How FormulaFusion ranks

Credibility firewall

Disclosure: FormulaFusion is operated by people with an interest in Dispatch, an owner of grid-connected battery sites that may host fusion plants. Rankings are never influenced by commercial conversations — see Methodology.

• Dispatch's landowner/grid-connection interest is disclosed on every public page of FormulaFusion and in this methodology.
• No company is ever ranked differently because of a commercial conversation.
• The Grid Fit table is the honest reason a landowner is running this — we lead with that, not around it.
• Density or evidence-grade changes from a company reply are never published without a source citation, even an internal one.

Championship A — design-point density (MW / m³)

• Time-averaged power only. Never peak/instantaneous. Pulsed FRC, Z-pinch and laser-inertial concepts run at high duty-cycle peak power but low time-averaged power — ranking on peak would flatter them by 1–2 orders of magnitude.
• Volume boundary = ISLAND, not core. Island = everything inside the controlled/exclusion area required for that concept: plasma chamber + blanket + shield + coils, and (where applicable) the laser bay, cryoplant, pulsed-power hall or capacitor bank. Ranking on core-only volume would flatter laser-inertial and pulsed-power concepts, whose real footprint lives outside the chamber.
• Power = net electrical (MWe) where stated as a design target; MWth flagged separately and not mixed into the same rank.
• Every figure is a company design target unless marked otherwise — none is a measured net-power result, because none exists yet in the private sector as of mid-2026. That is exactly why Championship B exists.
• Numbers are FormulaFusion's own derivation from public specifications wherever possible, not company self-submissions. Company-submitted data is accepted and displayed, but always tagged with its evidence grade.
• As of Phase 0 no private developer has publicly disclosed a matched (power, island volume) pair, so no density is published yet.

Championship B — evidence grade

Measured on hardware of that class (net-power result on a full-scale or pilot machine). No private company holds an E1 for net electrical power as of mid-2026.

Subscale demonstration achieved AND a larger machine of that design is under construction.

Design point published with peer-reviewed or third-party technical backing (e.g. a national-lab collaboration, published physics basis).

Design point is company-stated only, with no independent technical validation available publicly.

Inferred by FormulaFusion from partial or indirect disclosure (funding size, general description, aggregator listings). Weakest grade — treat as provisional.

Rule: never blend the evidence grade into the density number. Publish them side by side — e.g. “41 MW/m³ (E4)” — so the reader can see the gap between ambition and proof.

Data confidence (separate from evidence grade)

Evidence grade rates the company's technical claim. Data confidence rates our sourcing of the row itself.

High

Primary source: company press release, SEC/regulatory filing, named-partner joint announcement.

Medium

Reputable secondary source (trade press, industry association report) but not company-primary.

Low

Aggregator/directory listing only; founding year, approach or funding inferred; needs direct verification.

Grid fit — field definitions

Footprint (ha, FOAK)

Land area required for the first-of-a-kind plant, fenced/controlled area, not just the building footprint.

Cooling type

Open-loop water / closed-loop water / air-cooled / not disclosed. Drives siting compatibility with ex-BESS sites.

Grid connection class

Distribution vs transmission voltage; approx MVA required at steady state.

Startup import load

Parasitic power draw during commissioning/ramp — it can exceed a BESS site's existing import capacity even where export capacity is ample.

Conversion type

Thermal (steam cycle) vs direct energy conversion. Thermal plants need cooling water and civil works closer to conventional fission/CCGT; direct-conversion plants do not.

Ramp capability

How fast net output can change — relevant to whether the plant is baseload-only or can also serve balancing/BESS-adjacent roles.

Status of this dataset

• Phase 0 draft — under right-of-reply. Companies are being sent their own row and asked to correct or fill the pending cells.
• Every entity row shows its as-of date, its listed sources, an evidence grade and a data-confidence rating. Assertion-level provenance (per-fact validity windows) is planned, not yet stored.
• Dutch text on entity pages is a machine draft awaiting editorial review; the English row is authoritative. Six of the fifteen concept 101s are drafts awaiting physicist review.