Fireside Q&A with Qblox
Amit: Jules/ Niels, tell me a little bit about each of yourselves, and about how Qblox came about?
Jules & Niels: Thanks for hosting us Amit, it's great to be able to share the Qblox story with the QT community and quantum enthusiasts alike.
We met each other ten years ago during our bachelor’s at TU Delft in the Netherlands. While each following our individual professional careers we were always intrigued by the idea of one day founding a deep-tech company, which for a while had not matured beyond late-night conversations. About four years ago, Niels’ research in quantum computing at Leonardo DiCarlo’s reputed QuTech lab changed this.
At QuTech, it was realised early that the use of general-purpose lab equipment for the control of quantum computers is not a scalable approach due to their rigidity and complexity. This leads researchers around the world to spend a large part of their time working around these issues instead of being able to focus on their actual research.
The technological concepts that were developed at QuTech to properly battle this seemed fit for commercialisation and scalable for the upcoming generations of quantum computers. This foresight ticked all the boxes for a mission with the massive impact that Jules and I were looking for. Jules joined in and Qblox conception followed soon after in 2018. This has now led to the commercialization of fully-integrated control electronics with a wide range of proprietary technologies aimed to catalyse the development of quantum computers.
Amit: What specific problems are you (and Qblox) trying to solve? How are you solving it better than anyone else?
Jules & Niels: Currently, Quantum control stacks are assembled using general-purpose electronic equipment—arbitrary wave generators (AWGs), data acquisition cards, vector sources, etc. This approach leads to bulky, expensive, and error-prone setups reminiscent of early classical computers. Currently, it takes three 19-inch electronics racks filled with over $350k worth of electronics to control just five qubits. That is more than $60k and 50cm of rack space per qubit. With commercial QCs requiring thousands of qubits, this approach is practically and economically unsustainable.
Working at one of the world's leading Quantum research institutions (QuTech), we were confronted with limitations of current control stack. We founded Qblox after deciding that it was time to redesign the Quantum control stack from the bottom up.
Instead of trying to force general-purpose electronic equipment into performing qubit control, we created an entirely new architecture tailored to the peculiar requirements of qubits. We removed all the unnecessary clutter—buttons, screens, etc. from the device design. Then, we miniaturized and integrated the electronic components. Moreover, aware of the sensitivity of qubits, we developed a new class of AWGs with noise levels 4x lower than the best alternative on the market.
The Qblox Cluster can control up to 20 qubits, combining the complex web of many ducked-taped instruments into one slim fully integrated modular instrument. A case in point, calibrating a 50-qubit device requires fine-tuning 2,000 parameters or more. The task can take up weeks or even months for the first tune-up. Qblox architecture has many ways to speed up these routines by orders of magnitudes, saving integration teams significant amounts of time and money.
Amit: As a start-up, what are some of the biggest challenges you face? Finding talent? Getting buy-in for your value proposition and concept? How are you tackling this challenge?
Jules & Niels: This is a great question, building a startup is not a simple task; we have embraced the challenge of being entrepreneurs and the journey so far has been eye-opening and exciting. Naturally, we have faced numerous challenges over the past two years.
Being a spinoff from QuTech and TU Delft, we have access to a lot of talent, which we value. A noteworthy challenge was the ability to prove that our technology does what it is supposed to. We are able to overcome this challenge with QuTech being next door as well as several testing customers around the world, who are validating our prototypes as we speak.
We have spent many late hours in the workshop to design the electronics and software. To turn an idea into a prototype is tremendous work, that can take multiple iterations. Coming from an experimental quantum lab ourselves, we know how frustrating it can be when experiments fail due to bugs or problems in the tools we use. The products are designed to continuously run within experimental quantum setups for years on end and we are putting an enormous effort in testing and verifying both the hardware and software to make sure the products are ready for this.
Furthermore, like any young company, having the right amount of funding to fuel growth is fundamental to success. So far, we have been able to fund the company with a mix of revenue, grants and loans which resulted in a steep growth. For the upcoming phases, we are reevaluating the mix of private and public funding. We see creating impact with quantum computers as a marathon rather than a sprint. So, we are careful in selecting the funding partners that align on this vision.
Amit: There is a lot of hype about Quantum Computing at the moment. What is your take on where the industry will be in 2 years, 5 years, and 10 years?
Jules & Niels: So far, quantum computers are experimental devices— having more impact in the lab than in the market. This is about to change. IBM claims that quantum computers will be market-ready in the next three to five years. Several other big tech companies— Google and Intel—have been experimenting with quantum computing for more than a decade.
As the industry matures (over the next ten years) towards the goal of building large-scale error-corrected quantum computers, the ecosystem will have to adapt. The industry has gotten to the point where the degree of complexity has increased so much that it becomes inefficient for individual players to do everything themselves.
