Why Direct Energy Conversion Uses Little Water
The burner (Aegis / MetroVolt) recovers electricity from charged particles directly, so it never needs the large steam-cycle cooling loop that dominates thermal-plant water use.
Most electricity today is made by boiling water. Coal, gas, nuclear fission, and even most concentrated-solar plants heat a working fluid, spin it through a turbine, and then must reject the leftover heat to a condenser. That condenser is where the water goes: cooling towers evaporate it, or once-through systems withdraw enormous volumes from a river or the sea. The thermodynamic reject heat is not optional in a steam cycle; it is set by the Carnot limit.
What direct energy conversion changes
The burner (Aegis / MetroVolt) is a D-3He tandem-mirror design in which the great majority of the fusion energy leaves the plasma as fast charged particles, not neutrons. Charged particles can be decelerated against an electric field and their kinetic energy collected as current. That is direct energy conversion. Energy that is turned straight into electricity never becomes reject heat, and so never needs a condenser or the cooling water a condenser demands.
Honest scope of the claim
This is a design-and-simulation result, not a measured plant. Two caveats keep it truthful. First, the burner is low-neutron, not aneutronic: the neutron fraction is 5.44%, and that neutron energy still deposits as heat in the first wall and blanket, which needs some cooling. Second, auxiliary systems, magnet cryogenics, and power electronics reject heat of their own. The correct statement is that direct conversion removes the dominant steam-cycle water demand, not that the burner uses no water.
The breeder (Hyperion) is different: it is a D-T machine whose value is tritium, helium-3, and 14 MeV neutrons, and it does carry a thermal load. Water context for the breeder is treated separately on the breeder cooling page.
A staged, honest timeline
This water advantage is a design-and-simulation result on the path to hardware, not a demonstrated plant figure. The breeder (Hyperion) leads the program with construction beginning Q2 2027 and first-of-a-kind first tritium around 2030; the burner follows. No hardware net-gain claim is made before FOAK, and the water case is presented on the same footing: a well-founded architectural argument about direct conversion, waiting on the machines actually being built to become a measured result.