At the inaugural AI Everything Kenya x Gitex Kenya conference in Nairobi, a single question haunted every panel and every talk: will Africa build the next wave of artificial intelligence, or simply consume it?
For all the talk of Africa’s youth bulge and digital leapfrogging, the continent accounts for barely 1.3% of global patent filings, and a candid panel of financiers, data chiefs, and educators admitted the gap is not just technical, but structural.
Inside a packed Sarit Centre hall, an investor from Goldman Sachs spelled out the cold arithmetic of the AI economy: if your intellectual property cannot stay home, neither will your unicorns.

African AI (Artificial Intelligence) ambitions collided with hard economic reality last week as a high-level panel at the inaugural AI Everything Kenya x Gitex Kenya conference warned that the continent risks becoming a net consumer of AI technologies rather than a generator of proprietary innovation.

Speaking before an audience of technologists, financiers and policymakers at Nairobi’s Sarit Centre, an assembly experts painted a picture of a region rich in talent and data but structurally ill-equipped to convert either into ownership, or wealth.

“Africa and Latin America combined account for barely 1.3 percent of global PCT filings,” said Ali Jazairy, Senior Counsellor at the World Intellectual Property Organization (WIPO) in Switzerland. A show of hands from the audience confirmed his suspicion: almost no one in the room had heard of the Patent Cooperation Treaty.

“If you are not able to secure those assets, you lose them,” Jazairy warned. “Then somebody else is going to take them.”

The skills illusion: knowledge without application

The disconnect between training and deployment was laid bare early in the discussion, which brought together educators, employers, investors and IP specialists to examine the talent-to-industry pipeline.

Nikki Germany, CEO of Moringa School Kenya, which has trained more than 20,000 people in AI, data and technology, argued that the debate about “building versus using” AI requires granular thinking. “There are layers,” she said. “Integrating AI into workflows requires one skillset. Building AI-powered applications, a healthcare assistant, a tutor, requires another.”

But she acknowledged a persistent weakness: practical application. “The learning that is happening must be applied,” Germany said. “Problem identification, problem structuring, creative problem solving, actual product development, and the ability to deploy in a group environment, that is where we are lacking.”

Hartnell Ndungi, Chief Data Officer at Absa Africa Kenya, went further, drawing on conversations with data scientists he was considering for hiring earlier in the week. “Anyone who walks into a room looking for a job in data and AI has knowledge of these skills,” he said. “Knowledge has been federated. You can get certifications online. The real struggle is separating people who have knowledge from people who can actually do the job.”

Ndungi said his organisation (Absa Bank) now uses live-case assessments to test candidates’ ability to apply claimed expertise. He also warned of a growing over-reliance on generative AI tools. “People who only fully rely on AI to solve a business problem cannot alter that solution to meet a changing environment,” he said. “Make sure you also have foundational knowledge.”

The IP blind spot

Jazairy, whose organisation oversees the international patent system, delivered what many in the room later described as a wake-up call. He noted that the “great acceleration” enabled by AI has lowered barriers for startups to become unicorns: from four unicorns per year globally between 2003 and 2013 to 148 per year in the last decade. AI unicorns now reach that status in 4.7 years, down from six to seven years, and the average age of an AI founder has fallen from 40 in 2020 to 29 just four years later.

“And yet,” Jazairy said, “not once during this summit day had anyone mentioned intellectual property [IP] until now.”

He defined IP not merely as a legal shield but as a strategic currency. “IP increases the level of confidence of investors. It creates secure pathways for licensing. If you don’t have a title to your technology, you cannot exchange it. It is like a house: if you don’t have the title, you cannot rent it or sell it.”

To move the needle for Africa, Jazairy proposed three structural changes: harmonised IP laws across African countries specifically for AI and software; high-quality patent examination on the continent; and university IP policies that encourage researchers to protect results before publication, supported by technology transfer offices.

‘Anything that slows you down is your enemy’

The investor’s perspective came from Richard Waitumbi, Managing Director at Goldman Sachs USA, who laid out the unforgiving calculus of global capital. “In theory, it is a simple set of criteria,” he said. “We look for businesses with $30 million in revenues growing very quickly, or $10-20 million in earnings with steady cash flows. The main question: can this business compound equity at 25 per cent per annum?”

But in the African context, Waitumbi said, structural headwinds repeatedly undermine that target. “Currency instability. Political instability that means days taken off work. Inability to hire enough people on the tech or sales side. Anything that slows you down is your enemy.”

He pointed to the fragmentation of the continent’s market as a particular challenge. “There is a very huge market in theory, $150 million to $200 million of opportunity. But there are over 50 countries, each with different policies. Tapping that takes more time. More time hurts returns!”

On exits, Waitumbi was blunt. “IPOs happen. Unicorns are born. But the main source of exit is selling to another financial sponsor or a strategic buyer. The fewer of those buyers exist in the ecosystem, the lower the likelihood of premium returns.”

Still, he saw reason for optimism. “The world is beginning to notice. Africa has two things: land, and a large talent pool. Those are increasingly the most precious resources.”

African AI: Do ‘We prefer to give our data away?’

Perhaps the sharpest exchange came when moderator Abdelrahman Elbadawy of the Nairobi Business Angel Network asked Ndungi about Africa’s richest behavioural and financial data sets, sitting inside banks, and why they have not become fuel for home-grown AI innovation.

Ndungi’s answer was damning. “What makes me sad is that when it comes to exploiting the value of AI, we prefer to take our data and place it into commercial AI tools to generate that value. We are not creating assets or tools on top of our data. We are basically giving our data away.”

He disclosed that Absa had developed an internal AI model called Citrus, a replica of ChatGPT built inside the bank’s own warehouse, trained on transactional data without exposing it to external platforms. But he acknowledged the limits of such efforts.

“If you look at our capability as a continent to create data centres with the right GPUs to process the data we own, we are still crippled. We have only a handful of data centres in Africa. That is why most people prefer commercial tools.”

Ndungi called for a new conversation around data-sharing ecosystems with appropriate governance. “To create data assets that can compete, you need data about your customer that will not sit with one organisation. It sits across multiple platforms. Governance and data-sharing treaties are the future, and we should start talking about them today.”

Keeping IP local: carrot or stick?

In the final segment, Elbadawy asked Waitumbi to design a capital scheme that rewards founders who build and keep intellectual property in Africa.

Waitumbi proposed two models. The first was contractual: writing into purchase agreements incremental equity or management option grants if IP stays on the continent. But he acknowledged the limitation. “If it is very clear that greater wealth-building opportunity exists elsewhere, most people will see through incremental incentives and take the path to greater self-economic wealth.”

The second approach, he said, was more promising, and required government action. “Governments have the power to impact taxes. Imagine a world where companies or investors that buy businesses with active IP on the continent are given reduced tax rates if the IP stays local. Taxes take away cash flow. Lowering them creates a win-win: entrepreneurs keep IP local, and investors now have an economic reason to want it to stay.”

Germany of Moringa School added a practical note. “Not everyone can run a startup and create their own IP. Many people go into jobs and create on the side. That is why we collaborate with industry to ground our curriculum in reality.”

As the session closed, the moderator observed wryly: “I wish we had this question in the morning when more officials were attending.” The audience, by then, had no such wish. They had heard enough to know that the race for AI is not just about code or compute, but about who holds the title deeds to the continent’s ideas.