Levitated dipole fusion reactors confine plasma in the magnetic field of a superconducting coil floating inside the vacuum vessel. The geometry is elegant — the confining magnet is decoupled from the vessel walls, making the reactor accessible, maintainable, and potentially cheap to operate. But deuterium-tritium fuel produces neutrons, and neutrons damage superconducting magnets. Previous designs avoided DT altogether, accepting lower fusion yields to protect the magnet.
Simpson and colleagues (arXiv:2602.20564) solve this differently. Instead of avoiding neutron damage, they design for it. The REBCO core magnet is heavily shielded — alternating tungsten and boron carbide layers radiate 92% of absorbed heat to the first wall while attenuating neutrons sufficiently. The abstract says the core magnet REBCO coil comprises a small 'sacrificial' section of about 20 percent, with a neutron-damage-limited lifetime of about one year, after which the core magnet is quickly removed from the vacuum vessel and replaced. (I originally wrote that the innermost 20 percent is the part closest to the plasma and is replaced alone; the abstract does not say where the section is.)
The damaged core magnet is then refurbished and reused, which the authors say reduces cost; the larger semi-permanent section is not described as lasting the reactor's lifetime. The core magnet is designed to produce peak fields of 23 T with peak mechanical strain below 0.4 percent. My description of replacing only the damaged part and keeping the rest is a simplification.
The study presents high-level designs for two feasible first-of-a-kind plants, not built prototypes; the larger is predicted to produce 667 MW of fusion power and 208 MW of net electric power. Whether the economics work is the authors' claim ('theoretically low electricity prices'), and my statement that a damage-proof magnet may not be possible is my own remark.
The general principle: when a system component faces inevitable degradation from the environment it operates in, designing it as expendable and replaceable can be more effective than designing it as permanent and resistant. Accept the damage; localize it; replace the damaged part. Sacrifice enables function that protection cannot.
Correction (2026-11-05): I checked this essay against the raw arXiv abstract of 'Deuterium-Tritium Levitated Dipole Fusion Power Plants' (arXiv 2602.20564). Right: REBCO core magnet, 23 T, tungsten and boron carbide shielding radiating 92 percent of the deposited heat to the first wall, a sacrificial section of about 20 percent with a lifetime of about one year, and the 667 MW and 208 MW figures for the larger design. Corrected: 'prototype plants', the figures applying to both, the location of the sacrificial section, and replacing the section alone. The sacrifice-over-protection moral is mine. The author name 'Simpson' is unverified.