Oxford Ionics: Shaping the Future of High-Performance Quantum Computing

Various technologies in quantum computing compete for supremacy. One approach that has consistently delivered high performance is trapped ions.

Ion trap quantum computing leverages the precise control of individual ions to encode, manipulate, and entangle qubits, ultimately aiming to execute quantum algorithms that have the potential to solve certain problems exponentially faster than classical computers.

Oxford Ionics takes a unique approach by trapping ions electronically instead of using lasers, enabling it to achieve record performance and making its technology compatible with semiconductor manufacturing. Founded in 2019 by Chris Ballance and Thomas Harty, it raised a $36M series A at the beginning of 2023, led by Oxford Science Enterprises and Braavos Investment Advisers and with the participation of Lansdowne PartnersProsus Ventures2xNTorch Partners, and Hermann Hauser (founder of the chip giant ARM). 

Learn more about the future of high-performance quantum computing from our interview with the co-founder and CEO, Chris Ballance: 

Why Did You Start Oxford Ionics?

Tom and I had done our PhDs in fundamental research around ion trap quantum computing at the University of Oxford. I had always found it intriguing to build stuff that gets used and solves an actual problem. And when you’re doing a startup, you need to do exactly that— build something useful in a short amount of time. 

Shortly after I finished my PhD, I gave a talk on Quantum Computing at the Royal Society and ended up sitting next to Hermann Hauser over dinner, where we discussed our shared disappointment over how many people were approaching quantum computing like a science project rather than the system integration project it should be (if you are using the right technology). Working on any single problem is a waste of time if you don’t keep the goal in mind to build a functioning quantum computer and thus solve the system integration puzzle. Hermann gave me a challenge: if I really thought I had the right way forward, I should start a company. A couple of years later, Tom and I co-founded Oxford Ionics to take on the challenge ourselves, in which Hermann invested.

How Will You Get to High-Performance Quantum Computing?

Advances in computing power are generally very valuable, and without doubt, quantum computing will be once we realize its potential. Building a quantum computer is like building a rocket to shoot for the moon: if you are not careful, you find yourself competing on metrics to build better airplanes, but they will never get you to the moon.

The main challenge to building useful quantum computers is errors. One approach is to add many more physical qubits to run error correction to produce some logical qubits. But physical qubits today are noisy and error-prone, so it doesn’t help much to add just a few more of those. You’ll need millions of physical qubits just to implement a few handfuls of logical qubits, and you’ll need about 200 logical qubits and more to address the first meaningful commercial use cases. The other approach is to understand the fundamental physics of qubits better and pick the best technology platform to eliminate error sources and build qubits with unprecedentedly low error rates. 

While large, established players have started scaling their quantum computers with noisy, error-prone qubits, they have struggled to achieve a quantum advantage or perform complex quantum algorithms. Our background in science and engineering told us that scaling something 10,000x will make it a lot worse if it’s not almost perfect from the start. So we focus on having the best possible qubits first! 

We both chose to do PhDs in trapped-ion qubits as we saw their inherent advantages over other techniques. Atoms and ions make great qubits for the same reasons they make great (atomic) clocks: their physics is easily controlled by just external magnetic and electric fields. So you get well-defined quantum properties and excellent reliability. Every atom is the same the universe over. 

Other players in trapped ion quantum computing use lasers to control ions. We take a different approach and control the ion’s quantum state electronically by applying currents to a chip. Controlling the ions as qubits electronically is trickier than using lasers, but it allows us to avoid some fundamental quantum errors that would come from using lasers. It also makes our technology compatible with established semiconductor manufacturing techniques that already produce microchips at scale.

We have developed numerous toolkits and IPs to make everything compatible with standard semiconductor manufacturing. Our chips will eventually be easier to fabricate than the average microelectronic chip at TSMC. And we demonstrated a number of world firsts, including coherence times on the order of minutes, single-qubit gate errors of less than one in a million, and world-record two-qubit gate fidelities.

We’re now tackling three work streams. First, scale quantum gates with an unprecedentedly low error rate to hundreds of qubits, reaching a quantum volume in the 100s. Second, develop the architecture of a 256-qubit quantum processor and build a prototype that we plan to make accessible through a web API. Third, build small-scale prototypes for our collaboration partners and customers to try and get feedback.

Besides having the best possible qubits, we focus on system integration and make everything from the chip to the software stack work reliably. It’s about optimizing not any single layer but everything in conjunction to unlock the orders of magnitude performance improvement we need to make quantum computing useful. Think of computers in the ’70s and ’80s, you got the best performance if you optimized compilers by hand for one particular hardware architecture. 

Many quantum startups either still deal with yesterday’s problems today or aim too far out, beyond the ten-year horizon, which is more a matter of fundamental research. We’re sitting in the sweet spot in between and want to bring our technology to the market within the next two to three years.

How Did You Evaluate Your Startup Idea?

The most important question is how you are going to generate value. Selling quantum systems is tricky in the near term as they don’t provide enough value to grow a business fast enough. Instead, everyone in quantum computing is currently working out objective metrics on how to measure progress. 

An analogy to the pharma industry, where performance metrics drive value, is very useful. Think of a quantum startup like a drug development company, where getting meaningful revenues can take a lot of time. Valuation is driven by IP, the know-how to solve a problem well, as a proxy for what value will be unlocked down the road.

What Advice Would You Give Fellow Deep Tech Founders?

First, any company is unique in its go-to-market strategy, so select investors who are aligned with your approach to generating value and entering the market. Misalignment here can be dangerous. 

Also, as a deep tech founder, you need to become a recruitment expert. It’s a much bigger worry than technical issues. As your startup grows twofold each year, you need to be two steps ahead of the game. Don’t hire someone for one or four years, but really think in 18 to 24 months timelines. Vet candidates through your network, hire for skills and move fast. The latest after two years, reevaluate whether everything checks out and the person can still solve your problems another three years down the road.