Program · section 03 · the open record

The open record.

Kronos is the fusion company that shows its work. Of 238 research studies in the program, the 31 below are fully reproducible — pre-registered, citation-audited, with their data — and that is the bar for appearing here. In-progress work stays in the Lab until it clears it.

238
research studies
31
public & reproducible
19/19
regression checks pass
7
hi-fi code adapters
5
open Zenodo deposits

Two machines, from the frozen anchors

Breeder — HYPERION

fusion gain Q
3.076
fusion power
85.0 MW
plasma current
9.66 MA
TBR lever · BR-L2-A7
1.03–1.06 → 1.18–1.20 solid-Be

A strategic-materials platform, not a power plant — success is gain, TBR, He-3 and neutrons; net electricity is a category error, excluded. Tritium-marginal: closes on-model (H-C1) with a thin 3–6% margin, and the solid-Be lever (1.18–1.20, BR-L2-A7) is the build basis.

Burner — Aegis / MetroVolt

engineering gain Q_E
1.318
neutron fraction
5.44 %
fusion power
4298 MW
plug potential φ_i
248.8 keV

One residual physics gate — the plug ion-confining potential (φi; 112× GAMMA-10, never demonstrated; gate B-C3). The plug coil is buildable (stress ratio 0.26–0.37 — that gate is retired). Closed-on-model (QE 1.318); demonstrated = the WHAM-scale plug-potential experiment, 2028–30. Fleet power figures are design points, never demonstrated.

Reproducible studies

public tier · by track

Aegis (burner)

8 reproducible
IDStudyArtifacts
A2Siting and survivability4 csv · 0 py
A3Personnel-proximity radiation5 csv · 0 py
A4He-3 fuel security1 csv · 0 py
A5Defense-grade RAMI5 csv · 0 py
A6Self-fuelled demonstrator4 csv · 0 py
A7DoD federal licensing3 csv · 0 py
A8Load-profile fit5 csv · 0 py
DF3Survivability hardening4 csv · 0 py
DEPOSIT 10.5281/zenodo.21746479

Burner — shared

3 reproducible
IDStudyArtifacts
BURNER-SHAREDBURNER-SHARED3 csv · 0 py
P56Burner closure audit8 csv · 0 py
gate1-thermal-barrier-pushgate1-thermal-barrier-push4 csv · 0 py
DEPOSIT 10.5281/zenodo.21746479

Hyperion (breeder)

18 reproducible
IDStudyArtifacts
H1Tritium self-sufficiency3 csv · 4 py
H10Steady state sustainment4 csv · 1 py
H11licensing5 csv · 1 py
H12environmental4 csv · 1 py
H2Bootstrap net-power go-no-go4 csv · 3 py
H3Materials facility4 csv · 3 py
H4Isotope Co-Production1 csv · 1 py
H5He-3 separation2 csv · 1 py
H7NT kappa stability delta-banking4 csv · 1 py
H8disruptions3 csv · 1 py
H9exhaust divertor engineering3 csv · 4 py
HC-AHC-A2 csv · 7 py
HC-BHC-B1 csv · 2 py
P56Platform mission4 csv · 2 py
R1Extrapolation ledger1 csv · 0 py
R3Upgradability1 csv · 1 py
R4Post-diction1 csv · 1 py
de-riskde-risk6 csv · 4 py
DEPOSIT 10.5281/zenodo.21746157

Robustness

2 reproducible
IDStudyArtifacts
BC1BC11 csv · 0 py
BC2BC21 csv · 0 py

The engine's high-fidelity adapters

One anchor-tested engine sits behind every number. Where a full high-fidelity code has not yet been run, the engine stands in with a documented reduced-order surrogate and says so — tagged [T], naming the code that will retire it. The source is open in engine/.

AdapterReal codesRetired by
EdgeSOLsolps, b2.5, eireneSOLPS-ITER (edge/SOL plasma-neutral transport)
Gyrokinetic operatinggene, cgyro, gs2, tglfGENE / CGYRO / GS2 / TGLF (gyrokinetic turbulence)
KineticMirrorcql3d, cql3d-m, cql3dmCQL3D-m (bounce-averaged Fokker-Planck, mirror)
Neoclassicalneo, sfincsNEO / SFINCS (neoclassical drift-kinetic)
NucDatanjoy, njoy2016, sandyNJOY / SANDY (evaluated nuclear-data processing)
StructFEAansys, comsol, elmerfem, elmerANSYS / COMSOL / Elmer (structural FEA)
SyncTransportspece, cyneqSPECE / CYNEQ-class (synchrotron radiation transport)

Open deposits

Breeder — Hyperion (v2: 21795620)

10.5281/zenodo.21746157

46 studies flow into this deposit · CC BY 4.0

Open on Zenodo →

Burner — Tandem Mirror

10.5281/zenodo.21746479

128 studies flow into this deposit · CC BY 4.0

Open on Zenodo →

REBCO Magnet & Tape

10.5281/zenodo.21842514

3 studies flow into this deposit · CC BY 4.0

Open on Zenodo →

Direct Energy Conversion

10.5281/zenodo.21842864

3 studies flow into this deposit · CC BY 4.0

Open on Zenodo →

AI-ML-Quantum Control

10.5281/zenodo.21842371

2 studies flow into this deposit · CC BY 4.0

Open on Zenodo →

Go deeper

Roadmap & milestones

Dated commitments and the gates each stage must clear.

Open →

Readiness & risks

TRL/MRL by system and what would close each open risk.

Open →

Computing

HPC & quantum compute, the algorithm library, the AI-native stack.

Open →
Honest gates — carried, never softened No economics appears anywhere in this public record. Two-tier closure (ratified 2026-08-26): breeder = closes-on-model as a platform (net electricity is a category error); burner = closed-on-model (Q_E 1.318 + buildable coil), one residual physics gate (the plug potential). All three products close on the frozen physics; the distance to demonstrated closure is three named experiments, not open physics. Headline numbers are read from the frozen anchor file by construction — not hand-typed.
KRONOS FUSION ENERGY · Kronos ModelConceptual design and simulation study; no machine has been built