We’re raising $30,000 to build a shared AI training server that will serve students for the next five years, students in Dzaleka Refugee Camp in Malawi, Kakuma in Kenya, Kampala, rural Rwanda, Ethiopia, and even underserved neighborhoods in Boise.

Right now, when a student wants to train an AI model, it can cost $5, $10, sometimes $50, just for one try. Most of our learners don’t know if their program will have funding next month. So they hesitate. They simplify their ideas. They stop exploring.

But AI isn’t learned in one go. It’s messy. One need to try, fail, adjust, try again, maybe 30 or 50 times—until something clicks. Without the freedom to experiment, that learning never happens.

We’ve watched students build incredible projects… that never leave their laptops. Why? Because putting an app online needs a live server, and hosting costs add up fast. We can’t afford $100 a month per project across multiple countries. So their work stays hidden. No users. No feedback. No real-world test.

And that’s a problem because, employers don’t ask for theory. They ask, “What have you built that people actually use?”

For ten years, we’ve used onsite laptops. Now looking tat current trends, a good laptop for AI costs around $3,000 to $5,000. Even if we bought ten, each student would still be working in isolation. Their project might be brilliant but it would die the day the course ended.

What we are proposing is One powerful, shared server

With it, a student in Kakuma can train a model at the same time as someone in Kampala. A developer in Boise can debug code with a peer in rural Zambia. A health app built in Malawi can go live and serve a local clinic. A farming tool made in Kenya can be used and improved by farmers in Tanzania.

They deploy. People use it. They learn what works, what breaks, and how to fix it.

That’s how they would become job-ready. That’s how they would gather confidence to start their businesses.

Here’s what we’re building

Four NVIDIA RTX A5000 GPUs
128GB RAM
10TB storage
Server-grade CPU, power backup, secure global access

It’s real lab-grade hardware. Once it’s running, students anywhere can use it for development, training, inferences and deployment with no extra cost.

The numbers:

$24,000 covers all hardware, shipping, and setup
$6,000 keeps it running the first year (internet, power, software, cooling)
Just $3,000/year after that

Total: $30,000 to serve over 600 students across five years.

Using cloud services for the same work would cost more than $54,000, and students would still second-guess every training run. Owned infrastructure removes that fear.

The impact is that students will build language models in Swahili, Kirundi, Amharic or any language of their interest. They’ll make health tools that work on basic phones. Business apps for street vendors. Weather predictors for smallholder farmers. And when they apply for jobs, they’ll show portfolios with live, maintained applications, exactly what employers want or they may turn those apps if they got traction to become businesses.

Background:

We started in Dzaleka Camp in 2015 and today, we’ve trained over 4,000 young people. Hundreds have remote tech jobs and hundreds went on to university. Also we have worked closely to UNHCR, SNHU, Microsoft 4Africa , and others. We’re a registered 501(c)(3) which mean your gift and donations are tax-deductible.

Every quarter, we’ll share updates on who trained, what they built, who got hired, what’s working.

We’re asking you to help us give students the one thing they’ve been missing: a place to build, test, fail, fix, and ship without limits.

If that matters to you, please join us.

Contact: info@takenolab.org | takenolab.org

takenoLAB - Technology Learning and Building Solutions.

Talent is everywhere. Opportunity shouldn’t depend on where you were born.

Where technologies get created

Help Us Build Real AI Infrastructure for Under-served Communities - Together