Roadmap

A dated plan, with the gates named.

Kronos publishes not just where it's going, but what has to be true at each step — including what isn't proven yet.

2026Now

Open the physics

Two peer-review papers, two open Zenodo deposits, and a live simulation anyone can run. The breeder design point is solved and reproducible — gain Q 3.076 — and the burner's closing point is published with its conditions.

What it establishesThe physics closes at demonstrated confinement — the hard, novel part — with the open gates named, not hidden.

Q2 2027Targeted

Breeder construction begins

Construction of the first breeder (Hyperion) unit starts — a ~3.5-year build (target ~3 years) moving the breeder physics onto hardware, with the first-of-a-kind unit online — first tritium — targeted ~2030 (2031 on a conservative planning basis). Three breeder units follow the learning curve: FOAK → NOAK → BOAK.

Gate to clearDemonstrate the breeder design point on hardware; advance the tritium-breeding blanket from the demonstrated ~1.7 kg/yr toward the 4 kg/yr target.

~2030First unit online

FOAK online — first tritium

The first-of-a-kind breeder (Hyperion) unit comes online — first tritium — 2031 on a conservative planning basis. This is where the paper-solved gain gets hardware validation; until then the machines are not yet built and the gain is solved on paper.

Gate to clearHardware validation of the breeder design point, then two further units down the learning curve — NOAK, then BOAK.

~2030 → ~2035The foundry

Three breeders, down the learning curve

Breeders are built few and deliberately — a foundry for strategic materials, not a power fleet, because the materials market is finite. Build time falls with each unit: FOAK ~3.5 yr → NOAK ~3 yr → BOAK ~2.5 yr, giving about three breeders by ~2035.

Long-range outlookThereafter roughly one new breeder every 4–5 years — on the order of 5–6 units by 2050, tapering as lunar helium-3 supply arrives (~2040). Outlook, not a committed milestone.

2032 → 2036Commercial

The burner fleet

2032 — begin building the first test burner: a ~100 MW D–³He generator demonstrator. 2036 — begin building the commercial burner fleet in ~1 GW units, sized for practical siting and the learning curve: Aegis for fixed defense installations and MetroVolt for data-center power. Build time falls FOAK ~3 yr → NOAK ~2.5 yr → BOAK ~2 yr.

Gate to clearThe burner's requirement-class closure — chiefly the end-plug density (n_p/n_c ≈ 16) — demonstrated, not just specified. Fleet scale is gated on helium-3: breeder-bred early, then lunar He-3 (~2038–2040).

The missionDecades out

Strategic supply & firm power, at scale

A domestic source of the materials the U.S. can't otherwise make, and firm power for AI, industry, and defense — the reason the two machines exist. Long-range outlook for 2050: on the order of ~90 burner units (~90 GW), contingent on helium-3 supply.

The north starFuel follows purpose, not platform: breed what's needed, burn it where it's needed.

See the gates in detail.

The full register of open questions — each with the condition attached — is public.

Open the honest gates →

Dates are targets for a conceptual design and simulation study; the integrated Kronos machines are not yet built, and certain results are requirement-class (specified, not yet demonstrated). No financial figures are published on this site.

A national-lab-caliber team.