There has never been a better time to be in the quantum computing business. “Some 10 years ago, it was not obvious that quantum computing was more than an interesting lab experiment. Since then, an entire globalised ecosystem has emerged,” says at French quantum computing start-up Alice & Bob, one of hundreds of firms in the sector. at Microsoft puts it even more succinctly: “Quantum computers are working.”
But working on what? Practical uses for quantum computers remain limited, with no sign yet of the long-promised ability to solve problems that can’t be tackled by traditional computers. To deliver on that promise, they need to become large enough to run complicated calculations and they must make fewer errors to ensure that those calculations are meaningful. The two issues also compound: adding more qubits, which are the building blocks of any quantum computer, to a device typically introduces more errors.
To circumvent this, researchers have begun to group physical qubits into “logical qubits” that can catch and fix errors as they happen. “You need to be able to detect errors, correct errors and do all that during computation,” says Svore. The basic path towards truly useful quantum computers, then, is to build a device with as many logical qubits as possible. But in reality, rather than there being a single route, competitors around the world are taking very different approaches in the hope of taking the grand prize.
Advertisement
By one measure, California-based start-up Atom Computing is ahead of the pack. It holds the record for the largest quantum computer yet, with 1180 qubits made from extremely cold ytterbium atoms, also known as “neutral atoms” because they have no electric charge. Another start-up, , has assembling 1110 atoms in its quantum computing processors, without using them for calculations, while a research team at the University of Science and Technology of China in Hefei has showed that artificial intelligence could be used to make such assembly faster and easier.
“There has been amazing progress. We got over the hump of ‘Can you build a system at all?’ to ‘Can you build it better?’,” says at Atom Computing. “I think neutral atoms are in the lead.”
But sheer numbers may not be enough. “Building useful quantum computers involves work beyond building better qubits,” says at Nvidia, which isn’t building its own quantum computing hardware but is partnering with several companies to develop the best ways to use them. Other stalwarts of the traditional computing industry have had the same idea: last year, Microsoft worked with Atom Computing to create 24 logical qubits linked through quantum entanglement, a necessary first step on the path to useful devices.
That wasn’t enough to beat Boston-based neutral atom start-up QuEra, which has demonstrated more than 40 logical qubits. But QuEra also can’t claim the logical qubit crown – that belongs to a start-up called Quantinuum, which has now created and entangled 50 logical qubits, taking the lead. Later this year, the firm will launch a quantum computer that will be able to encode a trillion times more information than its already record-breaking machine, says Quantinuum’s .
Flexibility is king
Quantinuum takes a different approach to its qubits, using charged ytterbium ions held in place by electromagnetic fields, rather than neutral atoms. These “trapped ion” qubits are also being pursued by companies like Oxford Ionics and Maryland-based IonQ. One advantage both types of hardware share is the ease of switching up the connections between qubits, says at IonQ, making them more amenable to faithfully executing many different algorithms, including different ways of connecting physical qubits into logical qubits for error correction. “Flexibility and versatility is king right now,” he says.
This flexibility is part of the reason why trapped ion and neutral atom backers are hoping to eventually outpace technology giants like Google and IBM. Google, in particular, looms large in the quantum computing industry, having been first to claim “quantum supremacy” – the ability to run a calculation that a conventional computer never could – in 2019. While this was later disputed, Google claimed it again in 2024 with a new chip called Willow that it said could carry out a specific computational task in 5 minutes that would take the world’s leading conventional supercomputer about 10 septillion years.
Both Google and IBM make their qubits from tiny superconducting circuits that have their own benefits, in that they can execute calculations more quickly than their atomic and ionic counterparts and are, at times, more reliable comparatively. With neutral atoms, some of the qubits always run the risk of transitioning out of their precisely calibrated, laser-controlled quantum states.
But these benefits may not be enough to truly pull ahead of the competition. The superconducting qubits are wired in place and only easily connect to their nearest neighbours. This makes it much more difficult to implement several of the error-correcting algorithms that have been developed more recently and to experiment with these codes further.
“Developments with new error-correction codes happened very quickly, and I’d be surprised if this was the end,” says Gamble. Bloom says he used to work with other types of quantum computers, but shifted focus to neutral atom qubits because they seemed to offer more solutions to the field’s fundamental challenges. Once the industry favourite, the superconducting approach may now be at risk of running out of road.
