A D-shaped ring of fire
Deuterium–tritium plasma circulates around the torus, never touching the wall. Shaping it into a tall "D" improves stability and lets more pressure be held by the same field.
- Core temperature
- 150M °C
- Elongation κ
- 1.9
[01] Why a new sun
Temperature at the Sun's core. Its gravity does the rest.
What a tokamak needs — ten times hotter, because magnets must stand in for a star's gravity.
Released by a single deuterium–tritium fusion. Millions of times more than any chemical reaction.
The Sun's total output. Earth intercepts about one two-billionth of it.
[02] The load is coming
AI data centres, electrified heat and transport, and new industry are arriving faster than new generation can connect. The bottleneck isn't only energy — it's firm energy, delivered where the load is.
Projected global electricity demand in 2050, up from about 27,000 TWh in 2024.
Source: Breakthrough EnergyGeneration and storage waiting in US interconnection queues at the end of 2024.
Source: LBNL, Queued Up 2025Median wait from interconnection request to commercial operation — double what it was fifteen years ago.
Source: LBNL, Queued Up 2025Designed to IEEE 738, CIGRE TB 601, IEC 61215, IEC 61730, UL 9540A, NFPA 855, IEC 62933, ASME BPVC Section III Division 4, and 10 CFR Part 30.
[03] One system, four machines
No single technology decarbonises a grid. Helionovus engineers four — and the software that dispatches them as one — so firm, clean power arrives every hour of the year.
A compact high-field tokamak: superconducting magnets squeeze a star's worth of physics into a machine the size of a house.
Perovskite-on-silicon tandem modules that split the spectrum: blue light converted on top, red and infrared below.
Sodium-ion for the evening peak, iron-air for the windless week. Built from abundant materials — no cobalt, no nickel.
Sensors that ride live conductors and compute what each line can really carry — every few seconds.
One platform to forecast, dispatch and account for every electron across all four machines.
[04] NOVA · Fusion
The Sun confines fusion with gravity. On Earth we use magnetic fields — and the stronger the field, the smaller the machine. NOVA is built around high-temperature superconducting magnets wound from REBCO tape, reaching 20 tesla at the coil.
Take the machine apart
Reaction Deuterium + tritium
D + T → 4He (3.5 MeV) + n (14.1 MeV)
Deuterium is drawn from seawater. Tritium is bred inside the machine itself, when escaping neutrons strike lithium in the blanket. Most of the energy leaves with the neutron — which is why NOVA's blanket matters as much as its magnets.
Deuterium–tritium plasma circulates around the torus, never touching the wall. Shaping it into a tall "D" improves stability and lets more pressure be held by the same field.
D-shaped coils of REBCO superconducting tape, cooled to about 20 K, wrap the plasma. Fusion power density scales roughly with the fourth power of the field — so doubling B shrinks the machine dramatically.
The central solenoid drives millions of amps through the plasma itself. That current adds a poloidal field, twisting field lines into helices that keep particles from drifting out.
A tank of FLiBe (Li₂BeF₄) surrounds the vacuum vessel. It stops the neutrons, carries their heat to the turbines, and breeds tritium: ⁶Li + n → T + ⁴He.
At the bottom of the "D", field lines are diverted onto tungsten targets that remove helium ash and heat fluxes comparable to a rocket nozzle's.
Why compact The field is the lever
Pfus/V ∝ β2 B4
Conventional superconductors top out near 12 T at the coil. REBCO tape keeps superconducting at far higher fields and temperatures, which lets NOVA reach ITER-class plasma performance in about a thirtieth of the plasma volume.
Specifications Design targets
Pilot and plant parameters are Helionovus design targets, benchmarked against published compact high-field tokamak studies.
A net-energy demonstration machine. Pulses of about ten seconds, designed to show fusion gain Q > 10 — ten times more fusion power out than heating power in.
A steady-state plant with demountable magnets, so the vacuum vessel can be replaced as a module. FLiBe carries heat to a conventional steam or supercritical-CO₂ cycle.
Non-insulated REBCO pancake windings, conduction-cooled with helium gas at about 20 K rather than liquid helium at 4 K — a large saving in refrigeration power.
