Four Ways to Build a Fusion Reactor (and Which One Wins on Price)

A big debate in the field of nuclear fusion is which concept would win on economics. Tokamak, stellarator or tandem mirror? Closed-field or open-field? Magnetic confinement or inertial fusion? Every fusion company has a favourite.

The trouble is, analyses aren’t comparing like with like. One team's tokamak costing may assume a friendlier magnet supply chain than the stellarator team down the hall assumed for theirs. So when someone asks the question that actually matters to an investor, a utility, or a regulator - which concept gets me to the lowest cost of electricity, at scale - nobody can really answer it, because the four models aren't built on the same foundation.

That's what we aimed to solve with this analysis. We took four fusion concepts - stellarator, tokamak, tandem mirror, and inertial fusion energy - and ran them all through the same pipeline, with the same first wall loading, the same blanket and shield materials, the same magnet technology (where applicable), and the same balance-of-plant assumptions. Same rules, four concepts, three plant sizes. Whatever differences fall out the other end are real differences in the physics and engineering, not artifacts of two analysts making different assumptions.

The first thing you notice when you actually build the geometry for all four concepts side by side is how differently they use space. A tandem mirror is long and thin - two plasma plugs on the ends of a central cell. A tokamak and a stellarator both curl the plasma back on itself into a closed loop, but the stellarator has to twist itself into a 3D shape to do it without external current drive, which means its coils are highly non-axisymmetric. IFE, by contrast, doesn't confine anything continuously at all; it's concentric shells around a target point.

The problem is that geometry drives volume, and volume drives cost. Our comparison

found that the tandem mirror's extended, open-field layout means substantially more vacuum vessel and blanket material than the compact, closed-field tokamak and stellarator needed for the same power output. You're paying for all that extra material before you've spent a thing on a single magnet or laser.

Once the geometry is built, we hand it to a standard 23-account cost structure; the same

accounting scheme used across the industry, so a magnet is a magnet whether it belongs to a tokamak or a stellarator. That consistency is what lets us actually rank the four concepts instead of just describing them.

The results, at 1, 2, and 4 GW net electric output, tell a pretty clear story:

● IFE comes out swinging. At 1 GW, it's the cheapest concept in the study - a compact

vacuum vessel and modest heating requirements give it a real head start. But that lead

erodes as plants get bigger, because the laser driver and target-factory costs climb

steeply with power.

● The tandem mirror starts slower and finishes strongest. It's not the cheapest at 1 GW,

but it improves with scale faster than any other concept, and by 4 GW it's the most

cost-competitive magnetic confinement option in the study.

● The stellarator sits in the middle the whole way. Its compact plasma and first-wall

geometry actually help it, but that advantage is offset by the cost of building the

non-axisymmetric coils.

● Tokamak has the hardest time. Magnet costs grow faster than power output across the whole range we studied, and the supplementary heating bill at large scale keeps growing right along with it. It's the concept people know best, and in this like-for-like comparison, it's also the one that struggles most to get cheaper as it gets bigger.

The punchline, if there is one: bigger is not automatically better, and "better" depends

enormously on which concept you're building and what problem you're trying to solve. A 1 GW IFE plant and a 4 GW tandem mirror plant are answering completely different questions, even though they'll both show up on the same LCOE chart.

Are you interested in a cross-comparison analysis as above, or perhaps a full NuPlant optimisation, for your own concept? The spec sheet that goes with this post lays out exactly what we need from you and what you'll get back. Fill in the form on the final page and send it to info@nttaudigital.com. If you'd prefer to talk it through first, you can request a demo, start a 30-day pilot, or get in touch with us directly. We'll come back to you within 3–5 business days.

View the spec sheet here:

nTtau_Digital_Spec_Sheets___concept_comparison.pdf

Download PDF • 1.36MB