The Finnish Ministry of Economic Affairs recently funded an innovation project for VTT Technical Research Centre of Finland to build the country’s first quantum computer.
VTT enlisted IQM, a homegrown startup, to help with the project, which began at the end of 2020 and will continue until 2024.
Owned by the Finnish state, VTT is one of Europe’s leading research institutions. It plays the crucial role of taking what researchers learn in a range of scientific domains and making it ready for industry. The government firmly believes the best way to ready quantum computing for industry is to build a working quantum computer.
“When it comes to quantum technology, Finland has one of those unique opportunities where a small country has a whole value chain in place,” says Himadri Majumdar, programme manager for the Quantum Initiative at VTT. “Other countries also have strong ecosystems in quantum technologies, but in almost all cases they are working on a lot of different topics and many different platforms. Finnish researchers focus almost exclusively on the superconducting qubit approach, which they have been using for years and know very well.”
This will not be the first time Finland took quantum technology from research to industrialisation. They already did so for quantum sensors. Finnish spin-off companies have been producing sensors based on quantum technology since the 1980s and the 1990s, in the form of superconducting quantum interference devices (Squids), which were commercialised as essential components in brain imaging systems. Finnish startups also commercialised terahertz spectroscopy and terahertz imaging – quantum technologies used in space applications and in scanners at airports.
The country is now well positioned to play a significant role in the next generation of quantum devices and sensors – for example, atomic clocks scaled down to small dimensions and used in consumer devices. Given the success Finland has had with other quantum technologies, the government is hoping to get ahead of the curve on quantum computers.
“Now is the right time for us to lay the groundwork for bringing quantum computing to industry,” says Majumdar. “At the end of last year, we built a five-qubit computer. The ultimate measure of success is to run a programme on it and benchmark the results. We are developing the software stacks we will need to do this in early 2022.”
We don’t expect to solve any practical problems with five qubits. But the device can serve as an excellent proof of concept. The project team will then expand the computing capacity with 20 qubits in 2022 – and then with 50 qubits by the end of 2024, when they hope to solve real problems.
“We think the 2020s is a crucial decade for building the fundamentals,” says Majumdar. “This is when the race for making a higher number of qubits is happening. There will be two parallel paths. The first one is the one we have already started: building a computer with a large number of NISQ [noisy intermediate-scale quantum] qubits. The second path, which will also be taken during this decade, is to find ways of building pure qubits – that is, qubits that are not noisy and do not need error correction.”
Growing ecosystems in Finland
To assist in the project of building a quantum computer, VTT chose IQM, a Finnish startup that was founded in 2019 and now has 140 employees. “We act as a systems integrator,” says Jan Goetz, CEO and co-founder of IQM. “Our job is to take the different pieces and build quantum computing systems.”
One of the pieces they use is the cryogenic system from Finnish company Bluefors, which grew out of Finland’s long history of research in cold temperature physics. Founded in 2008, Bluefors eventually found a niche in quantum computing and is now the world’s leading provider of the cryogenic enclosures used to keep superconducting qubits at temperatures very close to absolute zero.
“Since we built Finland’s first quantum computer this year, we have seen a few other startups emerge,” says Goetz. “Algorithmiq is one of them, and Quanscient is another one that just very recently formed. On top of that, several companies from outside of Finland have seen an opportunity here and are now part of the local ecosystem. With this combination of homegrown startups and the local subsidiaries of foreign firms, we now have a nice ecosystem of organisations forming around quantum computing.”
While virtually all industrialised nations in the world recognise quantum computing as a strategic technology, Finland is particularly well positioned to embrace the new paradigm. The government is hoping to increase the advantage through investment – and some of the local companies and research organisations are also benefiting from EU initiatives, as well as the venture capital that is now flowing into Finland to cash in on the country’s skill set.
Research and educational ecosystems are also sprouting up, with plans to hire more scientists and professors. VTT, Aalto University and Helsinki University are founding members of a research community called InstituteQ, which focuses on developing world-class quantum expertise and helping business make use of quantum computing.
The Finns are acutely aware that Finland can never be a Silicon Valley. The economy just isn’t big enough. Finnish startups therefore know from the beginning that they must ready their products and services for export – and this is what makes homegrown Finnish companies so strong on the world market.
“As for IQM, we want to be the main supplier for supercomputing centres and for companies that can afford their own quantum computers,” says Goetz. “As systems integrator, we deliver a full system. But the system, of course, will contain more than just IQM parts.
