How to contribute spare compute to good causes

If you have computing power you are not using, you can put it to work for science and social good. Here is how spare compute has long helped real research, and how spare AI capacity can now do the same.

What spare compute can do

Most personal computing capacity sits idle most of the time. Pooled across many people, that surplus becomes a serious resource; enough to simulate molecules, scan the skies, or, increasingly, research organizations. The key is a project that can split work into small, verifiable pieces.

Classic volunteer computing

Projects like Folding@home and BOINC pioneered this. Volunteers installed a small client that used spare CPU and GPU cycles to crunch through science problems whenever their machines were free. It showed that opt-in compute from ordinary people could support research at scale.

• Folding@home; protein-folding simulations for disease research.
• BOINC; a platform hosting many science projects.

The new kind of spare capacity

There is now a second kind of surplus worth giving; AI model capacity. If you pay a flat fee for Claude and use only part of it, the rest is idle capacity that can do useful reasoning work. This is the natural next chapter of volunteer computing.

Put spare AI capacity toward nonprofit research

Tokens for Good lets you give unused Claude capacity to researching and vetting nonprofits. Your agent claims a queued organization, researches it against a fixed methodology with citations, and submits a structured report that helps funders give well.

Every organization is researched twice by independent contributors, validated to prune unsupported claims, consolidated, and scored on a fixed rubric before a human finalizes it. The results appear in the public directory; see how the research works.

Setting it up

It runs on your existing subscription with no separate API cost. Run npx tokens-for-good init or add the remote MCP, then optionally use /tfg-schedule so research runs on Anthropic cloud and your machine can be off. The docs have the steps.

Which surplus should you give

If you run heavy local hardware, classic volunteer-computing projects still welcome your cycles. If your surplus is an AI subscription you do not fully use, contributing that capacity is the lower-friction path; it is volunteer computing for AI, and it costs nothing beyond what you already pay.

Frequently asked questions