olly - Fotolia

Breaking the noise barrier: The startups developing quantum computers

Noise limits the scalability of quantum computing. Computer Weekly speaks to startups that are turning down the volume

This article can also be found in the Premium Editorial Download: Computer Weekly: On a mission to track space junk

Today is the era of noisy intermediate scale quantum (Nisq) computers. These can solve difficult problems, but they are said to be “noisy”, which means many physical qubits are required for every logical qubit that can be applied to problem-solving. This makes it hard for the industry to demonstrate a truly practical advantage that quantum computers have over classical high-performance computing (HPC) architectures.

Algorithmiq recently received $4m in seed funding to enable it to deliver what it claims are “truly noise-resilient quantum algorithms”.  The company is targeting one specific application area – drug discovery – and hopes to work with major pharmaceutical firms to develop molecular simulations that are accurate at the quantum level.

Algorithmiq says it has a unique strategy of using standard computers to “un-noise” quantum computers. The algorithms it is developing offer researchers the ability to boost the speed of chemical simulations on quantum computers by a factor of 100x compared with current industry benchmarks.

Sabrina Maniscalco, co-founder and CEO at Algorithmiq and a professor of quantum information, computing and logic at the University of Helsinki, has been studying noise in quantum computers for 20 years. “My main field of research is about extracting noise,” she said. “Quantum information is very fragile.”

In Maniscalco’s experience, full tolerance requires technological advances in manufacturing and may even require fundamental principles to be discovered because the science does not exist yet. But she said: “We can work with noisy devices. There is a lot we can do – but you have to get your hands dirty.”

Algorithmiq’s approach is about making a mindset shift. Rather than waiting for the emergence of universal fault-tolerant quantum computing, Maniscalco said: “We look for what types of algorithms we can develop with noisy [quantum] devices.”

Making the most of noise

To work with noisy devices, algorithms need to take account of quantum physics in order to model and understand what is going on in the quantum computer system.

The target application area for Algorithmiq is drug discovery. Quantum computing offers researchers the possibility to simulate molecules accurately at the quantum level, something that is not possible in classical computing, as each qubit can map onto an electron.

According to a quantum computing background paper by Microsoft, if an electron had 40 possible states, to model every “state” would have 240 configurations, as each position can either have or not have an electron. To store the quantum state of the electrons in a conventional computer memory would require more than 130GB of memory. As the number of states increases, the memory required grows exponentially.

This is one of the limitations of using a classical computing architecture for quantum chemistry simulations. According to Scientific American, quantum computers are now at the point where they can begin to model the energetics and properties of small molecules, such as lithium hydride.

Room temperature

In November 2021, a consortium led by Universal Quantum, a University of Sussex spin-out company, was awarded a £7.5m grant from Innovate UK’s Industrial Strategy Challenge Fund to build a scalable quantum computer. Its goal is to achieve a million qubit system.

Many of today’s quantum computing systems rely on supercooling to just a few degrees above absolute zero to achieve superconducting qubits. Cooling components to just above absolute zero is required to build the superconducting qubits that are encoded in a circuit. The circuit only exhibits quantum effects when supercooled, otherwise it behaves like a normal electrical circuit.

Significantly, Universal’s quantum technology, based on the principle of a trapped ion quantum computer, can operate at much more normal temperatures. Explaining why its technology does not require supercooling, co-founder and chief scientist Winfried Hensinger said: “It’s the nature of the hardware platform. The qubit is the atom that exhibits quantum effects. The ions levitate above the surface of the chip, so there is no requirement on cooling the chip in order to make a better qubit.”

Read more about quantum computing

  • Quantum computing is still at an early stage of development, but there are a handful of use cases where nothing else matches the horsepower it offers.
  • For the Finnish government, now is the time to start preparing for the day when quantum computers will have practical value.

Just as a microprocessor may run at 150W and operate at room temperature, the quantum computer that Universal Quantum is building should not require anything more than is needed in an existing server room for cooling.

The design is also more resilient to noise, which introduces errors in quantum computing. Hensinger added: “In a superconducting qubit, the circuit is on the chip, so it is much harder to isolate from the environment and so is prone to much more noise. The ion is naturally much better isolated from the environment as it just levitates above a chip.”

The key reason why Hensinger and the Universal Quantum team believe they are better placed to further the scalability of quantum computers is down to the cooling power of a fridge. According to Hensinger, the cooling needed for superconducting qubits is very difficult to scale to large numbers of qubits.

Industrial scale

Another startup, Quantum Motion, a spin-out from University College London (UCL), is looking at a way to achieve quantum computing that can be industrialised. The company is leading a three-year project, Altnaharra, funded by UK Research and Innovation’s National Quantum Technologies Programme (NQTP), which combines expertise in qubits based on superconducting circuits, trapped ions and silicon spins.

The company says it is developing fault-tolerant quantum computing architectures. John Morton, co-founder of Quantum Motion and professor of nanoelectronics at UCL, said: “To build a universal quantum computer, you need to scale to millions of qubits.”

But because companies like IBM are currently running only 127-qubit systems, the idea of universal quantum computing comprising millions of physical qubits, built using existing processes, is seen by some as a pipedream. Instead, said Morton: “We are looking at how to take a silicon chip and make it exhibit quantum properties.”

Last April, Quantum Motion and researchers at UCL were able to isolate and measure the quantum state of a single electron (the qubit) in a silicon transistor manufactured using a CMOS (complementary metal-oxide-semiconductor) technology similar to that used to make chips in computer processors. 

Rather than being at a high-tech campus or university, the company has just opened its new laboratory off London’s Caledonian Road, surrounded by a housing estate, a community park and a gym. But in this lab, it is able to lower the temperature of components to a shade above absolute zero.

James Palles-Dimmock, chief operation officer at Quantum Motion, said: “We’re working with technology that is colder than deep space and pushing the boundaries of our knowledge to turn quantum theory into reality. Our approach is to take the building blocks of computing – the silicon chip – and demonstrate that it is the most stable, reliable and scalable way of mass manufacturing quantum silicon chips.”

The discussion Computer Weekly had with these startups shows just how much effort is going into giving quantum computing a clear advantage over HPC. What is clear from these conversations is that these companies are all very different. Unlike classical computing, which has chosen the stored program architecture described by mathematician John von Neumann in the 1940s, there is unlikely to be one de-facto standard architecture for quantum computing.

Read more on Chips and processor hardware

Data Center
Data Management