantum utility” closer to reality. A race is on to develop machines that are accurate enough to faithfully model the behavior of complex real-world phenomena and deliver a leap forward in fields as varied as drug development, financial modeling and artificial intelligence.
What’s the appeal of quantum computers?
They can do things that classical computers can’t. Google revealed in December that its latest quantum processor, Willow, had solved a problem in five minutes that the world’s most powerful supercomputers wouldn’t have been able to solve even if they had been working on it since the universe began.
Experimental quantum computers are typically given tasks that would confound a conventional computer because there are too many variable inputs. Their greatest potential is for modeling complex systems involving large numbers of moving parts whose characteristics change as they interact with one another. They might, for example, replicate the behavior of molecules to accelerate the development of new medicines, or simulate the decisions of economic actors and financial intermediaries to make market forecasting more accurate.
Quantum computers are not expected to be of much use in the laborious but simpler work fulfilled by most of today’s computers, which process a relatively limited number of isolated inputs sequentially on a mass scale.
Who is building quantum computers?
D-Wave Quantum Inc., a Canadian-founded company with headquarters in California, became the first to sell quantum computers in 2011. International Business Machines Corp., Alphabet Inc.’s Google, Amazon Web Services and numerous startups have all created working quantum computers.
More recently, companies such as Microsoft Corp. have made progress toward building scalable and practical quantum supercomputers. Intel Corp. started shipping a silicon quantum chip to researchers with transistors, known as qubits (quantum bits), that are as much as 1 million times smaller than other qubit types.
Google and IBM, alongside startups Universal Quantum and PsiQuantum Corp., claim they will deliver a useful quantum supercomputer by the end of the decade. China is building a $10 billion National Laboratory for Quantum Information Sciences as part of a big push in the field.
How do quantum computers work?
They use tiny circuits to perform calculations, as do traditional computers. But they make these calculations in parallel, rather than sequentially, which is what makes them so fast. Regular computers process information in units called bits, which can represent one of two possible states — 0 or 1 — that correspond to whether a portion of the computer chip called a logic gate is open or closed. Before a traditional computer moves on to process the next piece of information, it must have assigned the previous piece a value.
By contrast, thanks to the “probabilistic” nature of quantum mechanics, the qubits in quantum computers don’t have to be assigned a value until the computer has finished the whole calculation. This is known as “superposition.” So whereas three bits in a conventional computer would only be able to represent one of eight possibilities – 000, 001, 010, 011, 100, 101, 110 and 111 – a quantum computer of three qubits can process all of them at the same time. A quantum computer with 4 qubits can in theory handle 16 times as much information as an equally-sized conventional computer and will keep doubling in power with every qubit that’s added. That’s why a quantum computer can process exponentially more information than a classic computer.
How does it return a result?
In designing a standard computer, engineers spend a lot of time trying to ensure that the status of each bit is independent from those of all the other bits. But qubits are entangled, meaning the properties of one depend upon the properties of the qubits around it. This is an advantage, because information can be transferred quicker between qubits as they work together to arrive at a solution. As a quantum algorithm runs, contradictory (and therefore incorrect) results from the qubits cancel each other out, whilst compatible (and therefore likely) results are amplified. This phenomenon, called coherence, allows the computer to spit out the answer it deems most likely to be correct.
How do you make a qubit?
In theory, anything exhibiting quantum mechanical properties that can be controlled could be used to make qubits. Many are made from semiconductors, with IBM, D-Wave and Google using tiny loops of superconducting wire. Some scientists have created qubits by manipulating trapped ions, pulses of photons or the spin of electrons. Many of these approaches require very specialized conditions, such as temperatures colder than those found in deep space.
How many qubits are needed?
Lots. Although qubits can process exponentially more information than classical bits, their inherently uncertain nature makes them prone to error. Mistakes creep into qubits’ calculations when they fall out of coherence with one other. Theorists are working to develop algorithms that can correct some of these errors. But an inevitable part of the fix is adding more qubits.
Scientists estimate that a computer needs millions – if not billions – of qubits to reliably run programs suited for commercial use. The current record for qubits connected is 1,180, achieved by California startup Atom Computing in October 2023 — more than double the previous record of 433, set by IBM in November 2022. Sticking enough of them together is the main challenge. As a computer gets larger in size, it emits more heat, which makes it more likely that qubits will fall out of coherence. Google’s Willow chip was seen as a breakthrough as the error rate fell even as more qubits were grouped together.
When do I get my quantum computer?
It depends on what you want to use it for. Academics are already solving problems on 100-strong qubit machines through the cloud-based IBM Quantum Platform, which the general public is able to try out (if you know how to develop quantum code). Scientists aim to deliver a so-called “universal” quantum computer suitable for commercial applications within the next decade.
One potential downside of the enormous problem-solving power of quantum computers is how easily they might crack classical encryption systems. Perhaps the best indication of just how close we are to widespread quantum computing is that governments are signing directives and businesses are pouring millions of dollars into securing legacy computing systems against being cracked by quantum machines.
(Updates to include location of D-Wave Quantum headquarters in second question)
©2024 Bloomberg L.P.