AI-driven optimization of non-planar stellarator coils

Stellarators, originally hindered by the extreme engineering complexity of their three-dimensional, non-planar magnetic coils, are experiencing a renaissance due to advanced artificial intelligence and supercomputing capabilities. Unlike tokamaks, stellarators do not rely on an internal plasma current, theoretically eliminating the risk of violent plasma disruptions and allowing for continuous, steady-state operation.

Type One Energy recently published its Infinity Two Physics Design Basis, using advanced computational models to validate a self-consistent physics framework for a commercial pilot plant. The document builds on decades of quasi-symmetric optimization research and positions the company to move from simulation to component procurement.

Concurrently, companies like Thea Energy are circumventing traditional complexities entirely by engineering stellarators that utilize computer-controlled arrays of planar coils, vastly simplifying the manufacturing process and improving plant maintainability. The approach trades geometric purity for engineering tractability — a bet that could compress stellarator construction timelines by several years.