14 August 2023
Solving quantum computing's wiring dilemma: How to build a 1,000-qubit computer without tens of thousands of wires
One of the biggest challenges facing the quantum industry is how to wire the chips up as we look to meaningfully scale the size, power and potential of quantum computers.
Today, small-scale quantum computers can (just about) get away with connecting qubits to one or more individual control lines. As soon as you look to scale the technology and ramp up the qubit count, however, the number of lines, and thus the number of wires and electrical interconnects to the chip needed to maintain pace quickly becomes unwieldy. To the point of impossibility.
Scaling today's proof of concepts into powerful QCs with 1,000 qubits would require on the order of 10,000 control lines. Just connecting this many signals to a chip is beyond the bleeding edge of the microelectronics industry, and that's before we get into how we get all those signals to the chip in the first place. That it’s challenging for the commercial computing industry to solve – with its decades of legacy and trillions of dollars of R&D – highlights just how significant the scale of the problem is for the nascent quantum computing industry.
It’s widely accepted that the solution to the wiring problem is integrating control components into the chips. However, it hasn’t been clear how to do this in a way that allows you to build these chips using existing fabrication technology, without taking up too much chip space, consuming too much power, needing too much bandwidth to control, or impacting the efficiency of the QC as it scales.
Wiring is a scary problem - see the 50-qubit state of the art QC from Google on the left - but if you're WISE about it the number of control lines doesn't need to increase with the number of qubits. The OI system on the right has enough control lines for ~1,000 qubits.
Today, we’re presenting a solution that solves the challenge of how to wire a quantum computer in a way that's consistent with all of these constraints. A solution that enables a 1,000-qubit QC to be run on just 200 control lines using chips and quantum systems – the infrastructure and controls around the chip – that are available today.
We call this architecture WISE (Wiring using Integrated Switching Electronics) and, although we’re of course biased, we believe it marks a major step forward in the journey towards truly scalable quantum computers.
WISE is a new wiring architecture for trapped ion QCs that solves for the power, bandwidth and footprint problems of integration and fabrication in one fell swoop.
Illustration of possible electrical wiring of a 1000-qubit chip. The QPU (right) combines an ion trap IC with integrated capacitors and switches, and requires a footprint of 8mm × 22mm (excluding interconnects). The QPU is controlled using ∼ 200 electrical inputs, delivered from ∼ 200 off-chip sources via wire bonds.
Ion traps typically use ~10 electrodes to control each ion qubit, each of which is driven by an independent DAC, which converts digital signals into the analogue voltages needed to control the ions. While DACs have been integrated into ion traps before, the amount of power, space and data they require makes incorporating many thousands into an ion trap infeasible.
In our WISE approach, we instead manage this by integrating multiplexers into chips. This allows us to share the DACs between lots of ions, which is great because multiplexers are very small components, consisting of only a few transistors, and take up very little power.
Until now, it has been an open challenge as to how to build a computer that runs efficiently when a small number of DACs are shared between many electrodes, and in a way that maintains this efficiency as it scales. In our scientific paper outlining this solution, we introduce a new idea called "dynamic electrode parallelism" which allows us to use a small number of DACs to control all the ion qubits in a scalable way using standard CMOS fabrication processes.
This means we can still move the qubits around at will – and thus get full connectivity – but in a way that lets us use one DAC for hundreds of electrodes at the same time. It’s like replacing lots of cars with a bus or public transport; it helps ease congestion and uses fewer lanes, while all the people (or in this case, ions) still get where they need to go.
If you’re interested to go deeper into how the technology works, read our paper, How to wire a 1000-qubit trapped ion quantum computer.
Integration is our business
Integrating control in a scalable way, as we've shown with WISE, was always a natural next step for us because integration and scalability are core to our mission at Oxford Ionics.
Our electronic qubit control technology allows us to replace the lasers traditionally used to control trapped ions with electronics integrated directly into silicon chips. Our relationship with top-tier foundry Infineon allows us to build this technology out at scale on a CMOS-compatible production line. WISE fits naturally within this approach, replacing tens of thousands of wires with simple circuitry integrated into a silicon chip that can be produced at scale.
The discussions around quantum computing focus a lot on qubits - ions, photons, superconducting circuits. Yes, qubits are the heart of a quantum computer, but they're also the cheap and easy part – a thousand atoms doesn’t cost a lot!
The real magic comes from how we integrate the qubits into a system of wires and chips to make a computer. The choices we make here are what ultimately determine whether today's small-scale prototypes can truly and effectively scale into tomorrow's supercomputers. It’s this process of integration, and picking the combination of technologies that scales, which sits at the heart of our approach at Oxford Ionics.