“We built the heart ourselves, which is the quantum processor, and then a little bit of the control electronics and part of the software. The software is best described as a firmware stack, but all the rest we just assemble,” he says. “We buy the cryogenics from Bluefors, we buy cables, and we buy amplifiers. Then we bring it all together.”
IQM manufactured the qubits for the five-qubit prototype and will continue up to the 50-qubit computer, which is expected to be a working system that can solve real problems. IQM has its own fabrication line, which it uses to build the processor, starting with bare silicon wafers. They also use the Otanano national infrastructure, which features the largest R&D cleanroom in the Nordic countries and is jointly run by VTT and Aalto University.
A new usage model will one day arise
One good way of illustrating how quantum computers might be used is to consider how Google Maps finds the best path. This is a very compute-intensive problem. If you request this on your smartphone, it’s not your smartphone that calculates the path. Your smartphone only communicates the problem to a server somewhere in a datacentre. The path is calculated on some powerful computer and the answer is transmitted back to your phone.
Quantum computing services will probably be offered to consumers in this way in the future, with most users completely unaware of what is involved. Quantum computing will also help companies with R&D using a similar model. Companies that want to find new materials can request modelling and simulation services, and some parts of those services will be performed by a quantum computer in the cloud; others will be performed by a classical computer.
IBM and other companies already offer quantum computing services on the cloud. But those services are used by researchers and are often limited to simulating quantum computing. Researchers can test algorithms on the simulators – and those who have a few qubits themselves can compare the results of the simulator with what they get on their prototype quantum computer.
It’s not yet clear how a practical system will offer services to application developers and end users. One approach is to have specific libraries – for example, a chemistry library that can be used to simulate new molecules. Application developers need only access these libraries to develop a solution that will help companies with R&D. At run time, the library transfers the work to a supercomputing centre that does the work. When the supercomputing centre gets a task, it separates the parts that go to the quantum computer from those that can be better performed on a classical computer. To do this, it will need a scheduler.
“Something very similar is already occurring for AI algorithms,” says Goetz. “People use GPU [graphical processing units] to accelerate CPU clusters. Certain problems run very well on GPUs, but not well at all on CPUs. These problems are separated and assigned to the appropriate processing units.
“To have the libraries, of course, you need to have the algorithms and the compilers in-between, and that’s a tricky topic right now,” he says. “We’re not yet at the point where we have a large-scale universal quantum computer where you just have one type of compiler that compiles everything for a standard architecture.”
Quantum computers are far from generic. Writing a program requires knowledge of the architecture of a given device – including the quality of the qubits and the distances between them. Coherence and fidelity are the most important factors to consider.
“Let’s say on the processor you have a few bad qubits,” says Goetz. “You want to avoid them in your calculation and let them only do very minor tasks. In the future, maybe we will have a system of feedback between the processor and the actual compiler, so the compiler can generate programs that fit the computer. But for now, we’re still in this phase where people really need to get their hands dirty and map the two worlds together.
“To help developers, we are building a kind of firmware that will provide standard software interfaces,” he says. “Right now, we’re integrating into Google Cirq, IBM Qiskit and Atos QLM [Quantum Learning Machine]. These are the three main software layers on top. Anybody with software that runs on top of those layers will be able to run on our machines.”
The first practical applications of quantum computing
As part of the project funded by the Finnish Ministry of Economic Affairs, a separate team in VTT, the quantum algorithm team, is developing algorithms to be used on the quantum computer. Materials modelling is one example of an application area they are working on. VTT intends to take a few such examples to test the algorithms on the five-qubit systems and compare the results with a simulation.
Like many other organisations trying to build a practical quantum computer, VTT is looking at two broad types of applications. The first is to solve complex optimisation problems that exist in many industries – problem domains, such as energy distribution, process control and fleet management. The second is to predict the structures and properties of molecular formations much more accurately and effectively than before, accelerating drug discovery and the development of new materials.
“Nobody knows whether the first practical applications of quantum computing will be in finance, in medicine, materials science or some other area,” says Majumdar. “But one thing that’s for sure is that it will evolve very quickly.
“A trend we are already starting to see is buyers and end users of the technology (BMW, Goldman Sachs and others) tend to create a triangle of companies, consisting of a hardware company, a software company and themselves as users. This triangle develops a highly customised solution around a specific use case. This trend will continue for quite a few years because quantum computers are very specific and machine agnostic algorithms are a long way off. Everything will be highly tailored in the beginning.”
While there are still a lot of unknowns, one thing is clear: by building a local ecosystem that exports products and expertise, Finland stands a good chance of becoming a part of the European answer to the quantum computing technology coming out of the US and China.