On the Gates Foundation website, Ghanaian programmer Darlington Akogo talks about his mission to reduce the number of deaths in the West African nation, where there’s one doctor per 10,000 patients. His start-up, minoHealth AI Labs, helps doctors in Accra’s Korle-Bu Teaching Hospital, the largest in West Africa, accelerate medical results. He aims to create an open-source “digital doctor” to widen access to medical care to the country’s 35 million people. Despite getting approval from the Ghana Food and Drugs Authority, Akogo’s startup relies on funding from foreign partners, such as the Gates Foundation, while the government of Ghana grapples with rising inflation and budget cuts.

Akogo’s enterprise is one of the 2,400 African companies applying AI solutions in tech, agriculture, language, climate, and many more, in a bid to catch up with the Fourth Industrial Revolution already in full swing in the US and China. But in a continent where only 5% of the people capable of driving this revolution have access to computational power for research and innovation, Africa’s path to AI success is littered with a plethora of challenges.

Recently, more than 1000 AI enthusiasts from more than 95 countries converged on Kigali for the first-ever Global AI Summit on Africa. Organized by the World Economic Forum, which estimates the continent will get a miserly $2.9 trillion from the $19.9 trillion AI will generate by 2030, keynote speakers included Rwandan President Paul Kagame and his Togolese counterpart Faure Gnassingbé. Kagame called for investment and commitment. “Our strategy should be to go back to the drawing board and build a stronger foundation for connectivity,” he urged delegates at the summit. “Africa cannot afford to be left behind, once again, playing catch-up.”

Since 2022, when OpenAI rolled out ChatGPT and marked the turning point for AI, ushering in global adoption that has led to groundbreaking innovation and giant investments, there have been concerns that Africa will be left behind. Since the onset of industrialization in the 1800s, Africa has been playing catch-up due to myriad barriers, many of them intentionally imposed and maintained through colonial and post-colonial administration. In the advent of AI, Africa is still to get up to speed.

Following the two-day summit, attendees committed to a $60 billion fund, almost the size of all goods and services produced by Uganda, Africa’s 13th-largest economy, in 2024. The fund will be deployed to boost AI architecture, support African start-ups and machine learners, and encourage AI research domestication, the declaration document said.

Strive Masiyiwa—the Zimbabwean entrepreneur and head of Cassava Technologies, a company behind the continent’s first AI factory—told delegates his company will receive the first 3,000 units of the technology in May. The units will be mounted in Cassava’s data center in South Africa in June. Deployments of 7,000 more units will be made in Kenya, Nigeria, Morocco, and Egypt.

Artificial Intelligence happens when large datasets are collected, computed, and trained by using step-by-step processes (algorithms) aimed at solving problems or performing tasks. This requires graphic processing units (GPUs) that can perform quadrillions of calculations and functions within seconds. Nvidia, the American chipmaker, produces 85% of the GPUs that do the work. Alex Tsado, co-founder of Alliance4AI, an organization that rallies African AI engineers, told CNN that innovators’ access to the supercomputers will turbocharge computing power on the continent. Computer scientists will be encouraged to collect more data because they will be able to transform it, and there will be broad machine learning programs, he said.

Cassava Technologies might be paying up to $40,000 per Nvidia GPU to acquire some of its most potent AI-propelling hardware. That amount is almost 20 times the average yearly African household income, according to a World Bank report in 2023.

Although Google and Microsoft remain the biggest AI players in Africa, with data centers in Ghana, Kenya, and South Africa (Google has committed to $1 billion for digital infrastructure), tailoring technology to meet African needs remains a big challenge. For instance, Google, which holds more than 96% of the search engine market equity in Africa, only provides translation to 60 out of the more than 2,000 languages spoken by the continent’s 1.5 billion inhabitants. To a greater degree, there have been concerns about data mined by these giant companies following a 2018 Facebook data breach, and most of the servers are housed in the US, creating jurisdictional problems for African regulatory bodies.

