The stellarator is a magnetic confinement fusion device that uses only external magnetic coils to create a three-dimensional, twisted magnetic field for plasma confinement. Unlike tokamaks, which rely on plasma current to help create the confining field, stellarators use no net plasma current, eliminating current-driven instabilities like disruptions. The magnetic field is created entirely by external coils arranged in complex, three-dimensional geometries. Modern optimized stellarators, such as Wendelstein 7-X, use computational optimization to design coil configurations that minimize neoclassical transport and magnetic field errors. Stellarators offer real advantages: steady-state operation without current drive, no disruptions, and confinement that can rival tokamaks. Challenges—complex coil geometries, construction cost, and unproven scaling—are being addressed by new manufacturing approaches, AI-accelerated design, and a wave of public and private investment. This article reviews stellarator physics, historical development, current experimental and commercial landscape, and prospects for fusion energy.

Stellarators create a twisted magnetic field using external coils:

• Helical coils: Wind around the torus, creating the twist
• Toroidal coils: Provide the main toroidal field
• Poloidal coils: Shape the field and control plasma position

The field lines follow a twisted path, wrapping around the torus multiple times before closing. This creates closed magnetic surfaces that confine the plasma.

Unlike tokamaks, stellarators use no net plasma current:

• Field created entirely by external coils
• Eliminates current-driven instabilities
• Enables steady-state operation without current drive
• But: Requires more complex coil geometry

The twisted field creates nested magnetic surfaces:

• Closed surfaces: Confine the plasma
• Magnetic axis: Central field line
• Separatrix: Boundary of closed surfaces

Plasma particles follow field lines, staying on these surfaces and remaining confined.

One of 50 non-planar superconducting coils built for Wendelstein 7-X. Each coil is a unique 3D shape, precision-machined to create the twisted magnetic field that confines the plasma. Credit: Max Planck Institute for Plasma Physics / CC BY-SA 4.0.

Lyman Spitzer invented the stellarator in 1951. Early devices (Model A, B, C) at Princeton demonstrated plasma confinement and basic physics understanding, but achieved poor confinement compared to tokamaks once the Soviet T-3 results became known. Key references for this period and the rise of the tokamak include:

• Spitzer, L. (1958). "The stellarator concept." Physics of Fluids, 1(4), 253–264. AIP — original stellarator concept.
• ITER. "August 1968: A revolution in fusion" — the T-3 tokamak breakthrough that shifted the field toward tokamaks.

When tokamaks showed superior performance, most fusion programs shifted to tokamaks; stellarator research continued in smaller efforts aimed at understanding why tokamaks performed better. Useful references for tokamak development in this era:

• Wesson, J. Tokamaks (Oxford University Press) — standard reference on tokamak physics and history.
• ITER — iter.org for the evolution toward large-scale tokamaks (e.g. JET, TFTR, and ITER).

Computational optimization enabled far better stellarator designs:

• Optimization codes: Design coils for optimal fields (e.g. Grieger et al. 1992).
• Wendelstein 7-X: First fully optimized stellarator (first plasma 2015); performance comparable to tokamaks.

This period renewed serious interest in stellarators as a path to fusion.

Inside the belly of a star machine: the plasma vessel of Wendelstein 7-X during assembly, showing stainless steel cover plates and water-cooled copper backing plates that protect against plasma–wall interactions. Credit: Gwurden / CC BY-SA 3.0.

The Wendelstein 7-X (W7-X) stellarator at the Max Planck Institute for Plasma Physics (IPP) in Greifswald, Germany, is the world’s largest optimized stellarator and the main experimental basis for today’s optimism about the approach.

• Location: Greifswald, Germany (IPP)
• Construction: 2005–2014
• First plasma: 2015
• Cost: ~€1 billion (total project and site)

The device uses 50 non-planar, superconducting coils in complex 3D shapes. Official project information, publications, and visitor information: ipp.mpg.de/w7x.

W7-X has demonstrated:

• Excellent confinement: Comparable to tokamaks
• Steady-state operation: 30-minute pulses (limited by heating, not stability)
• No disruptions: Confirms a key advantage over tokamaks
• Good plasma control: Stable operation

These results validate optimized stellarator design and support scaling studies for future power plants.

