Kasi: We Put the AI in Space

In 2023, a foreign cloud provider shut down CyberRant overnight.

Fifteen thousand users. Revenue. Gone. An entire business built over years, serving real people, wiped out because the infrastructure lived on someone else's servers, under someone else's terms, in someone else's jurisdiction.

One email. No warning. No appeal. No recourse.

I sat with that for a long time.

And I made a decision that has shaped everything we have done since: Africa will never be at the mercy of foreign infrastructure again. Not for cloud. Not for AI. Not for the compute underneath it all. We are going to own the layer. All of it.

Today, I want to tell you exactly where that decision has led us.

We are sending the AI into space.

Not metaphorically. Not “cloud” as some distant server farm you are meant to imagine floating in the sky.

Literally. Physically. We are sending AI inference hardware into the stratosphere, 20 kilometres above the African continent, where it will float above the weather, above commercial aviation, and serve hundreds of millions of people below with latency that has never existed on this continent before.

This is Rantera Kasi.

Hybrid balloon platforms. Floating at 20 kilometres altitude. Powered entirely by the sun. Cooled by radiating heat directly to space. Serving AI to a 500km radius beneath them. Zero electricity costs. Zero industrial cooling. No undersea cables between the user and the answer.

Just Africa. And the sky above it.

Here is what nobody else figured out.

Everyone assumed AI required the heaviest infrastructure ever built. GPU servers weighing 35 kilograms per unit. Drawing 6,000 watts each. Requiring industrial cooling systems the size of warehouses. That is what the data centers look like today. That is what Africa has been told it needs to build, or borrow, or rent.

That assumption is only half true.

You need heavy infrastructure to train a model. To do the work of learning from billions of tokens, adjusting billions of parameters, running for weeks across thousands of chips. Yes. That is heavy, power-hungry work. Keep it on the ground.

But to run a trained model, to answer a question, generate a response, process a voice command, power an agent, you do not need a GPU farm. You need inference chips. Chips that weigh under 1 kilogram. Chips that draw just 5 to 50 watts. Delivering the same quality AI at a fraction of the weight and a fraction of the power.

Train the models on the ground. Fly the inference chips into the stratosphere.

That is the insight. The one nobody else assembled.

We assembled it.

The numbers that stop you cold.

A user in Lagos querying an AI model hosted in Europe right now waits 150 to 300 milliseconds for every single response. Every word. Every sentence. Every answer. That is not a minor inconvenience. That is the hard ceiling on everything that could be built. Voice assistants that feel alive. Agents that think in real time. Applications that simply cannot exist today because the physics do not allow it.

A Kasi platform above Lagos serves that same user at 5 to 15 milliseconds.

Ten to twenty times faster.

From space.

Zero electricity costs after deployment. The sun pays the power bill. Forever. The data never leaves African airspace. No foreign router. No foreign jurisdiction the data passes through. No morning email that can wipe it all out.

One platform. A 500 kilometre radius. Three platforms, positioned above Abuja, Nairobi, and Kigali, cover over 500 million Africans across 15 countries.

All of them. With AI that has never existed on this continent before.

Three cities. One continental footprint.

Abuja, Nigeria. West Africa hub. Rantera's headquarters city. The largest AI market on the continent. This is where we start.

Nairobi, Kenya. East Africa hub. Connected to the submarine cable via Mombasa. The best connectivity infrastructure in East Africa.

Kigali, Rwanda. Central Africa hub. Best power reliability. Cool climate at 1,500 metres altitude. The most tech-forward regulatory environment on the continent.

Together: 1.5 million square kilometres of coverage. 500 million people. AI that depends on no one outside the continent.

This is just Phase 1.

Kasi Phase 1 is stratospheric. Deployable from 2027. Ground gateways being established now. Platform design is in progress.

Phase 2 is Low Earth Orbit. Satellites at 400 to 600 kilometres altitude. Permanent orbital infrastructure. No balloon replacement required. Timeline: 2030.

Phase 3 is geostationary orbit. 36,000 kilometres. A permanent African orbital data center. In the sky. Forever. Timeline: 2032 and beyond.

We are not building toward orbit as an aspiration. We are building toward it one stage at a time, with proven technology, with a clear path.

The sky is not the limit. It is the foundation.

Why this is the most important thing we will ever do.

Every piece of infrastructure Africa has ever had, it waited for someone else to build. The railroads. The telecom cables. The cloud regions. Each time, we spent decades paying rent on access to our own data, our own communications, our own economy.

AI is not just another infrastructure layer.

AI is the layer underneath everything else. The intelligence that will run healthcare, agriculture, finance, education, government. The system that will make decisions about African lives, African businesses, African futures. If Africa does not own this layer, we will spend the next fifty years paying for access to our own future.

We are building it. Right now. From Africa. For Africa.

The models are live today. Hira is serving real conversations. The infrastructure is coming. And when it arrives, it will not be a server farm in someone else's country that Africa rents access to.

It will be floating above the continent. Powered by the same sun that has shone on Africa since the beginning of time. Answerable to no foreign jurisdiction. Owned by the continent it serves.

“Africa's AI. Powered from the Space.”

If you want to be part of what we are building, as a user, a developer, an investor, or a partner, start here.