SemiQon: Ultra-Compact Silicon Quantum Processors
We’re incredibly excited at Lunar Ventures to back SemiQon, our first investment in quantum computing, alongside Voima Ventures, Tiny.vc, and angels.
Quantum computers have the potential to solve computational problems in minutes that would take a classical supercomputer thousands of years to solve. This promise has fueled billions of dollars of investments into several dozens of quantum hardware companies. However, no quantum computer has demonstrated an advantage over classical supercomputers to date: current machines are too “noisy” and make too many errors that cannot yet be corrected.
Introducing SemiQon: spun out of the Finnish research institute VTT in the spring of 2023, SemiQon is building novel quantum processors with the potential to correct errors and make quantum computing a reality. By trapping electrons within silicon at a single point — a quantum dot — they can use the spin of these electrons as so-called silicon spin qubits to perform computations.
This approach allows SemiQon’s processors to reach unprecedentedly low levels of noise, a fundamental prerequisite for fault-tolerant quantum computing. Putting both the quantum processor and the classical control electronics on the same silicon chip also allows SemiQon to make their quantum processors ultra-compact and scalable to produce.
Read more in this article about why we are excited about the potential of quantum computing and SemiQon in particular!
Why Quantum Computing is Promising and Challenging
Quantum computers offer a fundamentally new paradigm for computing compared to classical computers: Leveraging the principles of quantum physics (e.g., superposition and entanglement), they can perform computations that no classical supercomputer could complete on a reasonable timescale.
This so-called “Quantum Supremacy” has been mathematically proven for some quantum algorithms. However, to be useful in real-world applications, quantum computers not only need to handle millions of qubits (a well-known requirement that prompted a race in the industry) but also be ‘fault-tolerant,’ i.e., resistant to noise and errors.
And fault tolerance is tricky: like every quantum system, quantum computers are noisy, i.e., various factors, such as thermal noise, imperfect control of the quantum system, and environmental interference, introduce small errors in every computation step and can skew the calculation result. It has nothing to do with the fundamentally probabilistic nature of quantum systems themselves but is simply because quantum systems need to interact with their environment to receive inputs or produce outputs, or perform gate operations.
Noise introduces errors in each calculation step that cannot be corrected as of today and distorts the calculation result. Current error correction techniques are limited: NISQ computers, for example, obtain their results by repeating the same quantum computation many times, thus averaging out the noise — but this is inefficient and works only if the noise isn’t too large. However, if the noise is small enough — which is attainable for SemiQon — and the error rate drops below a certain threshold (around 1% error per calculation step [1]), it might even be possible to perform error correction. Once fault-tolerant, quantum processors may truly outcompete classical supercomputers.
But even before such an inflection point, quantum computers could still unlock significant business value by solving computational problems in the fields of quantum chemistry and optimization. For example, IBM showed that quantum computers with thousands of qubits and a decently low error rate could run hybrid quantum-classical algorithms that may already provide value — even if there’s no mathematical proof for it [2].
The market opportunity for quantum computing is massive: McKinsey estimates that $700B in value can be created through quantum computing by 2035 through developing new chemicals and pharmaceuticals, optimizing industrial processes, and portfolios for the financial industry. And the success of Intel with CPUs and PL/M, and Nvidia with GPUs and CUDA, could be replicated by a quantum hardware leader with quantum processing units (QPUs) and quantum software.
SemiQon’s Approach to Quantum Computing
SemiQon builds silicon quantum processors: By placing electrodes on a silicon chip and applying a voltage between them, the electric field that’s generated traps electrons within the silicon in a single point — called a “quantum dot”. Once confined within that quantum dot, the electrons’ spins can be leveraged as qubits for quantum computations.
Such silicon quantum processors were first proposed as a concept by Daniel Loss and David P. DiVincenzo in 1997 [3], and since then, a lot of research has been dedicated to practical implementations. Research in the past has shown that silicon quantum processors could be very promising for quantum computing and could potentially achieve an error rate lower than the 1% threshold [3]. Yet, research has also shown that a specific form of noise, so-called charge noise, is the main limiting factor for silicon quantum processors [4].
For years, both large companies like Intel, and several quantum startups, have worked on developing silicon quantum computers. Building on more than a decade of research at the Finnish research institute VTT and one recent material innovation [5], SemiQon is following a differentiated approach:
The industry standard is to use metal electrodes, but the mismatch in the crystal and electronic structure between the metal and silicon leads to noise. SemiQon innovates by using poly-silicon to apply the electrodes, which avoid this noise, thereby achieving unprecedentedly low levels of noise, suitable for error correction, and even fault tolerance.
Current quantum computers typically require a separate electronics rack and lots of wiring, which makes them very bulky. SemiQon has demonstrated that they can bring the classical control electronics (needed for providing the input to the quantum computation, reading out the result, and potentially performing error correction) on the same silicon chip as their qubits. This will allow their quantum processors to be ultra-compact and ready to handle millions of qubits.
Last but not least, SemiQon is part of a world-class quantum ecosystem in Espoo, Finland, formed by Aalto University, VTT, and the VTT-spinout IQM (which raised €128M series A2 funding in 2022). The ecosystem has attracted international players such as QuantrolOx to set up local operations, and SemiQon is positioned to benefit from this flourishing cluster.
SemiQon’s next milestone is to deliver the first quantum processors to universities and research institutes, whose feedback will inform the development of the next generation of QPUs, which will then be used by system integrators to assemble entire quantum computers.
About Lunar Ventures
Lunar Ventures is an early-stage, deep-tech venture capital firm. We invest €300K — €1M at pre-seed and seed stages in startups that bring fundamental innovations to the market. We’re a team of founders, engineers, and scientists with hands-on experience in hard engineering and R&D.
If you’re a founder working on an early-stage quantum computing moonshot, we would love to talk to you and see how Lunar could support you, especially on tech strategy, hiring, and communicating very complex ideas. Please reach out to our physicist Benjamin on LinkedIn or via email: benjamin@lunar.vc
References
[1] Fowler et al., 2012. Surface codes: Towards practical large-scale quantum computation. https://arxiv.org/abs/1208.0928
[2] Moll et al., 2017. Quantum optimization using variational algorithms on near-term quantum devices. https://arxiv.org/abs/1710.01022
[3] Loss, DiVincenzo. 1997. Quantum Computation with Quantum Dots. https://arxiv.org/abs/cond-mat/9701055
[4] Yoneda et al., 2017. A >99.9%-fidelity quantum-dot spin qubit with coherence limited by charge noise. https://arxiv.org/abs/1708.01454
[5] Kuhlmann et al., 2013. Charge noise and spin noise in a semiconductor quantum device. https://arxiv.org/abs/1301.6381
[6] Bohuslavskyi et al., 2022. Scalable on-chip multiplexing of low-noise silicon electron and hole quantum dots. https://arxiv.org/abs/2208.12131
