Data Centers in Homes — the Eco Thesis (Consumer DC)
Data Centers in Homes — the Eco Thesis (Consumer DC)
Child of BuildDoc — MachineSync (download-and-sync runner + hardware funnels) — the vision page for the consumer data center: compute living in homes, done in a way that is genuinely ecological, not greenwashed. Teka grounding: help the human, help the earth. The MachineSync fleet IS the consumer DC — this page is the eco architecture of that idea.
The thesis: a data center's two sins are wasted heat and wasted distance. A home solves both — compute heat is home heat, and the compute lives where its person lives. Distributed beats centralized on ecology when the architecture is honest about it.
Why home DCs can be the eco option
1 — Heat is the product, not the waste. Centralized DCs spend enormous energy removing heat. A home machine's heat lands in living space: in heating season, every watt of compute is a watt of heating displaced — effectively near-100% energy utility (compute + warmth from the same draw). Precedent exists (compute-heater companies have heated homes and pools with server heat); we make it native: the AARON/Mac-class box as a quiet warm object that earns its watts twice. Honest flip-side stated plainly: in cooling season the equation inverts — so scheduling matters (below).
2 — Solar + storage pairing. Home machines can follow home generation: heavy jobs scheduled into solar peak hours, battery-buffered, grid-light. The fleet becomes a load that absorbs renewable surplus instead of demanding baseload.
3 — Temporal shifting (the queue is green). Most agent work is async by nature — spawns, long runs, batch research. The MachineSync queue can schedule deferrable jobs to off-peak/green-grid windows (carbon-aware scheduling: run when the grid is cleanest). Determinism in the substrate makes this a scheduler policy, not a hope.
4 — Hardware life extension. The fleet runs on machines people already own — Macs, Studios, repurposed boxes — extending useful hardware life instead of demanding new silicon. Every year added to a machine's life is embodied carbon amortized further. The AARON line designs for longevity and repair from day one.
5 — No marginal buildings, no marginal land. The consumer DC adds zero construction, zero new land take, zero diesel backup fleets. It uses rooms that are already heated, wired, and connected.
6 — Proximity. Compute beside its person trims transit energy and keeps data sovereign at home (privacy and ecology pointing the same direction — the local/offline model hosting wedge).
The honest constraints (eco claims must survive audit)
Efficiency-per-watt is worse than hyperscale for raw utilization — home machines idle more. Mitigation: the fleet pools idle capacity (your box works the queue when you don't need it), raising utilization toward DC levels while heat stays useful. Claims are measured via the telemetry spine (per-machine energy + utilization), never asserted.
Cooling-season inversion — summer compute in hot climates is anti-eco at home. The scheduler weights seasonal/regional reality: winter-heavy in cold regions, solar-hour-heavy in sunny ones.
Grid honesty — "green" depends on local grid mix; carbon-aware scheduling uses real grid-carbon signals, not vibes.
How it composes with the stack
MachineSync (the fleet + queue = the scheduling substrate) · Telemetry Spine (energy/utilization measurement — the eco claims' receipts) · CLOUD-189 (credits can price green-window compute cheaper — the Credits Rail earn loop meets the earth: run green, earn more) · AARON hardware (the purpose-built warm-quiet-repairable node) · Teka Campuses (campus microgrids as the mid-scale tier between home and hyperscale) · Frontier-Model Readiness (local BYOM models are the workload).
The line for the narrative
Centralized AI burns watts in deserts to heat nothing. The Cloud's fleet warms the rooms its people live in, runs when the sun does, and gets greener every winter. Harness and help the human, help the earth — in the same watt.
