Honest Roadmap To DC Power
MetroVolt is a design and simulation study today; the breeder is built first from Q2 2027, a test burner follows near 2032, and a commercial fleet near 2036.
The real timeline
No MetroVolt unit has been built, and no burner has produced net power. The Kronos program has a deliberate order: build the breeder (Hyperion) first, use it to breed the burner's fuel, and mature the burner in simulation until its physics gates can be retired.
Breeder construction begins Q2 2027, with first-of-a-kind first tritium near 2030; NOAK and BOAK units follow. A test burner is targeted near 2032, and a commercial burner fleet near 2036. There is no hardware net-gain claim before the breeder's FOAK, and the burner's commercial date depends on retiring its gates and standing up the He-3 supply.
For a data center, this roadmap means MetroVolt is something to design for, not to procure today. What exists now is a rigorous set of physics models, reproducible simulations, and engineering studies — plus a candid list of what must be true before a burner can power compute at all, let alone at Tier III.
The value of stating it this way is that every claim on these pages can be checked against the timeline and the gates. Nothing here asks a compute operator to believe a machine is ready that is not; it asks them to understand the design, the physics, and the honest distance to deployment.
- No MetroVolt unit built; no net-power hardware yet
- Breeder first: build Q2 2027, first tritium ~2030
- Test burner ~2032; commercial fleet ~2036
- Commercial burner gated on retiring the four physics/supply gates