Data-Center Cooling and the Water Context for MetroVolt

Large data centers consume water both directly for cooling and indirectly through the thermoelectric grid that powers them; on-site generation touches both.

Hyperscale data centers use water in two distinct ways. Directly, many use evaporative or water-assisted cooling to reject server heat. Indirectly, they draw grid electricity that is largely thermoelectric, and that generation consumes water at the power plant. A full water accounting for a data center therefore has an on-site term and an embedded-in-electricity term.

Where MetroVolt sits

MetroVolt is the burner (Aegis / MetroVolt) in a data-center housing — an on-site generator. Because it uses direct energy conversion rather than a steam cycle, the generation side of the water ledger avoids the condenser cooling that makes grid electricity water-intensive. That addresses the embedded term. It does not by itself change how the servers are cooled; server cooling is a separate design choice for the facility.

Honest boundaries

• The burner is low-neutron (5.44%), not aneutronic; its residual heat still needs closed-loop cooling.
• Availability is a live gate: design-and-simulation availability of ~0.86-0.995 is 30-100x short of hyperscale Tier III (0.99982), so the burner is not a sole Tier-III power source today.
• We make no claim that a data center becomes water-free; we claim the generation architecture avoids the dominant steam-cycle demand.

The availability gate matters here because a data center values uptime above almost everything. The truthful position is that MetroVolt is a promising on-site generation architecture whose downtime performance, as studied, does not yet meet Tier III on its own. See availability and siting honesty.