Diraq: Shaping the Future of Silicon Dot Quantum Computing

Diraq: Shaping the Future of Silicon Dot Quantum Computing

Quantum computers need to scale to many millions and ultimately billions of qubits to solve problems of global significance.

If you have tried to operate even a handful of qubits, you have noticed how hard that is—and how hard operating a billion qubits will be. Yet, the semiconductor industry has managed to control billions of transistors on a microchip. What if you could leverage existing semiconductor manufacturing techniques and scale to billions of transistors for quantum computing?

Diraq is a full-stack quantum computing company that aims to deliver the world’s first fault-tolerant quantum computer. It has established its core IP in the design and operation of silicon spin qubits, which are the same size as today’s transistors and compatible with CMOS foundry manufacturing.

Diraq was founded in 2022 by Andrew Dzurak, Scientia Professor of Quantum Engineering at the University of New South Wales in Sydney, Australia. In 2022, Diraq raised a $20M Series A led by Allectus Capital and this spring, it announced a Series A-2 extension led by Quantonation, bringing in an additional $15M in funding.  

Learn more about the future of silicon dot quantum computing from our interview with the CEO and founder, Andrew Dzurak:

Why Did You Start Diraq?

I got involved with quantum computing in the early days when I moved back to Australia after doing my PhD in Cambridge, UK, and became a postdoc at the University of New South Wales in the 1990s. I have a long history in research, from studying quantum effects in semiconductors to exploring nuclear spins in silicon as qubits for quantum computing. The more I dug into quantum computing, the more exciting it got, which led me to get involved with establishing the Australian National Center for Quantum Computer Technology at UNSW, founded and led from 2000-2008 by my former mentor, Professor Bob Clark. 

Over the past 25 years, it has become one of the longest-running research centers in quantum computing, helping hundreds of PhD students graduate, including Jeremy O’Brien, one of the co-founders of PsiQuantum. 

Together with Professor Andrea Morello, we demonstrated the first electron and nuclear spin qubits. But we soon figured that for quantum to have an impact, we would need to make quantum error correction work and, thus, find a way to manufacture millions, if not billions, of qubits in a scalable way. We knew how hard it was to produce quantum devices with a few qubits, so billions of qubits seemed like an enormous challenge. There needed to be some unique insight on how to make it happen. 

Yet semiconductor companies have already produced microchips with billions of transistors for classical, digital computing. Maybe there is a way to reuse semiconductor manufacturing techniques for quantum computing. Specifically, what if we could reuse existing transistors? 

We took conventional MOSFET transistors and reduced the number of electrons they contained to just one electron, and then showed that we could use these electrons as a qubit. Over the next decade, we built a ton of IP and demonstrated that we could manufacture high-fidelity qubits out of transistors, perform two-qubit logic gates, and develop a quantum computing platform compatible with semiconductor manufacturing.  

Finally, in May 2022, all of this research culminated in Diraq, which spun out of UNSW, Sydney, and raised a Series A round from Allectus Capital. We’re still working closely with the university and just opened a company lab on campus a few weeks ago. This way, we stay close to the latest advances in research while Diraq can focus as a company on leveraging semiconductor manufacturing to build a billion-qubit quantum computer. 

How Do You Build Silicon Dot Quantum Computers?

Our qubits are modified transistors, literally the standard MOSFET transistors used in 99% of all integrated circuits today. A metal gate electrode is brought on top of an insulating oxide layer, which sits on top of the silicon substrate connecting the source and the drain. Turning the gate voltage on and off controls the current flow between the source and the drain. 

What’s special about our technology is that we have found a way to reduce the number of electrons in such a transistor, even down to a single electron or a few electrons. These are confined to nanoscale regions within a silicon substrate, which are also called quantum dots. 

We found a way to encode information in the spin of the electrons with a quantum dot, make quantum dots interact with each other, and thus perform a two-qubit logic gate—one of the fundamental prerequisites needed to build a quantum computer. 

We need to combine many of these to build a powerful quantum computer: not only will we need many qubits to perform the actual quantum computations, but we’ll also need many qubits for quantum error correction. 

The big advantage of our technology is that since it’s already compatible with semiconductor manufacturing, we will be able to put a large number of qubits on a chip—even millions, which is about the order of magnitude we’ll have to reach to make fault-tolerant quantum computing a reality. 

Importantly, our qubits are much smaller at 20-30 nm than all other non-semiconductor-based quantum platforms: for example, superconducting qubits take 10s to 100s of microns. Packaging a million superconducting qubits will require a lot of space, and it will be hard to cool all of them down to cryogenic temperatures. This means scaling beyond a few 1000s of superconducting qubits while maintaining quantum entanglement across the entire chip and transferring large amounts of data between the qubits will be an enormous engineering challenge. 

Even if one could solve these challenges, the cost of operating such a giant quantum computer would also be massive, requiring thousands of dilution refrigerators and enormous costs for cooling and operating these. In comparison, our chips and the required cooling infrastructure are much smaller and cheaper, so we can envision thousands of quantum computers in a data center, augmenting classical compute. 

The major challenge for most quantum platforms is how to make many qubits interact with each other. Microchips faced the same problem when coordinating the operations of billions of transistors. Using metal gate electrodes between transistors, we’ve already demonstrated that we could use the same approach to turn the interaction between qubits on and off. Of course, interconnecting qubits will be more demanding than just standard transistors, requiring a more complex architecture. But we have developed a complete physical architecture for this, allowing for quantum error correction. 

While older semiconductor process nodes already allow for many useful things, we’re benefiting from all the advances that come with the most cutting-edge process nodes, smaller feature sizes, and more advanced manufacturing. 

How Did You Evaluate Your Startup Idea?

In the long term, our main focus is to manufacture universal, fault-tolerant quantum computers that can address all kinds of problems, and we expect to get there in the 2030s. But as early as 2028, we could see that our chips have thousands of qubits and target the first applications, e.g., in chemistry. Molecular design is certainly one of the domains where combining classical high-performance computing with quantum computing can provide early value. Once we have full error correction in place, we can address many more problems.

What Advice Would You Give Fellow Deep Tech Founders?

Focus on your end goal of leveraging technology to engineer a better product that customers need and will ultimately appreciate. Take advantage of scientific advances when they arise, but don’t get distracted by hype and focus on what your technology can do best.