How it works

From sunlight to answer.

Every inference at Node 01 follows the same path — photons in, tokens out, zero carbon in between. Here's the full pipeline.

01

Sunlight hits the array

A ground-mounted solar array tracks the sun across a south-facing slope. Photovoltaic cells convert daylight into DC power and feed it into a custom distribution bus inside the container — no grid connection, no fossil fuel backup.

02

Batteries store the day

Surplus solar charges a lithium battery bank throughout the day. The stored energy smooths cloud cover, covers evening demand, and carries the GPU cluster through the night — so inference doesn't stop when the sun goes down.

03

GPUs run inference

Inside the shipping container, a compact GPU rack loads open-weight models and runs inference on incoming requests. Cold alpine air is ducted through filtered intakes to keep silicon cool without compressors that would drain the solar budget.

04

AI served carbon-free

Responses travel back over a point-to-point wireless link to the internet. Every answer ships with 0 gCO₂ — metered per request — whether you reach Node 01 through chat or the REST API.

The result

A pipeline with zero carbon emissions.

No utility bill. No diesel generator. No water evaporated for cooling. Just sunlight converted into compute — and open models served to anyone with an API key.

Try Node 01