What EPCs Are Telling Us About Fusion – And Why We're Building NuPlant™ Around It

Simon Woodruff, CEO & Founder, nTtau Digital

Over the past year, we've spent a lot of time in rooms with engineering, procurement, and construction (EPC) firms who are actively trying to figure out how to build fusion power plants. Not fusion reactors — plants. The distinction matters, and it's exactly why we built NuPlant™. One of these engagements ran for the better part of a year. It started with a technical discovery session hosted by Fusion Energy Insights, where a senior engineering leader from a major global EPC laid out, in detail, how their firm was thinking about entering the fusion sector. We recorded it, transcribed it, and — rather than filing it away — we responded to it concretely: we mapped their stated needs directly against what nTtau Digital and our Fusion Advisory Services could offer. That response opened the door. The EPC came to visit us twice in Santa Fe, brought one of their fusion customers along, and walked us through their entire conceptual design process: the tools they use (SmartPlant 3D and similar platforms), the design stages, and the 30 to 40 documents their teams generate at each of those stages. We saw the first two stages in full. We classed the information, understood the workflow, and built out detailed mappings of exactly where our software and advisory capabilities plug in.

What came out of that process is a clearer picture than most fusion developers have of what a serious EPC actually needs before it will commit engineering hours to a fusion project. It's worth sharing, because if you're an EPC evaluating whether and how to move into fusion, you are almost certainly wrestling with the same set of problems.

The problem EPCs keep describing

Strip away the specifics of any one conversation, and a consistent pattern emerges across the EPCs we talk to.

Timing is the first fight. Most fusion developers haven't yet reached breakeven — scientific or engineering — and are understandably reluctant to spend money on plant-level engineering before their core physics is proven. But if an EPC waits for full R&D certainty before starting conceptual design, the commercial timeline stretches by years, because comprehensive design, procurement, and construction each add years of their own on top. Nobody wants to be the firm that added five years to a customer's roadmap by being too conservative — or the firm that got the plant layout catastrophically wrong by moving too fast.

Procurement and construction don't wait for engineering to finish. In an idealised world, engineering completes, then procurement starts, then construction begins. In the real world — and especially in a first-of-a-kind sector like fusion — these phases overlap out of necessity. Long-lead items like high-temperature superconducting magnets, vacuum vessels, and specialised power systems need to be ordered while the design is still evolving. Getting that phasing wrong means either idle capital or expensive rework when specs shift mid-procurement.

Fusion companies and EPCs speak different languages. A fusion startup's expertise sits in plasma physics and reactor engineering — not industrial-scale plant design, cost estimation, or regulatory pathways. An EPC doesn't design the fusion core, but it needs precise interface information: what comes in (coolant, feed gases, power), what comes out (steam, neutrons, effluent), and the boundary conditions in between. Without that interface clearly defined, EPCs are laying out balance-of-plant systems against guesswork.

Late design changes are expensive in ways that compound. Once procurement and civil works are underway, a change to reactor geometry or power output can force costly rework across piping, instrumentation, and structures. The tension is real: lock the design down too early and you risk building around a reactor concept that hasn't been validated; leave it too flexible for too long and every change ripples through a partially built plant.

Investors want commercial timelines that R&D realities don't naturally produce. The gap between a successful lab demonstration and a bankable commercial plant can run four to six years or more, plus commissioning. EPCs that can produce credible early feasibility studies — real costs, real layouts, real schedules — give fusion developers something concrete to show investors, and give themselves an early, low-risk way to prove value before committing to a full engineering contract.

None of this is unique to fusion, exactly. Every first-of-a-kind power technology has faced some version of it. What's different this time is the pace at which fusion developers are moving, and the number of EPCs now deciding, in parallel, whether and how to get in early.

Where the industry already is on digital plant design

This isn't a new problem for engineering firms in the abstract. Digital twins, model-based systems engineering, and integrated cost-schedule-risk platforms have been standard practice in conventional power, oil & gas, and nuclear new-build for years — SmartPlant 3D, AVEVA, and similar tools already carry decades of institutional workflow for exactly this kind of concurrent design, procurement, and construction management. The state of the art in power plant engineering has been moving toward earlier, more dynamic modeling: building a digital representation of the plant well before detailed design is locked, and updating it continuously as underlying technology data matures, rather than waiting for a "final" design freeze that first-of-a-kind projects can rarely afford.

What's missing is that toolchain adapted to a technology whose core physics is still being characterised in real time — where the "digital twin" has to represent a reactor concept that may still change, alongside a balance-of-plant that behaves like any other thermal power station. Fusion developers need a way to generate credible, cost-linked plant designs before their physics is fully locked, and EPCs need a way to plug their existing engineering workflows into that evolving picture without re-deriving it from scratch at every stage gate.

Why we built NuPlant™ this way

Every conversation we've had with EPCs — this one included — has pushed NuPlant™ in the same direction: a whole-plant simulation, costing, and optimisation platform that treats the fusion core as one evolving input among many, not the entire problem.

• Early-stage, cost-linked design so a fusion developer can generate a credible plant-level layout and cost basis long before Q≥1 is demonstrated — giving EPCs something real to scope against instead of a concept sketch.

• Interface-first modeling, so the boundary conditions between reactor core and balance-of-plant — coolant flows, power output, tritium handling, thermal loads — are explicit and updated automatically as the core technology matures, rather than being re-negotiated by hand at each design review.

• Modular, stage-gated architecture that mirrors how EPCs already phase engineering, procurement, and construction — so partial design maturity in one system doesn't force a full re-derivation of the rest of the plant.

• A standardised costing framework (we use a 23-account structure familiar to anyone who has worked conventional power cost estimation) so early feasibility numbers are directly comparable to the kind of estimates EPCs already produce for conventional and nuclear projects — the language investors and boards already trust.

We didn't build this in the abstract and then go looking for validation. We built it, in significant part, by sitting in a room with an EPC's engineering team, watching how they actually work, and mapping our capabilities against their real design stages and real document sets — thirty or forty documents deep, at every stage.

That particular engagement led into deeper licensing and contracting discussions, and while the timing didn't align for a near-term tactical contract at the time, the strategic relationship remains very much alive, and we expect to pick it back up. More importantly, the pattern it revealed has held across every other EPC conversation since: the firms serious about entering fusion are all wrestling with the same timing, interface, and cost-certainty problems, and all looking for the same kind of tooling to solve them.

If your firm is somewhere in that process — evaluating fusion customers, trying to work out how early is too early to engage, or figuring out how to plug your existing plant-design workflow into a technology that's still being characterised — that's precisely the conversation we want to be having. It's the conversation that built NuPlant™.