Within the next 3 years, artificial intelligence will move from the edge of your life to the centre of it. Not as a feature you choose to use. As the layer beneath every significant decision you encounter — the doctor reading your scan, the teacher marking your child's work, the bank deciding your credit, the nurse answering at three in the morning when a patient has breathing difficulties. The intelligence informing each of those judgments will increasingly be AI. What the AI knows, what it assumes, whose experience shaped its sense of the world — these are no longer abstract questions. They are the architecture of daily life as it is about to be lived.
"Every AI model inherits the mindset and prejudices of the society that trained it — the assumptions about who is credible, which names sound trustworthy, what illness looks like, whose discomfort registers as a problem worth solving."
The nuances and the prejudices travel with the weights. They are not documented anywhere. They are not declared in any terms of service. They were simply the data, and now they are the model. A continent without its own intelligence layer has already been decided for — at the layer that will matter most — by people who were not building for it.
A problem of this scale requires a response of a particular kind. Not a conference. Not a working group. Not a continental AI strategy document that arrives 3 years after the window has closed. On the twelfth of September 1962, John Kennedy stood at Rice University in Houston and demonstrated what that kind of response looks like: We choose to go to the moon. The technology did not exist. The path was not mapped. He chose the destination anyway, because ceding the frontier to someone else was a worse outcome than the difficulty of the attempt. 7 years later, two men walked on the surface of the moon.
What made Apollo work was not audacity. It was the discipline that followed the declaration. 400,000 engineers, scientists, and workers. Hardware assembled from hundreds of companies across 12 countries, the best expertise drawn from wherever it existed. What was American was not the supply chain. It was the decision — the choice to go, the ownership of the goal, and the governance of what the endeavour was for.
PhotonAI is making the same kind of declaration for Africa. Twice.
The window
The foundational models being trained today will be the cognitive infrastructure of the 2030s. Once they are scaled, embedded in products, trusted by hospitals and courts and governments — the cost of replacing them becomes what economists call prohibitive and what everyone else calls impossible. Africa has watched this happen before, with telecommunications, with financial rails, with cloud infrastructure. The window opens, someone else builds through it, and the window closes. The window for foundational AI is open now.
We are the new architects of the African dream, where we must think like governments and operate like a business.
At last Friday's SXSW in London, Chale Sekwalor articulated the organising principle behind everything described in this letter. The government's horizon — long, continental, oriented toward what a civilisation requires rather than what a quarter demands. The business's execution — decisive, capital-efficient, answerable to results. Separating those two modes is what has produced 60 years of plans without capital and extraction without infrastructure. PhotonAI was founded to refuse that separation.
Angel-1
PhotonAI's first foundation model begins training now. Angel-1 is being built from scratch — not fine-tuned from an existing Western architecture, not adapted with a new coat of African paint. From scratch. On African languages, African medical literature, African legal traditions, African agricultural knowledge. On the full, unabbreviated inheritance of 1.5 billion people who have, until now, been the subjects of other people's training data rather than the authors of their own. The name was chosen with care. Angel-1 is Africa's first guardian model. We are building it to last.
Building for Africa does not mean building alone. Apollo required 400,000 people and hardware from hundreds of companies. We will work with the world's best OEMs, chip manufacturers, data centre builders, hardware suppliers, cloud infrastructure partners, and research institutions — from wherever in the world they are found. Excellence does not observe borders. What belongs to Africa is the training corpus, the language diversity, the values embedded in the model, and the decision about what Angel-1 exists to serve. Those are the sovereignty. Everything else is a supply chain.
NiO
Africa has never launched its own commercial satellite, from its own soil. NiO will change that. The spaceport will sit on the equator — for reasons that are purely physical. Equatorial launches benefit from the full rotational velocity of the earth, roughly 463 metres per second of free velocity that temperate-latitude sites simply cannot offer. The Indian Ocean to the east provides thousands of kilometres of uninhabited overflight. The orbital mechanics are unambiguous.
What NiO carries into orbit matters as much as the fact of getting there. The constellation will serve 3 purposes simultaneously. First: sovereign launch capability for the continent — Africa's own access to orbit, on Africa's schedule, answerable to no foreign launch provider and subject to no foreign queue. Second: Boomchain nodes in orbit, propagating Boom's decentralised settlement layer — a blockchain-based commerce, payment and settlement system enabling the 2 billion adults in cash-based economies to transact without banking intermediaries — via laser inter-satellite links . Third: direct internet connectivity beamed to the 600 million Africans who live beyond the reach of any ground-based network — the farmers, the clinics, the schools, the markets — whose participation in the twenty-first century has been rationed by the megabyte for long enough.