The Wendelstein 7-X stellarator in its experimental hall at the Max Planck Institute for Plasma Physics, Greifswald. The device is roughly 16 metres across—about the size of a two-storey house. Credit: Siarhei Besarab / CC BY-SA 4.0.

W7-X is a physics and engineering test device, not a fusion reactor:

• No D–T operation: Uses hydrogen/helium
• Limited heating: Not at fusion-relevant power
• Research device: Focus on physics and engineering validation

Scaling to reactor size and net power remains to be demonstrated by next-step devices.

Stellarators can operate continuously: no plasma current to drive, no current-driven instabilities, and natural steady-state—unlike tokamaks. That is a major advantage for a power plant.

Stellarators avoid the disruption problem: no plasma current instabilities, more stable operation, and less risk of damage. That simplifies reactor design and operation.

Modern optimized stellarators achieve confinement comparable to tokamaks, low neoclassical transport, and good plasma control. W7-X has demonstrated this capability in practice.

The traditional challenges for stellarators are real—but AI, better simulation tools, and improved understanding of plasma dynamics and magnetic fields offer a realistic path to overcoming them.

Stellarator coils are highly complex: 3D non-planar shapes and millimeter-level precision. That said, AI-driven design and new coil strategies are changing the picture. Planar-coil approaches (e.g. Thea Energy) and optimization codes that run in seconds instead of days are reducing both complexity and cost. Better understanding of plasma and magnetic field behavior is making it possible to hit target performance with simpler or more buildable coil sets.

Small errors in coil positions can create magnetic islands and degrade confinement. Tolerances are tight, but improved simulation and real-time control are making it feasible to design and operate within those tolerances. Tools like STELLAR-AI and StellFoundry are accelerating the feedback loop between design and physics, so “precision” is increasingly something that can be designed in rather than only achieved by brute-force manufacturing.

Stellarators have not yet been scaled to reactor size. The new wave of pilot devices (Eos, Alpha, Infinity One) and power-plant designs (Helios, Infinity Two, Stellaris) is explicitly aimed at testing that scaling. Combined with better simulation and AI-surrogate models, the community can explore more design points and converge on viable reactor configurations faster than in the past.

Stellarators have had less funding and fewer large devices than tokamaks. The gap is closing: private investment, public–private partnerships, and shared tools (e.g. STELLAR-AI) are building a larger base of data and operational experience. Each new device and each improvement in simulation adds to that base.

• Steady-state: No current drive needed
• No disruptions: More stable
• Better control: No plasma current to manage

• Complex coils: More difficult construction (though new designs aim to simplify)
• Higher cost for traditional coil sets (new manufacturing may reduce this)
• Less developed: Smaller historical program (now changing)

• Simpler coils: Easier construction in conventional designs
• More developed: Larger program, more data
• Proven scaling: ITER demonstrates large scale

• Pulsed operation: Requires current drive
• Disruptions: Major operational and design challenge
• Current control: Complex plasma control

The full magnet system of Wendelstein 7-X: 50 non-planar coils (shaped) and 20 planar coils (flat) work together to create the optimized magnetic field. Future commercial designs aim to simplify this architecture using AI-optimized coils and HTS materials. Credit: Max Planck Institute for Plasma Physics / CC BY 3.0.

Stellarators are increasingly seen as a frontrunner for commercial fusion because of their inherent stability (no disruptions, steady-state). Private companies and public partnerships are advancing toward pilot plants and first-of-a-kind power plants.

Type One Energy is advancing the Infinity Two stellarator concept at 350 MW scale. In 2025 they signed agreements with the Tennessee Valley Authority (TVA) to site a fusion plant at the retired Bull Run coal facility near Oak Ridge, Tennessee. They submitted initial licensing applications in early 2026 and have raised significant funding (including backing from investors such as Bill Gates). Grid integration is targeted around 2030.

• Type One Energy – TVA and first fusion license application in Tennessee
• NucNet – Type One Energy and TVA advance 350 MW stellarator project

Proxima Fusion is focused on quasi-isodynamic (QI) stellarators. In early 2026 they signed a memorandum of understanding with the Free State of Bavaria, RWE, and IPP to build the world’s first commercial stellarator power plant in Europe. They plan a demonstration device (Alpha) by the early 2030s to achieve net fusion energy.

