Bill Gates came to Kigali last week and we treated it like a coronation. Three days of headlines, a tour of the National Health Intelligence Centre, a lecture up in Butaro, and enough handshake photos to fill a wedding album. The only thing missing was a public holiday.

ALSO READ: Kagame, Gates discuss cooperation in health, AI, agriculture

I get why. When a man of his caliber plants a tree in your country and comes back, years later, to water it, that lands. I almost forgave him for Windows Vista.

But read what Gates actually said in Butaro on July 22. Buried inside the compliments was a warning.

ALSO READ: Bill Gates urges early AI rollout in Africa to narrow health gap

Speaking at the first Paul Farmer Butaro Dialogues, he said that if we let things take their normal course, AI, like every advanced tool before it, ends up in the rich countries. The money is there and the data centres are there. Africa gets the technology last, if it gets it at all. We know that queue. We stood in it for vaccines.

The question I keep coming back to is: is Rwanda becoming the place where African AI gets built, or the place where other people's AI gets tested first?

ALSO READ: Bill Gates tours Rwamagana projects expanding access to healthcare

That leads to two very different outcomes.

The receipts point both ways. In January, Rwanda became the first country in Horizon 1000, a 50-million-dollar AI health programme from the Gates Foundation and OpenAI. The National Health Intelligence Centre (NHIC), launched in April 2025, now reads the country's health data in real time. On June 8, Cabinet approved a National AI Agency, the first of its kind on the continent. The AI Scaling Hub has pulled in around 25 billion francs. On paper, nobody in Africa is moving faster.

But look at the shape of it. We were picked for the pilot. The 50 million starts here. We are the launch site. Ask Florida how much of the rocket it gets to keep.

A hub builds. A launch site receives. We keep celebrating the second one and calling it the first.

If Rwanda becomes a real hub, the nation and companies like ours win. If it stays a testing ground, we still get work, just smaller, and always downstream of a model somebody else owns. Our accountant prefers the first option.

That is the part that keeps me up. Being first is not the same as being in charge. The 58,000 community health workers feeding data into the NHIC are doing real work, and the malaria they catch early is real. But the models reading their data sit somewhere else, and so does the computer they run on. We supply the ground truth. Someone else keeps the engine. Try asking one of these models something in Ikinyarwanda and watch it answer with confident nonsense.

Gates was clear this can go the other way. He wants Africa among the first places AI solves real problems, not the last, and he pointed to the University of Global Health Equity as the place to train these tools on local languages and local diseases. That is the hub version of the story. It is possible. It is just not automatic, and no number of visits delivers it.

So, watch the next announcement. When the next big AI programme lands in Africa, notice this one thing. Is Rwanda the country receiving it, or the country running it?

That one detail is the whole difference between a hub and a very well-run pilot.

The pass mark for Rwanda is whether, five years from now, the models get built here, or just tried here first.

The writer is an MIT-certified AI professional and the Head of Business and AI at Global Kwik Koders.