Thea Energy builds first stellarator digital twin for Helios power plant

Thea Energy’s planar coil stellarator design forms the basis of the Helios digital twin, developed in collaboration with NVIDIA, Synopsys, Argonne, and PPPL

(Image courtesy of Thea Energy)

Thea Energy has launched a multi-partner collaboration to develop a digital twin of its Helios fusion power plant – a system the company describes as the first of its kind for a stellarator design. Working with NVIDIA, Synopsys, Argonne National Laboratory, and Princeton Plasma Physics Laboratory, Thea Energy is applying AI surrogate models and physics simulation to accelerate development of the Helios design. The effort aligns with the U.S. Department of Energy‘s Genesis Mission, which focuses on leveraging AI to fast-track scientific advancement, including fusion energy development.

Stellarator digital twin combines AI and multiphysics simulation

Each partner contributes a distinct capability to the Helios digital twin. NVIDIA is integrating Thea Energy’s codes, models, and real-world data into a platform built on NVIDIA Omniverse libraries, enabling real-time analysis of power plant performance at scale. Synopsys is delivering a unified multiphysics framework using leading simulation software, focused on the breeding blanket – the component that converts fusion energy and shields the magnet systems. Argonne National Laboratory is contributing neutronics analysis and blanket design expertise, supplying datasets intended to bridge gaps in commercial blanket engineering and feed surrogate model training. PPPL is contributing high-fidelity plasma simulation codes and verified datasets to support development of a surrogate model within the digital twin platform.

“With the Helios digital twin, we can shorten development cycles and essentially run the system before we even put a shovel in the ground,” said David Gates, co-founder and CTO of Thea Energy. The approach requires “orders of magnitude less capital compared to traditional models.” The digital twin work extends Thea Energy’s use of AI-enabled software controls, which the company says support operational control and maintenance optimisation of its planar coil stellarator architecture.

AI surrogate models for plasma and blanket design analysis

PPPL’s Nate Ferraro said his lab is supplying plasma simulation codes and verified datasets to train a high-fidelity surrogate model under power-plant-relevant conditions, supporting design and analysis within the Helios digital twin. At Argonne, John Tramm said the GPU-accelerated platform “fundamentally alters the workflow for major power plant components,” enabling rapid feedback from complex physics datasets.

Thea Energy’s planar coil architecture uses arrays of flat, mass-manufacturable magnets rather than the complex three-dimensional coils of conventional stellarator designs. The company holds six DOE INFUSE awards and was selected as an inaugural awardee of the DOE Milestone-Based Fusion Development Program. Gates has stated a company target of delivering fusion power by 2035. Eos, Thea Energy’s large-scale demonstration system, is intended to achieve power-plant-relevant, steady-state fusion ahead of Helios, which the company targets for commercial operation in the 2030s.

Stay ahead in the fusion revolution explore more breakthroughs from leading innovators in clean energy technology.