In the early days of the semiconductor industry, companies were doing everything in-house, but over time, a rich ecosystem sprung up. IBM's quantum efforts leader Darío Gil says: “Okay, fine, at the beginning, you need to do all to integrate, but over time, it’s like, should we be in the business of doing coaxial cabling?”. Quantum computing companies are becoming integrators, and we are ready to provide them with the control stack to accelerate their development.
Amit: What do you think will be the killer app? The “email moment” for quantum computing?
Jules & Niels: The first killer apps of quantum computers will likely be in the ability to simulate nature at its smallest scale. You can think for instance of improving the efficiency of chemical processes like the production of fertilizers, an activity that is now needlessly energy consuming. But also, when it comes to the invention of new materials we can expect early applications. This will for instance lead to better solar cells.
A lot of technology is fundamentally based on quantum mechanical behaviour. Classical computers are notoriously bad at mimicking this behaviour while quantum computers are the natural choice. This leads to the often-mentioned exponential speed-up of quantum versus classical computers. Perhaps not the ‘killer’ applications that you would be hoping for but these will surely follow. Who would have predicted Snapchat when classical computers were born in the 1950s?
Amit: Tell me about some cool things about your technology.
Jules & Niels: One of the key aspects where our technology differs from alternative solutions is in how we create networks of different modules in the system which all have their individual qubit control and readout tasks. These modules can only work together properly if you guard two aspects. The first is in time management, creating almost complete predictability in the timing relationship between different qubit controllers. Even if they are meters apart they can be aligned within far less than a nanosecond.
The second is in creating a massively scalable infrastructure to share qubit measurement outcomes. We call this the quantum control highway. Many quantum applications require the control of qubit a to be dependent on the measurement of qubit b within just a few hundreds of nanoseconds. Doing this for just two qubits may sound challenging. But solving this for hundreds of qubits is one of the coolest challenges we have ever worked on.
Amit: Is your product “hardware agnostic” i.e. can it work with different types of Qubit technology (superconducting/ quantum dots/Ion traps/ photonic etc). What does that mean in practice?
Jules & Niels: Yes, while the systems mentioned are very different on the physics level, the control stacks actually have a lot in common. By building our systems up from modules, we can easily adapt the control stack to the needs at hand. By providing signals at a wide range of frequencies from ultrastable DC all the way up to the direct output of microwave frequencies, we have become a one-stop-shop for many systems. For some architectures (like trapped ions) additional lasers from third parties are required, however, interfaces to those (or other external equipment) are also provided.
Amit: What’s the “roadmap” for the next 12 months or so?
Jules & Niels: We are on a mission to create control stacks that enable first applications of quantum computing. This means controlling orders of magnitude more qubits, further reducing noise in the system and working on implementations of quantum error correction. We are doing this with a terrific team that’s rapidly expanding and fantastic partners. Exciting times ahead!
Learn more about Qblox here: https://www.qblox.com/
At Qblox we operate at the frontier of the quantum revolution. With a dedicated team of scientists, engineers and developers we are pushing quantum technology to support scientists worldwide with our scalable and low-latency qubit control equipment. We are the sole providers of quantum control stacks that can be easily integrated with our customers’ quantum computers. The stacks combine unlevelled noise performance, low-latency arbitrary control flows and can be scaled up to 100s of qubits. Our company is based in The Netherland and a spinoff of QuTech, which enables us to implement the latest scientific insights and take a position upfront in the worldwide race towards quantum advantage.
Niels Bultink, CEO Qblox: Niels has more than eight years of experience in experimental quantum computing with superconducting circuits performing the first feedback on solid-state qubits in 2012. His PhD research with Leonardo DiCarlo at TU Delft has led to more than ten scientific publications centred around fault-tolerant quantum computing with superconducting circuits. The work, that was part of the IARPA-funded QuSurf project has materialized in a complete redesign of control stacks. These stacks now enable the control of 50 qubits and beyond in a completely integrated package at Qblox. Niels is responsible for General Management, Sales, Fundraising, IP growth and HR.
Jules Van Oven, CTO Qblox: Jules holds a BSc. and MSc. degree in Physics at TU Delft where he graduated in the group of Lieven Vandersypen on the development of reflectometry setups to speed up the readout of spin qubits. After graduation, he became lead engineer of Innoseis, responsible for the development of a wireless sensor network for seismology applications. During 2016-2018, Jules rejoined QuTech as an electrical engineer. Within the lab of Leonardo DiCarlo, Jules was part of the team that developed the QuTech Waveform Generator and the Central Controller, specifically designed for fault-tolerant quantum computing. Both devices were the first in their kind leading to three patents. Jules is responsible for R&D, Operations, Logistics, and works with Niels on Sales and IP growth.