Quantum Computing
Quantum error correction has crossed important experimental thresholds. The Board question is how to time readiness against evidence, not vendor roadmaps alone.
What is Quantum Computing?
Quantum computing is being developed for a small set of problems where sufficiently capable, fault-tolerant machines may eventually outperform the best classical methods, including parts of molecular simulation and optimisation and, eventually, the mathematics protecting today’s public-key cryptography. Its raw components, qubits, are fragile and error-prone, so the field’s defining engineering problem is error correction: combining multiple unreliable physical qubits into a more dependable logical qubit. Recent experiments show meaningful progress. In December 2024, Google reported below-threshold surface-code error correction on Willow, and in November 2025 Quantinuum commercially launched Helios, which it reports has been used in demonstrations involving 48 logical qubits. Useful commercial scale remains unproven. The open question is no longer whether error correction can suppress errors in hardware, but whether it can be engineered at the scale, reliability, and cost that commercially useful applications require.
Why it matters to Boards
The discipline this technology asks of a Board is the separation of roadmaps from evidence. IBM and Quantinuum both name 2029 for significant fault-tolerant systems, but those dates remain company targets. The evidence is the error-correction record, and that record is moving: different architectures have demonstrated encoded qubits and increasingly capable error correction since April 2024, although logical-qubit counts are not directly comparable across machines. Three thresholds matter, and they do not fall together: error correction working experimentally, fault-tolerant machines engineered at scale, and those machines beating the best classical method on a problem worth money. Important experimental thresholds have been crossed. Large-scale fault-tolerant engineering remains a roadmap, and commercially useful advantage over the best classical method has not yet been demonstrated for the problem classes most organisations care about.
My current assessment is that pharmaceuticals, materials, and chemicals have the clearest theoretical connection through molecular simulation. Optimisation and risk applications in financial services remain plausible but more contested, because any advantage must beat exceptionally mature classical methods. For those sectors, a bounded readiness exercise is inexpensive now: identify the candidate problems, establish the best classical baseline, estimate the quantum resources each would need, and name the external milestone that would justify going deeper. For most other organisations, the appropriate posture is informed watching, with post-quantum cryptography treated as a separate and current programme.
The timeline
- Microsoft and Quantinuum demonstrate four encoded logical qubits with error rates up to 800 times lower than the compared physical circuits, with active error diagnosis and correction.
- Google announces Willow and reports below-threshold surface-code error correction, where increasing code size suppresses logical errors rather than compounding them. The results are published in Nature.
- IBM targets Starling, a fault-tolerant machine of around 200 logical qubits running 100 million gates, for 2029.
- Quantinuum commercially launches Helios, a 98-physical-qubit system it reports has been used in demonstrations involving 48 logical qubits. Logical-qubit counts remain difficult to compare across architectures.
- IBM targets Starling, a 200-logical-qubit system intended to run 100 million gates, while Quantinuum targets its universal fault-tolerant Apollo system. Both are company roadmaps rather than arrival forecasts.
Questions Boards are asking
When will quantum computers actually be useful?
In ranges, not dates. IBM and Quantinuum both currently target 2029 for significant fault-tolerant systems, and the error-correction record makes those roadmaps more credible than they once were. They remain company targets, and roadmaps in this field have slipped before. My rule is to credit each provider’s next published, testable milestone and treat the rest as ambition. Early useful workflows are likely to combine quantum and conventional computing, but a hybrid workflow is not evidence of advantage. The combined method must still beat the best classical alternative on cost, speed, accuracy, or a result that could not otherwise be obtained.
Should we be spending money on this now?
It depends on whether the organisation’s problems are among those quantum computing may eventually reward. In my current assessment, pharmaceuticals, materials, and chemicals have the clearest theoretical connection through molecular simulation. Financial-services optimisation workloads remain plausible but more contested because useful advantage must be demonstrated against highly developed classical methods. For those sectors, a bounded readiness effort is defensible: identify candidate problems, benchmark the best classical alternative, and test claims through remote access rather than capital purchase. For most other organisations, the appropriate posture is informed watching, with post-quantum cryptography treated as a separate and current programme.
Does quantum computing threaten our encryption?
Yes, and on a different clock from everything else in this Signal. Data protected by quantum-vulnerable public-key cryptography can be captured today and read when a capable machine arrives, so the migration to post-quantum cryptography is a live obligation regardless of when that happens. The standards are published and national timelines set. I cover that migration separately in the Post-Quantum Cryptography Signal, and it should be on the Board agenda whether or not you believe any vendor’s hardware roadmap.
How do we tell real progress from marketing?
Raw qubit count is not meaningful on its own. Error rates, logical operations, executable circuit depth, workload results, and comparison with the best classical method matter far more. Logical qubits are not directly comparable across architectures either, because encoding methods, error rates, and the operations they can sustain differ. The useful test for any claim is whether it was demonstrated or announced, what exactly was measured, and whether an independent party could reproduce or interrogate the result. Management should be able to apply that test to anything a supplier presents.
What should we watch over the next five years?
Watch three things: error correction improving across different architectures, systems moving from laboratory demonstrations towards repeatable engineering, and evidence that a quantum workflow can outperform the best classical alternative on a valuable problem. Vendor roadmaps matter because they create testable milestones, but they are not evidence by themselves. The Board’s job is not to predict the arrival date. It is to know which external milestone would change the organisation’s decision.
References
Quantum error correction below the surface code threshold
The peer-reviewed Willow result: logical error rates fall as qubits are added, the below-threshold milestone the field had pursued for decades.
Quantinuum Partners with Microsoft in New Phase of Reliable Quantum Computing with Breakthrough Demonstration of Reliable Logical Qubits
The April 2024 demonstration of four logical qubits with error rates 800 times lower than the underlying physical qubits.
IBM Sets the Course to Build World
The June 2025 roadmap committing to Starling, a 200-logical-qubit fault-tolerant system executing 100 million gates, by 2029.
Quantinuum Announces Commercial Launch of New Helios Quantum Computer that Offers Unprecedented Accuracy to Enable Generative Quantum AI (GenQAI)
The November 2025 launch of a commercial system delivering 48 error-corrected logical qubits from 98 physical qubits.
National quantum strategy
The UK’s ten-year strategy for a quantum-enabled economy, and the policy backdrop against which national quantum investment is being made.