Three startups have announced error-correction breakthroughs that could accelerate enterprise adoption of quantum computing.
Quantum computing is still in its infancy, easily beaten by traditional computers. One of the biggest challenges? The fact that quantum bits — qubits — are much more fragile than the bits in silicon computers, so a lot more redundancy is required. In fact, today’s quantum computers require thousands or even tens of thousands of qubits in order to create one, usable, functional, logical qubit.
The solution? Error correction.
“Error correction is vital for enterprise users of quantum computing,” says Yoram Avidan, CTO of Citigroup’s Innovation Lab and global head of Citi Accelerator.
Error correction ensures the accuracy and reliability of quantum computations, he says, which is particularly important for financial monitoring. “Error correction capabilities are crucial for enabling stable, predictable, and accurate quantum-based solutions, especially in the context of financial applications we are running in the bank,” Avidan says. “Error correction will definitely accelerate the adoption of quantum-based solutions in the bank.”
But while Citigroup waits for that to happen, it’s already experimenting with the technology that’s available. Citi is using Amazon Braket, a cloud-based service, to see how well quantum computers could handle portfolio optimization tasks. Amazon Braket supports a number of physical quantum computers, including computers from IonQ, Rigetti, Oxford Quantum Circuits, and QuEra.
It also offers simulators. The State Vector Simulator, for example, simulates quantum circuits of up to 34 qubits without any noise.
To adapt its portfolio optimization code to a quantum platform, Citigroup partnered with Classiq, a quantum computing software platform startup based in Israel. That gives Citigroup a headstart on the competition for the time when quantum computers finally scale up to useful size. And that day might be closer than expected, since several error-correction breakthroughs have been made in recent months, and some of them are about to hit the market.
Nord Quantique tries bouncing photons
Canadian startup Nord Quantique announced its breakthrough on February 8. According to the company, it was successfully able to demonstrate a way to improve the reliability of a single physical qubit by 14%, thus reducing the number of total qubits required for one logical qubit.
Company president and CTO Julien Camirand Lemyre says that they will have a full quantum computing system in two years and have a product that customers can buy in 2028. “We’re probably going to connect through the cloud platforms and also offer on-premise usage for some of our customers,” he says.
In addition to Amazon Braket, Google and Microsoft Azure also offer cloud access to quantum computers.
Nord Quantique’s quantum computer will have at least 100 logical qubits when it launches, Lemyre says. By comparison, today’s biggest quantum computer is IBM’s Quantum Condor, which has 1,121 qubits and sits in a 10-foot-tall and 6-foot-wide super-fridge. It’s unclear, however, how Nord Quantique’s logical qubits compare to Condor’s 1,121 physical qubits.
Typically, it can take 1,000 physical qubits — or more — to add up to a single usable logical qubit, but IBM is also experimenting with its own approach to error correction.
Nord Quantique claims that its error correction is more effective than some of the alternatives. Another advantage is the speed, says Lemyre. The company claims its system will operate with clock speeds at megahertz frequency, between 100 and 1,000 times faster than some competing systems.
Baptiste Royer, the University of Sherbrooke professor who invented the error-correcting method that Nord Quantique is using, says that it comes down to attaching a little container full of photons to a physical qubit.
The container is a piece of aluminum about the size of a walnut with the inside hollowed and polished to a mirror shine. Then a beam of light – or, more exactly, a bunch of photons in the invisible microwave spectrum – is bounced around inside. The photons are linked to the physical qubit, providing redundancy. Since photons are pretty small – in fact, they have zero mass – you can fit a lot of them in there, says Royer. Each additional photon offers more redundancy, and thus translates to better error correction, but, eventually, the benefits taper off, he says.
According to Royer, this approach fits particularly well with quantum computers based on superconducting circuits but may, in theory, be applied to other types of quantum computers as well. And there are lots of them, including quantum computers built on top of trapped ions, quantum dots, photons, and neutral atoms.
QuEra eyes magic state distillation
One of Nord Quantique’s competitors – one that already has a quantum computer on the market – is QuEra. It uses neutral atoms instead of superconducting circuits, and it, too, recently announced a breakthrough in error correction. “Some of our experiments took only eight physical qubits to make a logical qubit,” says Yuval Boger, the company’s CMO.
Then, in January, the company released a roadmap about when their new error-corrected computers will be available to the public. According to QuEra, it will offer a commercial quantum computer this year with ten logical qubits using its new error correction mechanism, based on transversal gates. These gates prevent error propagation between qubits, which reduces error rates.
