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

The Edge-to-Cloud Pipeline

The path a signal travels from a diagnostic feedthrough at the machine to the offline archive and back as a refined model.

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.

A signal's journey

Data in the Kronos architecture moves along a defined path from the machine to the foundation and back. It begins at the edge, a diagnostic signal through a cryo-rated vacuum feedthrough into the L1 control plane, passes through the L2 data fabric, lands in the L0 archive, is turned into models, and returns to the edge as compiled surrogates. The edge-to-cloud pipeline is that full circuit.

Outbound: from edge to foundation

On the way in, data is progressively aggregated. The edge sees raw, high-rate telemetry it must act on in microseconds; only a fraction of that, and derived features, flows onward for storage and learning. By the time data reaches L0 it is validated, contextualized, and archived, ready for the slow, thorough work the edge has no time for.

Inbound: from foundation to edge

On the way back, refined artifacts flow the opposite direction: retrained PINNs, GNNs, and surrogates are compiled, quantized, and deployed into the real-time twin and, where applicable, the edge. The pipeline is therefore a loop, not a one-way feed; the foundation continuously ships improved intelligence back toward the machine.

The defining feature of the pipeline is its latency gradient: microseconds at the edge, milliseconds in the twin, minutes to hours at the foundation. Each stage operates at the timescale its job demands, and data is escalated between stages as its purpose shifts from acting to learning.

The pipeline serves both machines identically in structure. Whether the source is the breeder's Mirnov coils or the burner's fast-ion trackers, the same outbound-aggregate, inbound-deploy loop applies, which is what lets one foundation support two very different machines.

Content reviewed August 2026 · design-and-simulation stage