VTT recently announced completion of Finland’s second quantum computer, which uses 20 superconducting qubits. The work, accomplished in partnership with IQM Quantum Computers, is another step on the roadmap to build a 50-qubit machine by the end of 2024.
The government set out with that end goal in November 2020, when it launched a project called “The Finnish quantum computer development action” with a total budget of over €20.7m. The first quantum computer, which uses five qubits, was completed in 2021. And the second one, of 20 qubits, was completed in October 2023, putting the country right on schedule to have a 50-qubit machine by the end of 2024. Moreover, given the progress, the Finnish government has upped its ambitions.
Now it wants to scale up to a 300-qubit machine – and it has increased the total budget to €70m to get there. Researchers expect to achieve quantum supremacy with the 300-qubit machine.
The new 20-qubit quantum computer is located at the same place as the five-qubit system – in Espoo, in the south of Finland at the VTT part of Micronova, the national research infrastructure for micro and nanotechnology. The 20-qubit machine is a technology demonstrator to learn more about the methods needed to scale up to 50 qubits.
Demonstrating progress and learning as you go
The two quantum computers developed so far in Finland are based on superconducting qubits. That choice of technologies is mostly because of a long tradition in research in superconductivity in the country – VTT has been working on superconducting sensors since the 1990s.
“There are so many alternatives,” says Pekka Pursula, research manager, microelectronics and quantum technologies, at VTT. “This includes ions, atoms and photonic quantum computers. Each of these has its relative advantages. But I think that currently, the superconducting platform is the most mature in terms of scaling up – including number and quality of qubits, control and so on. Having said that, it isn’t clear which technology will deliver the quantum advantage most efficiently in the long term.”
So far, VTT and IQM have learned new things at each step of the way. These learnings have allowed them to make improvements and boost their ambitious. One of the big learnings with the five-qubit machine was how to connect it to the LUMI supercomputer and how they could use a hybrid setup to hand off certain tasks from a supercomputer to a quantum device.
They also learned how to move from a 2D architecture to 3D, which is a more complicated arrangement – especially since the qubits need to be kept at very low temperatures. Keeping the qubits cold means bringing cooling elements to all the right places, which is much more difficult when components are stacked instead of being placed side by side.
“In a five-qubit QPU [quantum processing unit], all control and read-out lines to all qubits can be routed in 2D, so all qubits have access to the edge of the QPU,” explained Pekka. “In 20 qubits, this is not the case. A 3D superconducting integration process is needed to bring the control and read-out lines to all the qubits. We developed this and have now demonstrated it works.”
VTT and IQM are benchmarking their 20-qubit device against traditional simulations on supercomputers, comparing performance in solving well-known problems. They are also comparing with other quantum computers around the world. One of the biggest issues with quantum devices is the inherent noisiness. Even with sophisticated algorithms that correct errors, no two quantum computers act the same way, because each qubit has a unique behaviour and a unique error rate. Pursula said the 20-qubit machine was doing very well in that regard. “The median fidelities measured are 99.91% for single-qubit gates and 98.25% for two-qubit gates,” he said.
“We all know that IBM already has 433 qubits, and there are several others with higher qubit counts,” said Pursula. “But in quantum computing, it’s not about qubit count only. It’s about quality and speed also. In these terms, we measure up to the competition quite well. Secondly, in Europe, there are not many 20-qubit devices yet. We expect that next year, our 50-qubit device will be one of the largest on the continent.”
Aiming for the number three position in the world
“Quantum supremacy was demonstrated by Google with 53 qubits, but there is a lot of debate about that, so I will not promise supremacy with our 50-qubit machine,” said Pursula.
But once Finland has a 300-qubit quantum computer, it will probably be able to solve useful problems. Researchers expect to use it to solve problems in materials science, performing molecular simulations much faster than can be done on a traditional computer. They also hope to also use it to solve optimisation problems, using the hybrid computing approach to have the supercomputer hand off tasks to the quantum computer. “Whether quantum utility can be accomplished with hundreds, thousands, or any other number of qubits, the hybrid approach will be needed,” said Pursula.
What niche can Finland reasonably expect to occupy in the bigger ecosystem as quantum computer becomes a reality? Only time will tell. But according to Pursula, Finland’s goal is to be among the world’s top three countries in the field.
“It’s still early times for the field,” he says. “And let’s not forget that small countries can produce very big players, such as Nokia. The limited resources that are available just need to be focused on the best opportunities. And here I see an excellent opportunity that fits Finland’s long science background in superconductors, with the innovative mindset and can-do attitude of its people.”