“When you don’t build the solutions that you use in AI… there’s a high risk that (they) can harm you,” Tsado said, in an interview with Impact Newswire. The Nigerian worked on Nvidia’s product team, where he led cloud computing projects for Microsoft Azure, Google Cloud, and AWS. But he left in 2020. The company’s AI models at that time failed to recognize minority groups. Tsado believes Africans ought to design their own models. He hypothesized some AI discrepancies in the US:  “They build AI tools to become like the HR (Human Resources), to select people for jobs, (the tools) won’t (select) women (for)jobs in engineering, or (the AI models would) think that black people were (not) smart enough to get those kinds of jobs.”

To develop minoHealth AI Labs’ medical AI models, Akogo used data from Ghana, the US, and Vietnam. He crunched them from three races in three continents to prove the system’s reliability. This is similar to OpenAI, which harnessed texts and languages from different parts of the world to launch ChatGPT. Computing large datasets requires high-speed internet. Africa’s internet penetration rate stands at 43%. Morocco, Egypt, South Africa, Kenya, and Nigeria have the lion’s share. This means that most regions face the risk of missing out.

Take, for example, the Central African Republic (CAR). It has an abysmal 7.1% penetration and yet joined the AI pledge in Kigali.  It must drastically improve its satellite and optic-fiber networks to catch up. Despite the significant challenges CAR faces, from low connectivity to an electricity grid that a privileged 14.3% of Central Africans have access to, President Faustin-Archange Touadéra’s incessant belief that digital currency will usher in economic prosperity to the country is intriguing. He started a memecoin in February; its value plunged to 95% before it could mature. This was nearly three years after he had initiated bitcoin, becoming the second nation after El Salvador to make the currency a legal tender. In the bitcoin assessment document, seen by this reporter, the team proposed energy and connectivity improvements that CAR, it would appear, did not adhere to.

The CAR’s case provides an overview of how African leaders believed digitization, and now AI, would miraculously boost their economies without structural reforms. The economist and Nobel Prize winner Angus Deaton, in his book The Great Escape, lays down the marker for national development: “Economic growth requires investment in things—more machines, more basic facilities like highways or broadband—and in people, who need more and better education.” Deaton argues that an investment in research and development drives innovation. He further states that the people leading the innovation should not be subjected to expropriation, skewed judicial systems, and arbitrary taxes, all hallmarks of most African states.

The need for the basic facilities Deaton urges is reflected in Aliko Dangote’s struggle to build a $20 billion mega oil refinery in Nigeria. He established his own energy infrastructure to power construction, built a port to import the components he needed, and erected a heavy-duty vehicle plant that provided trucks to carry equipment. “In Africa, there is no infrastructure,” Dangote said, claiming it was easier to build such a project in Asia or the Middle East.

At the Global AI Summit, Masiyiwa said his other company, Liquid Intelligent Technologies, had laid down 120 kilometers of optic fiber, spanning Cape Town, South Africa to Cairo, Egypt. A chunk of workers at Cassava Technologies have been focusing on AI for the past seven years, he said. The speculation is that GPU acquisitions from Nvidia would help boost internet services provision.

AI’s huge potential, like the internet, will pose a test to African governments during elections and political unrest. “It will be a dangerous place if we involve Artificial Intelligence too much in our politics,” Kagame warned at the conference. Last year, protests and conflicts caused 15 African countries, including Kenya, Chad, Comoros, and Ethiopia, to either shut down or disrupt the internet. The practice has been a mainstay for the past decade in most African nations, including Gnassingbé’s Togo, which discontinued services when Togolese protesters stood up against a constitutional reform that extended his stay in power.

The summit did not detail where the $60 billion fund will come from. However, like the 2001 Abuja Declaration, when African states pledged to spend 15% of their budgets to elevate healthcare (recent data shows that only South Africa and Cabo Verde are living up to expectations), funding for the AI-for-good initiative might just rest in the hands of powerful multinationals. Black Rock, Microsoft, and the Abu Dhabi-funded investment firm MGX have already agreed to pool their resources in a $100 billion global AI investment fund. Though it mainly focuses on establishing data centers and energy infrastructure in the US, this could extend to Africa, in what many critics see as the West’s entrenchment of digital colonialism.