Fusion energy has mostly focused on tokamak reactors, the donut-shaped machines you've probably seen in headlines. But there's another design that researchers are taking seriously: the stellarator.
Stellarators look like twisted pretzels and are notoriously difficult to engineer. The complex 3D magnetic coils required to contain the plasma make them expensive and time-consuming to build. That's why they've been overshadowed by tokamaks for decades.
But stellarators have a major advantage. They can run continuously without the plasma disruptions that plague tokamaks. Tokamaks need pulsed operation and complex control systems to maintain stability, while stellarators achieve this through their geometry alone.
The nickname 'dumb machine' refers to this simplicity in operation, not the engineering. Once you solve the incredibly hard problem of building one, it just works without constant intervention. That could mean more reliable power generation if fusion ever reaches commercial scale.
For anyone following AI's energy demands, this matters. Training large models and running inference at scale requires massive, stable power sources. Fusion promises clean baseload power without the intermittency issues of solar and wind.
Germany's Wendelstein 7-X stellarator has been producing promising results, proving the concept can work at scale. Whether stellarators or tokamaks win the fusion race, the timeline for commercial fusion power still stretches years into the future. But the engineering progress is real, and the potential payoff for energy-intensive AI workloads makes it worth watching.
