How do you build a quantum chip?

I’ll come right out and say it: the title of this blog is a trick question with a simple answer. You don’t want to build a quantum chip – it’s too difficult and simply a bad idea. The better question to ask yourself is how do you tap into the existing trillion-dollar semiconductor industry to make classical chips that tell the quantum story. Anyone who has found their way onto this blog could guess that I am passionate about quantum computing, but what might be less well known is that I’ve spent much of my career solving exactly this question.

The challenge of building a quantum computer 

Across the quantum computing industry, many companies have already built small quantum computers – but they tend to resemble a quirky science experiment instead of a scalable, commercial technology. The real challenge is how we take these science experiments and engineer them at scale. Now, when trying to build millions of copies of a technology, the obvious answer is to do it on a chip. The past 70 years of compute history have demonstrated that chips are the only technology that can scale up a million-fold and still keep going.

Quantum computing companies have taken a few different paths to building quantum computers using chips. Some players have achieved an incredible bit of science by creating artificial qubits inside a chip. While these superficially look like a normal computer chip, they need to be kept at extremely cold temperatures to function – we’re talking a few millionths of a degree above absolute zero. Not only is this incredibly difficult to engineer at scale, it also means that the entire chip needs to be quantum. And if Schrödinger's cat taught us anything at all, it’s that quantum systems don’t like to be super big. 

Other companies rely on individual atoms held in a vacuum, separated from everything else including the chip. Individual atoms make great qubits: they are found all over the universe and are perfectly identical, meaning you immediately get rid of any tolerancing issues. After all, there’s a reason we make atomic clocks out of atoms! However, most companies require lasers to control individual atoms, and lasers are notoriously difficult to integrate into large systems – yielding bulky, noisy, and complicated platforms that prove impossible to scale.

What this means is that, in the attempt to find a path towards powerful quantum computing, the rest of the industry has had to resort to complicated technology with no room for engineering growth. Each individual component of the quantum systems must be state-of-the-art — with specialised fabrication needs, long lead times, bespoke vendors, and an elaborate platform that has to be reinvented with every upgrade. And as anyone who has had to scale technology before will tell you, it is this complexity that will kill you.

Using boring, classical chips to make powerful quantum computers 

Oxford Ionics has taken a different approach. Rather than trying to force quantum behaviour on a classical chip, we’ve separated the quantum from the classical. 

We start with what we know we can make work reliably, every time, without fail: atoms. Now, the challenge with atoms is that they don’t like to be controlled – but once you strip off an electron, your atoms have a charge that can be leveraged to trap them in one place. We then suspend these ions above a standard thumbnail-sized silicon chip using an electric field generated by this classical chip. This gives us the key advantage of protecting the quantum information encoded in our qubits from environmental defects (giving them long coherence times), whilst making them robust to the inevitable variations of manufacturing processes. But we just said that we need lasers to control individual atoms, right? Wrong. 

We took our engineering work one step further and created a novel technology called ‘Electronic Qubit Control’. This allows us to not only trap our ions with a classical chip, but also control them. At the core of this architecture is an integrated antenna built into the silicon chip. When an oscillating current is applied to the antenna, the qubits experience oscillating magnetic fields that drive the quantum gates, allowing us to implement all quantum operations using electronics instead of lasers. This eliminates the need for noisy high-power laser beams controlled by intricate opto-mechanical setups and the fundamental problem of off-resonant laser scattering on the quantum computer’s performance. 

While most quantum control architectures struggle to cope as the number of qubits increase, Electronic Qubit Control is scale-insensitive – allowing small and large quantum computers to be controlled with equal ease. Its capabilities extend to full connectivity and arbitrary parallel control of qubits in multi-cell devices, which provides the scalability and utility required to implement powerful quantum algorithms. What we’re left with is a simple system that traps and controls our qubits through electronics integrated directly into a silicon chip produced on a CMOS-compatible production line. Put more simply, our qubits sit on top of an incredibly boring chip. But this isn’t just a hypothetical science experiment – we’ve proven it works, and this approach has yielded the highest-performing quantum platform in the world. 

Scaling through repetition, not reinvention 

Figuring out how to control our qubits with electronics on a silicon chip took the better part of a decade. But the result is that we can now take advantage of the might and power of the existing semiconductor industry, which has a long history of producing incredibly reliable classical chips. 

Crucially, we can now build more powerful quantum computers without having to re-engineer the fundamentals each time. Instead, we simply replicate the discrete unit cells which contain all of the necessary components to control our qubits. To scale our size and performance, we just copy and paste these unit cells thousands of times into a silicon chip that is produced right alongside the chips in our laptops and iPhones. For our customers, this means we can learn and iterate fast, produce more powerful QPUs at speed, and upgrade systems by simply swapping out the QPU whilst the surrounding infrastructure stays exactly the same.

Despite the assumption that building quantum computers must require complex engineering, the key to unlocking true scalability is actually simplicity. By building powerful quantum technology on a standard classical chip surrounded by simple, repeatable infrastructure, we’re inching ever closer to realising the full potential of market-catalysing quantum computing. 

• For more information on Electronic Qubit Control, read our paper on how this architecture yielded the highest single-qubit and two-qubit gate fidelities in the world. 

• To learn about how we can use this architecture to scale, read our paper on how to wire a 1,000 qubit quantum computer.