Learn › Comparisons & deep-dives
Application · AI compute
Fusion for AI compute.
The bottleneck on large-scale AI is shifting from chips to power. Firm, clean gigawatts — sited where the compute is — is exactly the gap the Kronos burner is designed to fill.
- Demand
- Gigawatt-scale, firm, continuous
- Machine
- MetroVolt (~1400 m) · ≈ +2832 MWe
- Advantage
- Clean baseload without weather risk
- Siting
- Co-located with compute
Training and serving frontier models is increasingly limited by electricity: firm, dense, and available now. Grids are congested and renewables need storage to serve 24/7 loads. A steady-state fusion generator sited next to the campus sidesteps both problems.
MetroVolt is the Kronos machine for this: ≈ +2832 MWe of firm output from low-neutron D–³He, with part of the power produced by direct conversion. It is the same burner as Aegis, built long for grid scale.
Honest statusThis is a designed capability. The burner's closing point is requirement-class and the fleet is gated on helium-3 supply.