Beta — Coming Soon

The GoFundMe for AI compute

A global community donating spare GPU cycles to power AI research, healthcare, and agriculture projects across the Global South. Choose the projects you believe in. See exactly where your compute goes.

<1%
of global data centre capacity is in Africa — home to 18% of the world's population
$2.50–4.00
per GPU-hour on major cloud platforms — pricing built for wealthy enterprises
1 week
of intensive GPU access can exhaust a well-funded African doctoral researcher's annual budget

Brilliant researchers, locked out by infrastructure

AI is becoming foundational infrastructure — as critical as electricity. But access to the compute needed to build, train, and deploy AI systems remains concentrated among wealthy nations and well-capitalised organisations. The result is a systematic bias in who gets to build AI, and for whom.

$25,000–40,000
Cost of a single NVIDIA H100 GPU — equivalent to several years of faculty salaries at many Global South institutions, assuming export approval can even be obtained.
Free on OpenToken
Researchers and NGOs access compute at no cost, funded by a global community of volunteer donors who choose which projects their spare GPU cycles support.
3 barriers
Cost, sovereignty, and supply chain access. Export controls, foreign jurisdiction over data, and pricing optimised for enterprise budgets create compounding exclusion.

Donate cycles. Fund projects. See your impact.

OpenToken pools volunteer-donated GPU compute from people around the world and routes it to vetted research and humanitarian projects that need it most.

01

A Distributed Network

Gamers, university labs, and organisations worldwide donate idle GPU cycles through a lightweight agent. Sustainable infrastructure partners contribute upcycled servers. Everything flows into one intelligently managed compute pool.

02

Free for Those Who Need It

Researchers, NGOs, and social enterprises in the Global South access compute completely free. Projects are vetted for legitimacy and listed on the platform for donors to discover and support. Sustained by affordable paid access for startups and enterprises alongside the volunteer network.

03

Choose Your Impact

Pick the projects you care about — crop disease detection, maternal health, local language models. Your dashboard shows exactly where your cycles went: "Your 847 GPU-hours this month trained a crop disease detector for farmers in Uganda."

Philanthropy backed by economics

OpenToken combines donated compute with affordable paid access for startups and enterprises. Revenue from paid users funds platform operations; volunteer donations flow directly to free projects. The result is a model that doesn't depend on perpetual fundraising.

Volunteer donors
Unified Compute Pool
Free for researchers & NGOs
Affordable for startups & enterprises
Revenue funds operations

Built for the people who need it most

Researchers & Academics

PhD students, university labs, and research institutions in the Global South. Access free compute for AI/ML work — your research budget shouldn't determine what questions you can ask.

Startups & Builders

Building AI products for underserved markets? OpenToken provides affordable compute access so your budget goes to building, not to cloud bills.

NGOs & Social Enterprises

Organisations using AI for healthcare, agriculture, and education in developing contexts. Access compute completely free — funded by donors who believe in your mission.

Compute Donors

Gamers, university labs, crypto miners — anyone with idle GPU capacity. Donate your spare cycles to meaningful projects and see exactly where your contribution went.

The kind of work OpenToken makes possible

Projects like these need thousands of GPU-hours. On current platforms, the cost is prohibitive. On OpenToken, they're funded by a community of donors.

Illustrative examples — not active projects
Agriculture

Cassava Mosaic Disease Detection

Training a model to detect crop disease from smartphone photos, helping smallholder farmers across sub-Saharan Africa identify infections before they spread.

~10,000 GPU-hours
Healthcare

Swahili Language Model for Health Workers

Building a language model trained on East African healthcare terminology, enabling clinical decision support tools that actually work in the languages health workers use.

~15,000 GPU-hours
Maternal Health

Maternal Risk Prediction — Lagos

Developing a risk prediction system for maternal health clinics, trained on local data patterns to identify high-risk pregnancies earlier and reduce preventable mortality.

~8,000 GPU-hours

Infrastructure you can trust

Sustainable

Designed to minimise the environmental cost of AI.

Secure

Security protocols ensure jobs are securely run for both donors and recipients.

Transparent

Donors see exactly where their cycles go. Projects are vetted for legitimacy. Impact dashboards provide real accountability.

Community-Driven

Built on open-source infrastructure. Powered by a global network of volunteer donors. No single point of control or dependency.

Self-Sustaining

Affordable paid access for startups funds platform operations. Donated compute flows to free projects. No dependency on grants or perpetual fundraising.