A handful of converging trends—including a push by major industry players into African data centers, multilateral investments that reduce project risk, and vibrant grassroots research communities—are refashioning the notion that Africa can be home not only to the deployment of foreign models, but to homegrown systems that understand its languages, serve its markets and keep the value in local economies.
By Oluwatomisin Amokeoja
It’s Jade Abbott’s first time in Nigeria’s commercial capital Lagos—a long-time entry on her bucket list—and what she finds at a packed tech conference confirms her instincts.
“I am astounded by the scale. From a startup perspective, everyone here is trying something. Everyone is pushing,” Abbott, the CTO and co-founder of Africa-focused AI research and product lab, Lelapa AI, shares with FORBES AFRICA.
For the natural language processing researcher, the energy of Lagos captures something essential about Africa’s emerging artificial intelligence (AI) movement: ambition, collaboration, and urgency.
Among Abbott’s longtime collaborators is Alex Tsado, the Nigerian-born founder of Alliance4AI–a nonprofit advocating for African-led AI development, leveraging his working experience with American chipmaker, Nvidia.
Tsado says an initiative about building local graphic processing units (GPUs) is already attracting interest from development finance institutions and international philanthropies.
“The good thing is they’re beginning to see the importance of local GPUs,” Tsado tells FORBES AFRICA. “A year ago, they didn’t.”
He believes Africa’s path to what he calls AI sovereignty—the ability to control how AI is developed and used within a country—begins with building local computing power.
“With zero GPUs in your country, I don’t know how you are sovereign,” he says. “Because that means your engineers are paying in dollars to rent GPUs abroad and waiting weeks to get access.”
AI sovereignty has become a rallying cry for Tsado, Abbott, and a growing group of African AI researchers and entrepreneurs seeking to build AI for African contexts, not just with African users in mind.
Lelapa AI, based in South Africa’s biggest city, Johannesburg, builds tools for African languages. Abbott also co-founded Masakhane, a grassroots collective of researchers across the continent developing open-source language models.
Their focus is on making artificial intelligence understand, and reflect, Africa’s diversity.
“The world has been building AI models that are bigger and bigger and mostly trained in English,” she said to a room full of AI enthusiasts at the Moonshot by TechCabal conference.
“But the building blocks are wrong for African languages.”
Languages like South Africa’s Zulu or Nigeria’s Yoruba, she explained, use complex compounds and diacritics that most global AI models ignore.
“These systems just remove the accents or break words apart,” she said. “They weren’t designed for us.”
Masakhane’s work centers on creating and protecting African data. “If AI is the new electricity,” Abbott said, “then data is the oil that powers the engine. Historically, people have extracted Africa’s resources, processed them elsewhere, and sold them back to us. We can’t allow the same with data.”
To that end, these AI collaborators have launched a new data license called Esethu, which allows free research use within Africa but requires foreign companies to pay if they want access.
“We’re ensuring that our data is first-for-use for Africa,” she said.
“To get AI to work, you need several pillars—data, talent, policy, and buyers,” Tsado said at the same gathering. “But ultimately, you need compute.”
He recalled someone trying to run an AI model on a Raspberry Pi, a low-cost computer. “It almost went up in smoke,” he said. “That’s what happens when you don’t have the right infrastructure.”
Alliance4AI and its Lagos-based partner have banded together to help universities, banks, and incubators establish smaller GPU hubs that can connect through shared software—a decentralized approach to computing power.
“It doesn’t have to be a $200 million project,” he says to FORBES AFRICA. “Start with $1 million, show impact, and scale from there.”
Already, the World Bank’s private sector arm, the International Finance Corporation, is backing a $100 million investment in Raxio Group to expand data centers across several African countries, a landmark move that reflects the rising interest in local hosting and the recognition that the continent currently holds a tiny fraction of global data-center capacity.
At the same time, private firms and chipmakers are signaling a deeper commitment. Mid-2025 reports suggested partnerships and deployments that would place thousands of GPUs into African facilities, part of a broader push by companies like Nvidia and regional partners to close the compute deficit. Those moves, while nascent, mark a turning point for plans to run meaningful local models at scale.
Beyond infrastructure, the Alliance4ai FutureMakers program, a peer-learning program, helps university students teach themselves AI through structured study groups. The program combines lessons in technology, professional skills, and African history.
“At first, students wonder why we’re teaching history,” Tsado said. “But once they understand where we come from, their eyes brighten. They realize this isn’t just about getting a job abroad—it’s about rebuilding the continent.”
“The theory is that after three years, some of those students become lecturers. They start to put pressure on the universities to teach AI,” he added.
Abbott, meanwhile, warns that the continent’s growing reliance on Western AI tools could slow progress.
“A lot of money is being spent on tooling that isn’t built for us,” she says. “When those investments don’t pay off, people will think AI doesn’t work—when the real issue is that they’re not partnering with local players.”
Both Abbott and Tsado agree that Africa’s AI journey must be collaborative.
“If you want to go fast, go alone,” Abbott says, quoting an African proverb. “If you want to go far, go together.”
Source: Forbes
