# Power & Grid > Canonical HTML: https://initkoa.org/infrastructures/kristal-farms/power-and-grid > Markdown mirror: https://initkoa.org/infrastructures/kristal-farms/power-and-grid/index.html.md > Route: /infrastructures/kristal-farms/power-and-grid > Source: app/infrastructures/kristal-farms/power-and-grid/page.mdx > Generated: 2026-06-13T02:25:42.872Z [Open the HTML page](https://initkoa.org/infrastructures/kristal-farms/power-and-grid) "How Kristal Farms turns local hydro power into dependable compute and community heat: short-distance distribution, metered interfaces, grid-safe operations, and emergency resilience.", # Power & Grid Kristal Farms is built around a simple idea: **use local hydro power on-site**, distribute it over **short distances**, and convert it into two useful outputs: - **Compute capacity** (exported by fiber) - **Heat** (used locally for buildings and food production) The power system is designed to be **low-loss, auditable, and grid-safe**—with clear operational rules that keep community needs first. ## The basic power flow **Hydro plant → short MV feeder → village substation → feeders to pads + heat equipment** Instead of building long transmission corridors, Kristal Farms concentrates interconnection and distribution in the village: - A **short medium-voltage (MV) feeder** links the hydro source to a **new village substation** - The substation distributes power to: - each **compute pad** (tenant containers) - the **heat system** equipment (pumps, exchanger station, controls) This “village-sited” layout keeps losses low and keeps the system legible: you can see where energy goes. ## Metering and auditability (by design) Every major interface is metered so the project can publish an honest accounting: - hydro output - substation in/out - per-pad power handoff This supports: - transparent billing and loss accounting - seasonal reporting (including **diesel avoided**) - operational governance (how much power is serving heat vs compute) (See also: Metrics & dashboard.) ## Grid-safe operations (no surprises for the village) Compute loads can be “spiky” if you energize large systems all at once. Kristal Farms avoids this through: - **power quality monitoring** (voltage stability, harmonics, events) - **coordinated start/stop sequencing** (pads are energized in a staged way to prevent inrush and flicker) - feeder protection with **selective coordination** (a fault on one pad should isolate that pad—not the whole site) - proper grounding and protective devices per feeder This is a key principle: the farm must behave like a **well-mannered industrial neighbor**, not a disruptive load. ## Heat-first load management Kristal Farms does not treat compute as the “master load.” The operating rule is: **community heating needs take priority**. That shows up in how power is managed: - pads are added in phases, matching IT power to available heat sinks - workloads can be shaped to follow heat demand (more batch work in cold periods; throttling when heat loops are saturated) - contracts and procedures encode curtailment rules: lower-priority compute yields before public heating is shorted (See also: Heat-first design.) ## Resilience (without normalizing diesel) The village may already have diesel generators. Kristal Farms keeps them as **emergency backup for critical loads only**—for rare outage events. In practice: - the normal operating state is hydro + on-site operations - diesel is reserved for essential services (e.g., clinic / NOC) during exceptional failures - compute can be safely curtailed under defined procedures This is “resilience without dependency”: backup exists, but the system is not designed to burn fuel to function. ## Safety and seasonal readiness The power system is treated as civic infrastructure, not a private lab: - strict lockout/tagout procedures at the substation and pad connections - routine pre-season testing (e.g., before winter ramp-up) to validate the full heating + electrical chain under load (See also: Environment & safety.) ## Related pages - How it works - Tenancy model - Cooling & water - Go / No-Go checklist