Kristal Farms is a cold-climate, hydro-powered compute infrastructure designed to deliver two things at once:
Compute capacity (hosted in modular “pads”), exported by high-quality fiber.
Local community benefits, by turning server waste heat into building heating and greenhouse food production.
What makes it different
The core move is simple: put the compute in the village near heat users, not far away near the power source. Instead of building long transmission lines, the project uses local hydro on site and converts waste heat into something useful. Tenants lease black-box compute pads (power, cooling, fiber provided), while the village gains low-cost heat and improved connectivity.
Outcomes (what the village gets)
Lower-cost heat for public buildings first (and then homes), with a “heat-first” operating rule.
Greenhouse capability powered by recycled heat (food resilience and local jobs).
Better connectivity: fiber infrastructure that can also serve local institutions.
Transparent reporting through a public dashboard of outcomes (energy, heat, reliability, community benefit).
Operating model (what tenants get)
A strict black-box tenancy model: the host can monitor infrastructure health, but cannot inspect tenant data or internal operations.
Clear operating constraints: heat-first rules and environmental compliance are part of the deal.
How it works (high level)
Local hydro powers the site using a short connection to a village substation (avoiding long new transmission corridors).
Compute runs in modular pads (tenants bring hardware/software; the site provides the utilities).
Waste heat is captured and routed to local heat sinks (public buildings first, then homes, plus greenhouse and thermal storage).
Work is exported by fiber with monitoring focused on availability/latency and separation between tenant traffic and community traffic.
Heat-first rule
The operating priority is:
reuse → store → reject, with community heating prioritized.
Tenancy boundary
The host provides utilities and physical security up to the pad boundary, and monitors only infrastructure metrics—not tenant content.
What gets measured
PUE, WUE (~0 by design), useful heat delivered, HUF, diesel avoided, and uptime/occupancy (plus network and community benefit indicators).
Governance (so “benefit” is enforceable)
Project Council (overall strategy and dispute resolution)
Environment Committee (environmental compliance and incident review)
Kristals Council (public-interest knowledge commons program, topic selection, validation)
Phasing
Phase 1 proves the basics: compute running, useful heat flowing, data exported by fiber; with thermal storage and greenhouse readiness.
Scaling happens only after go/no-go criteria are met (reliable fiber, safe commissioned power, functioning heat reuse, signed agreements, seated governance, and a public dashboard).