In 2025, it expects to have a new kind of error correction mechanism, magic state distillation, to allow for a quantum computer with 30 logical qubits. Magic state distillation allows for a greater variety of gates, which is a crucial step towards building universal quantum computers instead of today’s special-purpose ones.
And, in 2026, the company expects to release a 100-logical-qubit computer, which will push quantum computers beyond the limits of what classical computers can do.
The downside? QuEra’s computers are slower than the alternatives. “We don’t want to intermingle error correction with speed,” says Boger. “Speed is a different matter, because different technologies run at different speeds.
QuEra has had a quantum computer available on Amazon Braket since November of 2022, he says.
It can take a few months to get a computer from the lab to production, says Boger, so some early customers will be able to get it via the cloud this year. “The waitlist is already open,” he says. It will be available to everyone next year, he adds.
Customers can also order an on-premises version of the computer. “If they place an order today, they can get it in 2025,” he says. “It takes us a year or a year and a half to deliver because they’re made to order.”
These won’t be world-changing computers, he says. “If you want to simulate a quantum circuit of 10 qubits, you could do it on your cell phone.” But early customers will be able to start learning how error correction works, how to write quantum software, and start testing algorithms.
Alice & Bob devise cat qubits
A third quantum computing startup, Alice & Bob, announced their new quantum error correction architecture in January.
Alice & Bob is also using the bouncing photons trick, says CEO Théau Peronnin. “But instead of bounding around a mirrored ball, it’s bouncing around an electronic circuit,” he says. IBM and Google are also experimenting with this, he adds, but they only use single photons.
Alice & Bob, like Nord Quantique, uses multiple photons. It also uses software, namely low density parity check code. IBM is experimenting with this approach in its quantum computers, as are other companies.
But Alice & Bob’s biggest breakthrough in error correction comes from cat qubits – named after Schroedinger’s Cat – which reduce the number of dimensions that noise comes from.
Normally, says Peronnin, qubits can switch from, say, zero to one – just like regular bits – when they’re not supposed to. They can also shift phase. Alice & Bob’s cat qubit error correction technique almost completely eliminates the possibility of switching from a zero to a one. “By doing that we’re slightly degrading the performance on phase shift, but only a small amount,” he says.
This reduces the total number of physical qubits required to make one logical qubit by a factor of 200. So, for example, it would take 20 million qubits for Google’s state-of-the-art computer to run the encryption-breaking Shor’s Algorithm, he says. “With us, it would take slightly fewer than 100,000 quantum bits.”
Plus, this approach is faster than what QuEra offers, he adds. “What QuEra has done is absolutely beautiful but, at the end of the day, we’re building a computer,” he says. “And speed matters.”
Alice & Bob doesn’t have a commercially available computer out yet, but it will be up on one of the big three cloud providers in the next couple of months, he says. “Currently, we only serve special clients on-premise,” he says.
Which error correction is best?
The best strategy, says Sam Lucero, chief quantum analyst at Omdia, would be to combine multiple approaches to get the error rates down even further.
For example, with Nord Quantique’s current error-correcting system, instead of needing 1,000 physical qubits for one logical qubit, they might, say, only need 100. “That’s very substantial,” Lucero says.
But adding error-correcting coding can reduce the total physical qubits needed even further, he adds. “There are dozens of these codes,” he says. “This type of code seems to allow for even more reduction in the number of physical qubits needed to make one logical qubit based on the coding efficiency.”
The bigger question is which type of qubit is going to become the standard – if any. “Different types of qubits might be better for different types of computations,” he says.
This is where early testing can come in. High-performance computing centers can already buy quantum computers, and anyone with a cloud account can access one online.
Using quantum computers via a cloud connection is much cheaper and quicker. Plus, it gives enterprises more flexibility, says Lucero. “You can sign on and say, ‘I want to use IonQ’s trapped ions. And, for my next project, I want to use Regetti, and for this other project, I want to use another computer.’”
But stand-alone quantum computers aren’t necessarily the best path forward for the long term, he adds. “If you’ve got a high-performance computing capability, it will have GPUs for one type of computing, quantum processing units for another type of computing, CPUs for another type of computing – and it’s going to be transparent to the end user,” he says. “The system will automatically parcel it out to the appropriate type of processor.”