That isn’t to say Google’s efforts have been in vain. The company has demonstrated that adding more physical qubits to Willow’s logical qubits increases their error-correcting abilities, a vital step in making large-scale computers.
Scaling up
Meanwhile, IBM’s Condor quantum processor only has 59 qubits fewer than Atom Computing’s record-breaking machine and the firm is on track to breach 4000 qubits in 2026. To that end, IBM is developing quantum computer components that will enable it to connect existing devices into larger and more powerful machines. The company believes this will also allow them to implement more error-correction codes than direct competitors like Google.
at Rigetti Computing, which also specialises in superconducting qubits, says that, in his view, superconducting quantum computers haven’t hit a dead end yet. There is value in this type of quantum computer “as we speak”, he says. Rigetti Computing sells a ready-to-ship quantum computer with 9 qubits, along with access to a larger 84-qubit quantum processor, and Rivas says the firm has sold quantum computers to both governmental labs and commercial businesses, mostly for further explorations of the technology.
Alice & Bob also makes its qubits from superconducting components, but its basic design differs from others by prioritising the suppression of errors even before creating logical qubits. Because of this, Alice & Bob researchers believe they could reach fully error-free quantum computing with thousands of qubits where their competitors may have to build millions. They haven’t demonstrated any logical qubits yet, but aim to have a truly useful quantum computer by 2030.
Similar five-year timelines loom large for many competitors in the race towards building a useful quantum computer, but California-based quantum start-up PsiQuantum has the most ambitious plan among them. The firm has foregone demonstrations and experiments with several-qubit devices and is aiming to present a large-scale, supercomputer-like quantum computer in 2027. Its qubits will be made from single particles of light, or photons, and the team has focused on integrating traditionally complex components, such as lasers and lenses that control those particles, onto semiconductor chips that could be easily manufactured at the industrial scale.
“We’re in the business of setting very challenging timelines for ourselves and we have rational, evidence-based reasons to believe that that is doable,” says at PsiQuantum.
Other quantum computing companies that specialise in using photons have taken a more traditional path. In 2021, demonstrated a photonic quantum computing chip that could execute several algorithms, while another photonic quantum computing start-up, , already sells a 12-qubit machine featuring a modular design that sets it up for easier upgrades in the future.
With so many qubit platforms to choose from, who is likely to pull ahead? at the California Institute of Technology, a long-time watcher of the industry, is tentatively betting on atoms. The prospects of making many of them and being able to connect them in just the right way for your quantum algorithm makes them the most promising, he says. “I think what a [neutral atom quantum computer] might be able to do with a few times 10,000 qubits would be comparable to what, in a superconducting quantum computer, might require hundreds of thousands of qubits.”
But the best qubit may be the one you never even notice. Today’s quantum computing researchers want the engineers of the future to think of their devices as just one other great computing resource, alongside traditional supercomputing or AI, rather than exotic machines where it may still be important to understand the nitty-gritty of hardware. “My hope is that eventually, no one will ever have to think about physical qubits ever again,” says Bloom. At that point, quantum computers may not be just working, but tackling problems that could truly change the world.
The surprising uses for a quantum computer
While the quantum computing industry is still figuring out the best way to build its devices (see main story), other industries are already trying to put them to work in surprising places. For instance, Cleveland Clinic, a non-profit medical centre in Ohio, houses one of IBM’s quantum computers – the first in the world to be uniquely dedicated to healthcare research. Elsewhere in healthcare, biotechnology company Moderna, now known for its mRNA vaccines, has already used IBM quantum computers to calculate the behaviour of certain molecules that may be relevant for its drug-development process.
Banks are big quantum backers, with HSBC, JP Morgan Chase, Goldman Sachs and Wells Fargo all hosting their own quantum divisions, which are tasked with exploring quantum algorithms for price optimisation and ultra-secure transaction verification processes. In fact, in 2024, Quantinuum partnered with Mitsui & Co., one of Japan’s largest trading companies, to demonstrate a lab-scale test of quantum tokens – a bit like a quantum version of cryptocurrency.
Automotive giant BMW has also been invested in and exploring quantum computing since 2017, with an eye towards using these new computers to develop novel materials and optimise the company’s logistics. But not all of these quantum experiments work out – in 2024, for example, Chinese e-commerce and cloud-computing company Alibaba and the country’s leading internet search provider Baidu both closed their quantum computing research labs.
