The Standard We Hold Ourselves To

High standards at Kronos are a concrete practice: get the physics right, quantify what you do not know, and let anyone check the work.

A standard is a practice, not a slogan

Kronos designs its machines digitally before any metal is cut. That makes the quality of the analysis the quality of the product. A high standard is therefore not an aspiration we hang on the wall; it is a set of practices we can point to on every number we publish: rigorous physics, an explicit error budget, verification that the code solves the equations, validation against reality where reality exists, reproducible pipelines, a frozen design point under change control, and predictions written before the test that could disprove them.

Why we start here

Fusion has a long history of numbers that could not survive scrutiny. We think the antidote is boring and specific: state every assumption, carry every uncertainty, and mark the boundary between what simulation has settled and what only hardware can settle. The breeder (Hyperion) and the burner (Aegis / MetroVolt) are design and simulation studies today, not built machines, and we say so on every page.

The pages in this section describe each practice in turn: how we verify code, how we build an error budget, why we freeze a design point, how peer review and pre-registration work, and why we treat all of this as a moral commitment rather than a checkbox. Read together, they describe an organization that would rather be caught being careful than be caught being impressive, and that treats the analysis behind an unbuilt machine as the machine's first and most testable component.