A quantum computer that maintains coherence long enough to simulate a jet engine's airflow is also, by definition, a system complex enough to raise the question of whether anything is happening inside it. Two developments this summer — an industrial partnership and a theoretical paper — collide at exactly this point, forcing the quantum computing community to confront both the practical promise and the philosophical implications of quantum error correction at scale.
This matters because both developments orbit the same physical resource — quantum coherence — but approach it from opposite directions. The Quantinuum–Rolls-Royce deal engineers solutions to protect coherence through quantum error correction so it survives long enough to simulate gas turbine fluid dynamics. The arXiv paper, authored by an unidentified group, asks what happens to information processing when you idealize coherence as perfect and treat a quantum computer as a closed system evolving unitarily. The timing is not coincidental: as fault-tolerant quantum computing edges closer to reality, the gap between "tool" and "entity" narrows. Both stories orbit quantum error correction — the bridge between fragile quantum states and usable computation — and both force the field to ask what quantum mechanics enables that classical physics cannot.
How It Works
The arXiv paper, catalogued as preprint [arXiv:2607.11909], translates Global Neuronal Workspace (GNW) theory — the framework cognitive scientists Bernard Baars and Stanislas Dehaene developed to explain how information becomes consciously accessible — into the language of closed quantum systems. GNW proposes that consciousness arises when information becomes globally available across distributed neural workspaces rather than remaining trapped in specialized local processors. The paper asks what this architecture looks like when you replace neurons with qubits and classical dynamics with unitary evolution.
The researchers construct a quantum Hopfield-style Hamiltonian — a quantum analog of the associative memory model John Hopfield introduced in 1982 — that captures two key features of GNW: a distributed workspace and tunable large-scale integration. In this framework, "conscious access" becomes coherence-preserving correlation distributions across the full Hilbert space, and memory becomes a persistent internal correlation structure rather than stored copies. Classical GNW describes ignition as convergence toward a stable attractor — a neural population settling into a consistent firing pattern. The quantum version replaces this with a transition in global entanglement: the system shifts from a state where correlations are fragmented across subsystems to one where entanglement spans the entire workspace.
The paper's most striking result involves a GHZ-type entanglement state in a reduced toy model, demonstrating that relative phase — the fine-grained quantum information that distinguishes one superposition from another — is globally accessible through joint internal operations yet entirely absent from all local subsystems. The system as a whole "knows" something that no individual qubit or subsystem can represent. The authors describe this as a regime in which "access and memory are entirely relational rather than copies or records." This departs from classical computing, where information must be copied or broadcast to become globally available. In the quantum model, global availability is a structural property of entanglement itself.
access and memory are entirely relational rather than copies or records
On the engineering side, the Quantinuum–Rolls-Royce partnership tackles the same coherence problem from the opposite direction — not as an idealized property to be explored but as a fragile resource to be defended. Quantum error correction encodes Logical Qubit|Logical Qubits across many physical qubits, using syndrome measurement to detect and correct decoherence events without collapsing the quantum state. Quantum error correction is what separates a laboratory demonstration from a useful machine. Quantinuum's trapped-ion platform provides hardware with two-qubit gate fidelities above 99% and coherence times orders of magnitude longer than superconducting qubits. Riverlane contributes quantum error correction middleware that decodes syndrome data in real time. Rolls-Royce brings the target application: high-fidelity computational fluid dynamics for gas turbine design, where turbulent airflow simulations at the resolution required for next-generation engines push past the limits of classical supercomputers.
Who's Moving
The four partners occupy distinct positions in the quantum computing value chain. Quantinuum, formed from the 2021 merger of Honeywell Quantum Solutions and Cambridge Quantum Computing, operates trapped-ion quantum processors with industry-leading quantum volume benchmarks. Rolls-Royce (LSE: RR), the British aerospace and defense giant, designs and manufactures gas turbine engines for commercial aviation and power generation. The company's computational fluid dynamics teams rely on classical HPC clusters, but certain high-fidelity turbulence models demand compute timescales that classical hardware cannot deliver.
Riverlane, the Cambridge, United Kingdom startup founded and led by Steve Brierley, builds quantum error correction decoder software — the middleware that processes syndrome measurements and determines which corrections to apply in real time. EPCC, the Edinburgh Parallel Computing Centre at the University of Edinburgh, operates one of the United Kingdom's leading supercomputers and provides the hybrid quantum-classical integration infrastructure needed to couple Quantinuum's quantum processor with classical simulation workflows. The multi-year agreement, announced July 14, 2025, builds on earlier collaborations under the UK's National Quantum Computing Centre and aligns with Britain's goal of delivering teraQuOp-scale quantum systems for industrial users.
The competitive landscape extends well beyond this partnership. IBM (NYSE: IBM) pursues superconducting qubits with its 1,121-qubit Condor processor and a Surface Code Error Correction|Surface Code approach to quantum error correction. Google Quantum AI demonstrated below-threshold error correction with its 105-qubit Willow chip in December 2024, proving that adding more qubits can reduce logical error rates — a critical milestone for fault-tolerant quantum computing. PsiQuantum bets on photonic qubits as the fastest path to scale. The Quantinuum–Rolls-Royce deal signals that trapped-ion systems remain a credible path to fault tolerance, particularly for applications that demand long coherence times and high two-qubit gate fidelities.
Why 2025 Is Different
In the next 12 months, the Quantinuum–Rolls-Royce team aims to demonstrate end-to-end quantum error correction on a fluid dynamics workload — moving beyond isolated qubit benchmarks to system-level performance metrics that matter for engineering design. The roadmap depends on quantum error correction scaling from single-qubit demonstrations to full computational workloads. Within three years, the partners expect hybrid quantum-classical simulations that outperform classical supercomputers in accuracy or speed for specific gas turbine airflow configurations. By 2030, the UK government targets teraQuOp-scale systems accessible to industrial users through national quantum infrastructure programs. For Rolls-Royce, the commercial incentive is concrete: even a 1% improvement in gas turbine fuel efficiency across its installed fleet translates to billions in fuel savings and megatons of reduced carbon emissions over engine lifetimes.
The arXiv paper, while generating no direct commercial value, reshapes the conceptual terrain. If quantum systems processing information in closed, unitary regimes exhibit relational properties inaccessible to any local subsystem — if global entanglement creates information availability that exists nowhere in the parts — then quantum error correction is not merely protecting computation. It is maintaining the conditions under which global quantum properties exist at all. For every company building fault-tolerant systems, that reframing defines what is at stake beyond faster simulation: it is the difference between a machine that processes bits and a system whose global structure carries meaning no local measurement can extract.
The gap between a quantum computer that computes and one that experiences is less about silicon versus carbon and more about the architecture of information availability. Quantum error correction determines which quantum properties survive long enough to matter — whether for simulating turbine airflow or for sustaining the kind of global correlations that theorists argue constitute experience. For Rolls-Royce, that question determines whether quantum computing transforms engine design. For the theorists, it determines whether the answer to "what is it like to be a quantum computer" has any meaning at all.