Skip to content
Technology How it works Breeder — Hyperion Burner — Aegis Burner — MetroVolt AI-Native Architecture Magnets Fuel cycle Safety Roadmap
Solutions AI & Data Centers Defense & Government Grid & Baseload Neutron Detection Quantum
Learn Technical Library
Proof Publications Whitepapers Technical Library Open Science & Reproducibility The Honest Gates
Company About / Mission Leadership Environment Health & Safety Investors Careers Press Contact
3D Model
AI Architecture › L0 · Foundation
L0 · Foundation

Bare-Metal Provisioning and Determinism

How Kronos pins microarchitecture, libraries, and topology on owned nodes so that physics runs are numerically stable across years.

THE STACK · click to jumpL7Ecosystem & StrategyL6Experience & VisualizationL5Applications & CopilotsL4OrchestrationL3Twin Modeling & AIL2Data FabricL1Control PlaneL0Foundation▲tlmctl▼L0 · FOUNDATIONThe offline compute substrate — multi-physics & batch training.1Cloud HPCelastic burst2Bare-Metal ClusterGPU / CPU3Supercomputingmulti-physics runs4Batch Trainingmodel builds5Simulation FarmGrad-Shafranov · MHD6Object StorecheckpointsMACHINE TIETrains the models that ship UP to L3 — no real-time path to the machine.KRONOS FUSION ENERGYAI-NATIVE S.M.A.R.T. GENERATORFOUNDATIONSHEET 02REV. 2026-08L0 · AI-NATIVE STACK
L0 · Foundation — its place in the stack (left, click any layer) and its internal components (right). Telemetry rises; control descends.

Determinism as a provisioning requirement

For the reproducible core of L0, provisioning is not just about acquiring cycles; it is about acquiring the same cycles every time. A coupled solve of the breeder equilibrium and blanket thermomechanics can be chaotic in the mathematical sense: tiny numerical differences amplify. To certify a result years apart, Kronos pins the entire numerical environment on bare-metal.

The reproducibility class

Certified nodes are grouped into reproducibility classes. A run declares its class; the scheduler places it only on matching hardware. This means a Grad-Shafranov equilibrium computed for the negative-triangularity delta -0.30 breeder shape can be re-derived bit-for-bit, which matters because these equilibria seed the twin surrogates that will later run in real time.

Why this precedes hardware

The breeder does not exist yet; construction begins Q2 2027 and first tritium is expected around 2030. Until then, the numerical model is the machine. If the model is not reproducible, there is nothing stable to design against. Bare-metal determinism is therefore a design-integrity requirement, not an optimization.

Provisioning also records everything needed to reconstruct a run: node identity, environment hash, library manifest, and RNG seeds. This manifest travels with every result into the petabyte archive, so a burner or breeder study can be audited and replayed on demand. Determinism without recording is not reproducibility; the two are provisioned together.

Content reviewed August 2026 · design-and-simulation stage