NiO is currently evaluating several sites along the equatorial band: São Tomé and Príncipe, Malabo in Equatorial Guinea, and Malindi in Kenya, among others. We will say plainly why we have not yet announced a single location. Every government that wants this facility will know it is in competition for it. That is by design. The country that offers the best combination of infrastructure, regulatory framework, land terms, and sovereign partnership will earn the spaceport and everything that comes with it — the jobs, the orbital economy, the institutional prestige of hosting Africa's first commercial launch site. We expect the conversations to be vigorous. We welcome them.
The same logic applies to the data centre programme. 4 sites are planned across the continent. The governments that move fastest, offer the strongest data residency frameworks, and partner most seriously on sovereign computing will be the ones that host them. We are not distributing this infrastructure for the sake of geographic balance. We are placing it where the conditions are best. Countries that want those conditions on their territory know what to do.
Kennedy went to the moon because the frontier was there and someone had to claim it. Africa's frontier is both cognitive and physical. Angel-1 claims the first. NiO claims the second.
RFP‑2026‑001 is open
PhotonAI invites proposals from suppliers, OEMs, hardware manufacturers, data centre builders, satellite component makers, power and cooling specialists, and cloud infrastructure partners from anywhere in the world. We are building 4 data centres across the continent, training Africa's first foundation model, and constructing the NiO equatorial launch facility — site to be confirmed following our evaluation of candidate locations. The full RFP documentation and technical specifications are available at photonai.ai/rfp.html or by writing to rfp@photonai.ai. The scale of the procurement is significant. The timeline is deliberate. The standards are uncompromising.
Selection will be governed by two criteria: the best available technology and the most innovative deployment and financing models. Geography is not a criterion. Existing relationships are not a criterion. We will work with whoever brings the best solution from wherever they are. The Apollo programme did not restrict its hardware to companies headquartered in Houston. PhotonAI will not restrict its supply chain to companies headquartered in Africa.
To register interest and receive the full RFP documentation: rfp@photonai.ai
The Series A is open
PhotonAI is raising its Series A to fund the first phase of Angel-1 training, the NiO spaceport programme, and the data centre buildout across the continent. Investors in this round are funding the cognitive and orbital infrastructure of a continent during the narrow window before that infrastructure is permanently set by someone else.
The companies that built the great infrastructure layers of the last thirty years built the layer beneath all products — and the returns compounded for decades because the thing being monetised was a dependency rather than a preference. Data centres, fibre, cloud: each was dismissed as capital-intensive and speculative until it wasn't. Angel-1 and NiO are that kind of investment, twice over, for a market of 1.5 billion people, at the moment the window is still open. The people who funded Apollo did not do so because they had modelled the return on lunar real estate. They funded it because they understood what was at stake, and what would happen if someone else got there first.
To request the investment memorandum: invest@photonai.ai
Unlike the United States, China or Russia — each with one head of state — Africa has 54. To build technology infrastructure that truly serves this continent, you cannot think like a company. You must think like a government, and execute like a business. That is what our teams at Photon AI and Boom are doing. Three calls. One moment.
To the scientists and researchers: the most interesting unsolved problems in AI are here. Building a model that genuinely understands Hausa — not as a translation layer from English, but as a language with its own grammar, its own pragmatics, its own embedded logic. Training a medical model on the actual African disease burden. Solving multilingual inference across 3,000 languages, not the 40 the major labs have treated as commercially significant. African or not, we are indifferent to your passport. We care about your work. Come and do it with us.
To join the research team: research@photonai.ai
To the 54 governments of Africa: you do not need to agree with each other to act. The country that builds sovereign AI infrastructure now will be setting the agenda in 10 years. The one that waits for continental consensus will be importing intelligence from the one that didn't. Both things remain true simultaneously: coordination is worth pursuing, and it will not arrive in time for this window. Acting now does not foreclose it. Not acting now forecloses everything.
Kennedy did not say the moon was close. He said it was chosen — because that challenge would serve to organise and measure the best of our energies and skills.
Angel-1 is chosen. NiO is chosen. Africa's cognitive and orbital sovereignty is chosen — on a timeline, with partners from everywhere, on Africa's terms.
We choose to go.