Thea Energy Patent Portfolio Analysis

Money arrives twenty-five months after the paperwork

The Series B came from Brevan Howard Macro Venture, ALJ Investments and Alumni Ventures, at [amount undisclosed in brief]. Thea Energy was founded in 2022. The interesting number is not the round size but the gap: the last public filing is dated 23 July 2024, twenty-five months before this money landed. Everything the company has filed since then sits inside the eighteen-month publication window and is invisible to anyone reading the record today.

That gap cuts both ways. It means the visible portfolio is a snapshot of what the company believed in its second year, not what it believes now. It also means the three families on the public record were filed before the round, before the validation, before the hiring push. They are what a small team chose to protect when protecting things was expensive and nobody was watching.

One solid tick, two thin ones, same year

Three families, all filed in 2024, all via WO. One is granted and enforceable today. The other two are published and pending, with first grants expected around 2027. That is the entire public estate. For a company in its third year this is neither thin nor thick; it is early, and it is unusually concentrated in time.

Read the titles and the concentration makes sense. A method of designing substantially planar electromagnetic coils using curve torsion as an objective function. A planar coil stellarator with removable field shaping units. A method of designing planar coils using explicitly planar parameterizations. Two design methods and one machine built from what those methods produce. This is not three ideas. It is one idea, filed three ways, with the apparatus claim already granted.

One idea, filed three ways, with the apparatus claim already granted.

Fusion on one side, magnets on the other

The claims sit across G21B and H01F. Fusion reactors and magnets. That pairing is the whole thesis in two classification codes: the company says it develops fusion technology that simplifies construction and operation of fusion power systems, and the filings say the simplification is in the coils.

Stellarators are conventionally built from coils that twist in three dimensions and are brutal to manufacture. Filing on planar coils, and on the mathematics that produces them, is a claim on making the hard part easy rather than on the reactor itself. The removable field shaping unit in the granted family is the physical expression of that: geometry pushed out of the coil and into a component you can swap.

Six names, one on the load-bearing two

Six inventors are named across the three families. Thomas Kruger appears on two of them. Charles Pendleton Stuntz Swanson and Andrew Thomas Cote appear on one each. That is a shallow spread for a company now hiring across a hundred roles, and it tells you the design-method work came from a tight group rather than a department.

The hiring plan points the same direction. Twenty-nine hardware roles and nineteen in software and AI, including an integrated modeling intern and a fusion plasma physicist working on diagnostics and fast-ion transport. Modelling and magnet hardware are exactly the disciplines that produced the 2024 filings. Whatever is sitting unpublished is likely to be more of the same claim space, filed by a larger group.

Small circle, but standing somewhere specific

On the same CPC groups over five years: Hefei Inst Physical Sci Cas with 188 families, Hengdian Dmegc Magnetics with 89, Southwestern Inst Physics with 65, Ningbo Inst Materials & Eng Cas with 58, Jinhua Cimeng Intellectual Property Services with 51, Murata Manufacturing with 42. Thea Energy has three. On count alone the comparison is not close, and count alone is the wrong measure here.

Those large portfolios span magnet materials and tokamak programmes across a wide field. Thea Energy's three describe one manufacturable geometry and the optimisation that finds it. Narrow claims in a busy class survive on scope, not volume. The useful external signal is citation: Proxima Fusion has cited this work twice. A direct peer in stellarator development has read the file and the examiner record reflects it.

Narrow claims in a busy class survive on scope, not volume.

What the unpublished eighteen months contain

So the position, stated plainly: one granted family that is enforceable now, two pending behind it, all three pointed at the same narrow proposition that stellarator coils can be planar and the design method to get there is worth owning. That is a real moat and a small one. It holds if the granted claims reach the geometry rather than only one implementation of it.

The record stops in July 2024. Between that date and this round the company hired hardware and modelling people in numbers that dwarf the six inventors on file. If the filing velocity continued, the public portfolio will look materially different by 2026. If it did not, three families is what the moat is.