Idle machines, woven into one living compute cloud.

The AI boom is starving the planet of compute, and new datacenters take years to build. Mycelia grows one instead — weaving the GPUs and CPUs already sitting idle in millions of machines into a single, living network.

No new silicon. No new datacenters. Just the machines we already have.

Active nodes
GPUs online
Network TFLOP/s
Jobs running
How it works

Three steps to grow the network

A node is born, it fruits compute, it earns. The same loop, repeated across millions of machines, becomes one organism.

01

Connect

Install the lightweight agent or join from your browser. Your machine becomes a node the moment it comes online — no configuration, no lock-in.

02

Contribute

Donate the cycles you aren't using. The scheduler grows work toward idle capacity and retreats the instant you need your machine back.

03

Earn

Every completed slice of compute settles into MYC credits — redeemable, transferable, and proportional to the work your node fruited.

Earnings

See what your idle machine could fruit

Estimates settle in MYC and convert at the live network rate. Your node only works when it's otherwise idle, so this is yield from hardware you already own.

Assumes 1 MYC ≈ $0.12 · electricity $0.17/kWh · 30-day month.

GPU tier
Idle hours / day8h
1h24h

Est. monthly

1,632MYC

Net after power

$179

−$17 electricity

What you can run

Built for work that loves to spread

Anything that parallelizes thrives here — the network grows toward the work and reclaims itself when the job is done.

Rendering

Distribute frames and bake light across thousands of nodes at once.

/render

Batched AI inference

Fan out high-throughput inference for models that don't need a single host.

/inference

Scientific sims

Molecular dynamics, climate models, Monte Carlo — embarrassingly parallel by nature.

/compute

LoRA fine-tuning

Train adapters and run sweeps across the network's pooled GPU memory.

/training