610,000 dead in 2024. 95% in one region. AI hasn't cured malaria, but it's cut drug-discovery timelines from years to months, and African researchers are now building the tools themselves. A personal look at what's actually changing, and what isn't yet.

No investor pitch. No bond rating. No roadshow. China's 15th Five-Year Plan is funding a nationwide AI data center buildout because the state decided to, not because a lender said yes. Part 3 closes Money in the AI Ecosystem: what that speed actually costs, and what, if anything, it offers Africa.

$570 billion. That's the debt quietly financing AI's physical buildout in 2026, nearly four times last year's pace, much of it structured to never touch a hyperscaler's own balance sheet. Africa isn't short of the capital to do something similar. It's short of the plumbing. Part 2 of Money in the AI Ecosystem breaks down both.

Two companies just took 43 percent of every venture dollar invested worldwide, in six months. Africa's entire startup ecosystem raised less in a year than that in a week. Part 1 of Money in the AI Ecosystem looks at what venture capital's concentration actually costs, and proposes two ways Africa can build something better than what it's copying.

Two companies just absorbed 43 percent of all global startup funding on the planet. Meanwhile, Africa holds 18 percent of the world's population and less than 1 percent of its data center capacity. This is the opening piece in a new three-part series tracing how money actually moves in the AI ecosystem, in venture capital, infrastructure debt, and state funding, and what Africa needs to build instead of copying.

A stranger on a flight home from Atlanta once described, without knowing it, the exact problem at the center of Africa's AI moment: the people most affected by a decision are too often the last ones in the room where it's made. Drawing on new research from Chinasa Okolo published in Science, this issue traces Africa's long history of exclusion from global science diplomacy, and makes the case that Africa cannot afford to wait for perfect continental consensus before claiming its seat at the AI negotiating table.

I almost deleted the email that changed my life. It sat in my spam folder for weeks - a recruiter trying to get a broke graduate student in front of her for an internship interview he didn't know he'd already half-earned.

This week's Sovereign Stack is a departure from the usual five-pillar analysis. It's a personal story about the relationships that quietly move us from where we are to where we're trying to go - a friend's offhand mention of a career fair, a recruiter who refused to let a missed email be the end of the story, a mentor who made a new country feel survivable, and a goodbye to a mother that almost didn't happen in time.

Read the full story: "The Email I Almost Deleted."

No one independently checks a frontier AI lab's safety claims the way regulators check a bank's books - yet. I break down what's missing, why the Big Four are quietly racing to fill that gap, and why Nigeria's 1956 oil discovery at Oloibiri is the clearest warning Africa has about governing transformative resources too late.

One Dutch company controls 100% of the machines that make advanced AI chips. One Taiwanese company manufactures 71% of the world's semiconductors. And the materials for the next generation of chips - gallium from bauxite - sit largely in African soil. The question is whether Africa captures any of the value.

Africa generates the AI queries. Africa pays the bandwidth bills. But the servers doing the thinking sit in Amsterdam, Virginia, and Singapore. The Stack Sovereignty Test turns to its second pillar - Compute - and asks the question that determines everything else: who owns the machines that think for you?

This one's a little different from my usual AI and tech content — it's personal. It's the story of a year that tested our faith, our finances, and our patience in ways I never saw coming. I'm sharing it because I believe someone needs to hear it. I hope it encourages you.

Africa is producing more AI-capable graduates than ever. And losing more of that capacity than it can afford. The Stack Sovereignty Test reveals why talent isn't just one of five pillars - it's the variable that determines whether all the others can be built. The window is open. It will not stay that way.

The Stanford AI Index 2026 has spoken. Three countries made the most dramatic gains in global AI rankings - none of them G7 nations. Their secret was not money. It was timing, political will, and legal architecture built before it was needed. Africa's window is open. The blueprint is sitting right there in the data.

Across Africa, a quiet argument says AI is a Silicon Valley toy - a distraction from feeding, healing, and employing our people. It's dangerously wrong. From offline crop-disease apps lifting Kenyan farmers' yields 30%, to AI ultrasounds saving mothers, to 230 million projected digital jobs, AI is already moving the needle. The only unforgivable response would be to look away.

Kenya just turned away a billion dollars - because the lights could not stay on.

Microsoft and G42's landmark data centre deal collapsed under the weight of a grid that wasn't ready. It is not a Kenyan failure. It is an African signal.

The AI race runs on electricity. Africa must decide - right now - whether it builds that future or watches someone else build it instead.

Africa survived colonialism. Survived exploitation. Survived being written out of every major economic revolution in modern history. We cannot afford to let AI be the next one. The race is already underway - infrastructure being built, rules being written - mostly without Africa in the room. That has to change.

The IEA is clear: no electricity, no AI. For Africa - youngest population, richest renewables, fastest-growing digital adoption - that's not a warning, it's an opening. The infrastructure decisions made in the next few years will determine whether Africa shapes the AI era on its own terms. The window is open.

A single AI data center consumes up to 100 megawatts. Global power demand from data centers will grow 128% by 2026. The grid cannot keep up. For the developed world, that's a crisis. For Africa, it's a crossroads. The decisions made today will determine who builds the next economy.