Fusion can't run away: any disturbance cools the plasma and the reaction stops within seconds. There is no long-lived high-level waste. In the United States, fusion machines are regulated under the byproduct-material framework (10 CFR Part 30) rather than as fission reactors.
[05] LUMEN · Solar
Silicon turns high-energy blue photons into the same voltage as red ones and loses the difference as heat. LUMEN stacks a 1.68 eV perovskite cell on a silicon heterojunction cell, so each half of the spectrum is converted by the material best suited to it.

Perovskites are crystals that can be printed from solution at low temperature — thin enough to sit on top of a silicon wafer without replacing it.
Spectrum Where the photons go
Records: industry-certified cell results, 2025–26. Limits: detailed-balance calculations under AM1.5G.
Architecture Inside a LUMEN cell
[06] UMBRA · Storage
Four-hour batteries cover the evening. They can't carry a grid through a cold, still, overcast week. UMBRA pairs sodium-ion for daily cycling with iron-air for multi-day gaps — both built from abundant, low-cost materials.
Illustrative hourly model: 1 GW average demand; 6.5 GWp solar; days 3–5 overcast at 15–25% of clear-sky output. Sodium-ion 1.5 GW / 6 GWh at 90% round-trip; iron-air 1.5 GW / 120 GWh at 45%, entering the week 85% charged.
Sodium-ion cells in 5 MWh containers. Same factory processes as lithium-ion, but with no lithium, cobalt or nickel, and they keep working in deep cold.
Discharging, iron pellets breathe in oxygen and rust. Charging reverses it. The chemistry is slow and lower in efficiency — and very cheap per kilowatt-hour, which is what multi-day storage needs.
| Chemistry | Best duration | Cell energy density | Cycle life | Key materials | Where it fits |
|---|---|---|---|---|---|
| LFP lithium-ionIndustry standard | 1–4 h | 160–205 Wh/kg | 6,000–10,000 | Li, Fe, P, graphite | Daily shifting, frequency response |
| Sodium-ionUMBRA-S | 2–6 h | ~160–175 Wh/kg | 10,000+ | Na, Fe/Mn, hard carbon | Daily cycling, cold climates |
| Iron-airUMBRA-L | 24–150 h | Low (stationary) | Multi-decade | Fe, water, air | Multi-day weather gaps |
| Solid-stateEmerging | 1–4 h | 300–500 Wh/kg | In qualification | Li metal, ceramic | Vehicles first, then grid |
Figures are typical published ranges for each chemistry as of 2025–26.
[07] HALO · Grid
Lines are rated for the worst case: a hot, still, sunlit afternoon. Most hours are cooler and windier, and the conductor could safely carry far more. HALO measures the line itself, so operators can use the capacity that's really there.
Dynamic line rating 230 kV · Drake ACSR · 24 h
Average capacity unlocked across this summer day, computed from the IEEE 738 heat balance. Field studies report 10–40% depending on season and terrain.
Of new line built. Reconductoring and DLR together are the fastest capacity a transmission owner can add.
Hardware HALO sensor
[08] Heliostat · Software
A heliostat is a mirror that tracks the Sun. Ours tracks everything else: generation, storage, lines and load, with a live digital twin of each site, weather-driven forecasts and dispatch that respects the grid's real limits.
Open HeliostatEvery panel string, battery rack and line span, modelled and live.
Weather-driven generation and load forecasts, from the next five minutes to next week.
Charge, discharge and curtail against price, carbon and dynamic line ratings.
Ask why output dropped, what tomorrow looks like or what a new asset would change.
[09] How we work
We start from what the laws of nature allow — detailed-balance limits, plasma scaling laws, electrochemical potentials — and work back to what can be manufactured.
Every claim is tested on real hardware at meaningful scale: magnet coils at full field, modules on outdoor racks, storage on live feeders.
Bankable warranties, open data interfaces and designs certified to the standards utilities and regulators already use.
20 T HTS test coil. HALO on 1,200 km of line. LUMEN pilot line.
First 100-hour UMBRA-L system energised on a utility feeder.
NOVA-1 first plasma.
NOVA-1 targets Q > 1, then Q > 10.
First NOVA-P site selected and licensed.
First fusion electrons delivered to the grid.
[11] Contact
Utilities, data-centre developers, industrial off-takers, investors and engineers — tell us about the problem you're solving.