The first essay in a series exploring why Africa’s greatest AI advantage is not compute, but people.
I was watching a video last night of a honey badger standing its ground against a lion.
It is a small animal, rarely more than sixteen kilogrammes. The lion outweighed it by at least fifteen times, enough to end almost any other contest before it began. The honey badger did not run. It squared its body, lowered its head, and waited.
This is not a rare clip. There is a whole genre of honey badger videos on the internet, honey badgers facing down lions, honey badgers eating puff adders and cobras whole, honey badgers digging into places nothing that size has any business going. People find them funny, and they are funny, in the way that any mismatch between size and confidence is funny. But I must have watched six or seven of these videos in a row last night, and somewhere around the fourth one, I stopped laughing and started paying attention.
Because the honey badger is not brave in some abstract, admirable way. It has not decided to be fearless. It is behaving rationally, given what it actually has.
Its skin is thick and, more importantly, loose. Loose enough that when something larger clamps down on it, the honey badger can twist inside its own hide and bite back before the grip tightens. Its front claws are disproportionately large for its body, built for digging through packed earth and, when required, through an opponent. It carries a partial resistance to snake venom that lets it survive bites that would kill almost any other animal its size. It has never once, in its entire evolutionary history, had the option of winning through mass. So it did not evolve toward mass. It evolved toward everything else. Skin. Claws. Nerve. A biochemistry that shrugs off poison.
The forest does not care about fairness. Everything in it is eating everything else, and for most small animals, survival means invisibility. Hide well, run fast, never be seen by the thing that could kill you in one motion. The honey badger took a different path entirely. It looked at what it could never have, size, and built its whole existence around what it already had instead.
I have not been able to stop thinking about this animal, because I think Africa is standing exactly where it stands.
The lion in the room
Let me say the uncomfortable thing plainly. Africa is not going to out-build the United States or China on artificial intelligence infrastructure. Not this decade, and probably not the one after it.
The continent holds around 0.6% of global data centre capacity. Active capacity across the whole of Africa sits at roughly 360 megawatts, against a global active capacity of 55,000 megawatts. Even if every announced African data centre project is completed on schedule, analysts expect our global share to hold roughly steady rather than grow, because the rest of the world is expanding faster than we are.
Some frontier AI labs now spend more training a single model than many African governments spend on technology in an entire year. None of that capital is arriving in Lagos or Nairobi or Kigali at anything close to that scale in the foreseeable future.
This is not defeatism. It is the same arithmetic the honey badger faces every single time it meets a lion. Mass versus mass, it loses before the encounter starts. There is no version of that fight it wins by trying to become larger.
So it does not try. It leans, instead, into everything else it has.
What we actually have
Iyinoluwa Aboyeji, who co-founded Andela and Flutterwave and now runs Future Africa, has been making an argument I keep returning to. He draws a distinction between two futures for artificial intelligence. One he calls Generative AI, the Silicon Valley model, capital-heavy, built to produce autonomous output and, in many of its most ambitious forms, to remove the human from the process entirely. The other he calls Assistive AI, or human-augmented AI, a model where the technology exists to amplify and multiply human capability rather than replace it.
His argument is that Africa should not try to compete in the first category. We do not have the capital density, the compute, or the concentration of frontier research talent to win a race built on autonomous, labour-displacing systems. But in the second category, the augmented category, he argues the continent holds a genuine structural advantage. Africa has the world’s youngest population, and the fastest-growing working-age population of any region on earth. Where Silicon Valley treats engineers as a scarce, extraordinarily expensive resource to be economised through automation, an augmented worker in Lagos or Nairobi, paired with the right tools, can match Western output at a fraction of the cost, while keeping the human, not the model, at the centre of the value created.
The lion wins on mass. Silicon Valley wins on capital, chips, and scale. Neither the honey badger nor Africa can meet that on its own terms. But mass is not the only thing that determines an outcome in the forest, and capital is not the only thing that determines an outcome in the AI economy.
As routine work becomes cheap and abundant, judgment becomes scarce. The ability to interpret context. The ability to know when the model is wrong.
This is already visible in the data. Roles where AI automates the routine so that human judgment matters more are growing twice as fast as roles where AI simply makes a task easier for anyone to perform, with wage growth 42% faster in the first category than the second.
Compute is no longer the scarce resource.
Judgment is.
That is our thick skin. That is our claw. That is the venom we have already built a resistance to.
I think of one of our own operators, who now runs a function for a company thousands of kilometres away that once required three people. She did not get there by competing with AI. She got there by directing it, catching what it gets wrong, and bringing judgment to decisions the model could not make on its own. It is already happening, quietly, across African-led teams serving clients thousands of kilometres away.
What it takes to live like a honey badger
Naming an advantage is the easy part. The honey badger did not survive because someone pointed out that it had loose skin. It survived because its entire way of living is organised around that one fact.
If Africa is going to make the human layer actually central to its AI strategy, rather than a comforting story we tell ourselves, five things have to change.
Education has to stop testing recall and start testing judgment.
Training has to happen outside universities as well as inside them.
We have to compete in the augmentation market instead of chasing the automation market.
We have to own the contextual data that turns individual judgment into collective intelligence.
And we have to stop selling Africa as cheap. The real advantage is judgment.
Each of these deserves its own full treatment. Over the coming weeks, I will take each one in turn.
The bet
The honey badger is not confused about what it is. It does not spend its short, violent life wishing it were a lion. It knows precisely what it has, and it has arranged its entire existence around that knowledge, without apology and without waiting for permission.
Africa spends a great deal of energy wishing it were somewhere else in this particular story. Wishing we had built the frontier models. Wishing we had the chip fabrication plants. Wishing we had arrived earlier at the table where these decisions are made.
We did not, and for the most part, that race is already decided.
But the race that will define the next decade is not the race to build the most capable model. It is the race to build the most capable humans to stand alongside those models, directing them, correcting them, bringing to them the one thing no amount of compute can manufacture. That race has barely begun.
Competing on price is a race Africa will eventually lose. Competing on judgment is one we can win.
The honey badger never won by becoming a smaller lion.
Africa will not become a smaller Silicon Valley.
We are underprepared. Those are not the same problem, and only one of them is fatal.
