Powering Advanced Intelligence, Cleanly

Large-scale computation is becoming a defining human capability — and its growing appetite for power can be met without a growing emissions bill.

Vision, not specification. This page describes a future that fusion could help make possible — not a product promise, forecast, or performance claim. The Kronos machines today are design and simulation studies: the breeder (Hyperion) and the burner (Aegis / MetroVolt). No hardware net-gain has been demonstrated; construction of the first breeder is planned to begin in Q2 2027, with first-of-a-kind first tritium targeted around 2030.

Compute is the new heavy industry

Training and running large models, simulating proteins and plasmas, rendering weather and climate at fine resolution — all of it runs on electricity, and the demand is rising steeply. Left on a fossil grid, that demand becomes a large new source of emissions. On clean firm power, the same growth in capability arrives without the carbon.

Data centers are unusual loads: they run flat-out around the clock and cannot tolerate interruption. That profile is a poor match for variable generation alone and a natural match for firm clean power delivered on site. The burner housing MetroVolt is aimed squarely at this case — direct energy conversion to electricity, sited where the compute is.

A virtuous loop

There is a further link worth naming. The same computation that fusion powers is also what makes fusion tractable: surrogate models, plasma-control learning, and materials search all lean on large-scale compute. Advanced intelligence and clean energy can reinforce each other — better models design better machines, which power more capable models.

Kronos treats this as a working relationship, not a slogan. The AI and control stack that runs the plasma is the same class of system that the plant is designed to power for others.

The caveat, stated plainly

• Direct energy conversion in the burner is a design choice, not yet a demonstrated plant.
• The burner is D–³He with a modest neutron fraction of 5.44%, still a simulation study today.
• Clean compute at scale is the aim; the machines must first be built and proven.

Follow the loop into computing and discovery, and the load story into the electrified world.