The Curvature‑Stabilised Fusion Reactor (CSFR) proposes a confinement architecture that replaces the tokamak’s force‑balance, current‑driven paradigm with geometry‑driven, current‑free stability. Tokamaks rely on large plasma currents to generate poloidal magnetic fields, but those same currents create tearing modes, kink modes, neoclassical tearing modes, edge‑localised modes, and full‑scale disruptions. These are not engineering defects but structural consequences of current‑driven confinement. CSFR removes the instability source by eliminating plasma current entirely and shaping confinement through externally defined curvature fields, phase‑aligned electromagnetic channels, and field‑defined boundaries.

The conceptual anchor is Earth’s magnetosphere, which has confined energetic plasma for 4.5 billion years without plasma current, without disruptions, and without reactive feedback. Its stability arises from magnetic geometry: curvature, field‑aligned structure, and closed field‑line topology. CSFR does not replicate the magnetosphere but applies the same principle of geometry‑governed, current‑free confinement at reactor conditions.

Recent experimental results strengthen the case. Wendelstein 7‑X demonstrated record stellarator performance in 2024, including an eight‑minute high‑power discharge with no disruptions, validating current‑free confinement at reactor‑relevant parameters. Advances in coil optimisation show that W7‑X‑class confinement can be achieved with far simpler coil geometries, supporting CSFR’s use of dynamically tunable curvature rather than fixed, highly complex stellarator coils.

CSFR stability arises from three mechanisms. Curvature matching places the plasma in favourable curvature throughout the volume, suppressing interchange and ballooning modes. Phase alignment synchronises external oscillatory fields with stabilising plasma eigenmodes, reinforcing stable structures and suppressing unstable ones. Field‑defined boundaries replace material walls as the confinement surface, reducing impurity influx and eliminating edge‑instability drivers.

The MHD energy integral provides the theoretical basis. In ideal MHD, the destabilising term is proportional to parallel current. By operating at zero plasma current, CSFR removes this term entirely, leaving only stabilising curvature and magnetic‑well contributions. Density limits are no longer governed by the Greenwald scaling, which depends on plasma current; instead, density is limited by geometry‑defined stability envelopes, consistent with stellarator behaviour.

The system architecture integrates curvature‑shaping coils, dynamic curvature‑modulation coils, rotational electromagnetic confinement zones, phase‑aligned plasma channels, and a field‑defined confinement volume. All subsystems are modular and designed for staged validation. Dynamic curvature control is a key differentiator: geometry can be adjusted in real time to track evolving plasma conditions, suppress emerging instabilities, and expand the stable operating envelope.

External context highlights the structural problem CSFR addresses. Tokamak ELM suppression in 2024 required machine‑learning‑driven adaptive 3D field optimisation, illustrating the operational tax of current‑driven confinement. Independent assessments in 2025 emphasised the compounding complexity of disruption management, ELM control, tritium breeding, and materials challenges.

CSFR defines explicit falsifiability criteria across curvature mapping, mode‑selective phase alignment, disruption‑free operation, density performance, and curvature‑modulation bandwidth. A staged development pathway outlines analytical modelling, electromagnetic demonstrators, plasma‑channel experiments, integrated prototypes, and a Q>1 demonstration at an estimated programme cost of $100–150M.