Quantum algorithms today regularly execute around a thousand steps before errors overwhelm them. Oxford Quantum Circuits CEO Gerald Mullally puts a date on when that number reaches a million — and with it, commercial advantage.

"We will hit millions of steps before an error takes over in the years 2028 to 2031," Mullally said. "That's the point where we move past the capabilities of classical compute in those narrow areas like anomaly detection. That will be our version of the ChatGPT moment, and that will be the point where the confidence will grow. We're going to hit billions, and we're going to hit trillions of quantum operations, and we'll start to then have this real commercial advantage for a corporate."

For boards, the parallel with large language models a decade ago is the point. What looked like a fledgling technology with a wide range of plausible applications reshaped entire industries. Quantum computing now occupies the same position, and the timeline to relevance may be as short as five years.

How does quantum differ from classical computing?

Classical machines encode everything in bits, which hold one of two states: 0 or 1. Quantum computers use qubits, which can be 0, 1, or a mixture of both — a state called superposition. That flexibility lets them perform many more tasks, fast enough to change how work gets done.

The hardware remains demanding. Qubits are erratic and prone to error; small fluctuations in temperature and atmospheric pressure cause malfunctions, and expensive isolation chambers are standard equipment. The analogy Mullally's industry offers: everyone might want a supercar, but the family runaround — classical computing — is often more reliable.

What is the money doing?

The field has already produced multi-billion-dollar valuations. IBM, Google and Microsoft run significant quantum divisions. Pure-play operators D-Wave Quantum, Rigetti Computing and IonQ trade at valuations in the multiple billions of dollars.

Europe is scaling up. In June, Oxford-based OQC closed the largest Series C ever raised by a quantum computing company on the continent: £260 million, or $345 million. The next stage, Mullally says, is commercial applications with a clear return on investment.

Where does quantum deliver first?

Cybersecurity and financial crime. "Some of the earliest use cases of quantum are the ability to conduct vulnerability analysis on your network, and also to be able to detect threats," Mullally said. "And that's really where the possibility of strengthening our cyber security defenses through the earliest use of quantum compute becomes possible."

Anomaly detection comes first in financial services, he added — specifically fraud detection and anti-money laundering. Classical systems struggle to identify relationships between data points; quantum changes the geometry of the problem.

"Quantum moves that data to a higher dimensional plane, like a sphere, rather than two points," Mullally said. "If it's a sphere, you can then see more easily that there's a relationship between data that you wouldn't be able to spot if it was classical compute."

Why the urgency on encryption?

The offensive side of the ledger. A scenario in which quantum computing eats through all classical encryption defenses is not outside the realms of possibility — and it could arrive within five years. By that point, every company and individual would need powerful quantum capability on their own side. Mullally frames the technology as giving businesses an "offensive capability" sufficient to head off the threat before adversaries get there first.

Can quantum police AI itself?

The near-term business case runs through the AI systems companies have already deployed. As AI agents go rogue with greater regularity, Mullally argues, quantum offers a well-armed hunter.

"I think quantum can have a profound impact on being able to identify the anomalous behavior of certain agents, and that's the rogue agent type," Mullally said.

OQC already has applications helping data centers operate more efficiently. Beyond security, the technology addresses problems classical computing handles poorly by nature: modeling human proteins and their relationship to disease, simulating climate systems, and tracking the interaction of sub-atomic particles in fluids — tasks where reality is more complicated than a series of ones and zeros.

The positioning carries weight with executives burned by AI spending that left bottom lines resolutely flat. Quantum, its advocates argue, is a different mechanism for problem solving — one closer to the complexity of the world it models. Mullally's bet: by 2028 to 2031, the error threshold hits a million steps, confidence follows, and the corporate buyers arrive.