• Proxima Fusion, RWE, Bavaria, IPP – MoU for first commercial stellarator power plant in Europe
• IPP – Memorandum of Understanding (2026)
• ANS – Proxima Fusion signs MoU with Bavaria, RWE, IPP

Thea Energy released preconceptual designs for the Helios power plant in late 2025, using many smaller, software-controlled planar magnets to simplify manufacturing. They aim to build a proof-of-concept device (Eos) in 2026, targeting commercial systems around 2030.

• Thea Energy – Eos integrated stellarator
• ANS – Thea Energy releases preconceptual plans for Helios
• DOE certifies Thea Energy fusion pilot plant preconceptual design (2026)

Patent filings for stellarator-related technology have risen sharply (from on the order of ~10 patent families in 2022 to dozens by 2025), with the United States holding a large share (~45%), reflecting strong private and institutional investment in the approach.

• AI and computing: New codes and platforms are simplifying magnet design and reducing cost. The PPPL-led StellFoundry initiative (2026) uses AI surrogates to test configurations in seconds or milliseconds instead of hours or days. STELLAR-AI (PPPL, 2026) accelerates fusion simulations across the community.
  • ANS – PPPL-led STELLAR-AI to advance fusion research
  • EurekAlert – Bringing fusion energy to the grid using AI
• High-temperature superconductors (HTS): Widely adopted to make magnets more compact and powerful, improving the economics of both tokamaks and stellarators.
• Quasi-isodynamic (QI) and optimized designs: Concepts such as SQuID (Stable Quasi-Isodynamic Designs) from IPP (2024 onward) meet power-plant physics requirements with low toroidal current and excellent confinement.
  • IPP – New stellarator design SQuID (Stable Quasi-Isodynamic Designs)
  • Cambridge Core – Quasi-isodynamic stellarator configuration towards a fusion power plant

AI advancements and better understanding and simulation tools for plasma dynamics and magnetic fields are a credible route to making stellarators practical. Faster optimization (StellFoundry, STELLAR-AI), improved physics models, and HTS magnets are reducing the historical penalties of coil complexity and cost. That doesn’t guarantee success, but it makes the “optimistic” case—pilot devices demonstrating net gain, then scaling to commercial plants—realistic rather than speculative. The same tools also lower the risk of the “cautious” outcome: even if the first commercial fusion plant is a tokamak, stellarator research and private projects will have advanced the whole field.

Stellarator technology continues to advance on the back of better optimization (including AI), improved construction and manufacturing, and lower costs. Pilot devices (Eos, Alpha, Infinity One) demonstrate net gain or strong performance; scaling to commercial plants (Helios, Infinity Two, Proxima’s Stellaris) proceeds. Stellarators become a viable—and for some applications preferred—route to steady-state fusion.

Stellarators remain a serious parallel path: they contribute to fusion science and may reach net energy in the 2030s. Technical or economic hurdles could still favor tokamaks or other concepts for the first commercial plants—but the same AI and simulation tools that help stellarators also benefit the whole fusion portfolio. Stellarator research and private investment deliver lasting value either way.

Coil complexity, cost, or scaling could still limit stellarators; tokamaks or other approaches may prove more practical. Maintaining a diversity of approaches—with clear milestones and peer review—is the right strategy. The difference today is that the tools to de-risk that path (AI, simulation, better plasma physics) are no longer science fiction; they are in use and improving.

Panoramic view of Wendelstein 7-X. The machine that proved stellarators can work—now the question is whether the next generation can produce power. Credit: Max Planck Institute for Plasma Physics / CC BY 3.0.

1. Max Planck Institute for Plasma Physics. Wendelstein 7-X. ipp.mpg.de/w7x — Official W7-X project site: status, results, and publications.
2. Nührenberg, J., & Zille, R. (1988). "Quasi-helically symmetric toroidal stellarators." Physics Letters A, 129(2), 113–117. ScienceDirect — Early quasi-helical symmetry work.
3. Drevlak, M., et al. (2019). "Optimized stellarators with resonant magnetic perturbations." Plasma Physics and Controlled Fusion, 61(1), 014018. IOP Science — Recent optimization of stellarator configurations.