BEIJING, Sept 12 (China Economic Net) - In May, Ethiopian academic Kedir Yassin Hussen spent six days teaching colleagues and students in a computer laboratory at the University of Gondar. He had just returned from an AI computing programme in China and was bringing its curriculum home, according to FlagOS, the software project used in the programme.

It was a small step towards a large problem. The African Union’s continental AI strategy identifies shortages of skilled people and computing resources as barriers to wider adoption. The question for China-Africa cooperation is whether it can help local institutions keep teaching, adapt AI to their needs and sustain the services they build.

Nigeria is one of the countries seeking that help. Johnson Tamunopreye Bareyi, director of e-government at Nigeria’s communications ministry, told China Economic Net that Nigeria wanted to work with China, which he called a global leader in AI, to build local capacity in both people and infrastructure. He was speaking on September 9, on the sidelines of the China International Fair for Trade in Services in Beijing.

The ministry already worked with Chinese counterparts on the green and digital economy, and its permanent secretary had been invited to China a few months earlier, Bareyi said. Nigeria intended to make AI development part of that cooperation.

The country had drawn up a national AI strategy, he said, and now needed to put the technology to use, particularly in government services. Banks and other private companies already used AI, though there was scope to do much more.

The strategy calls for affordable local computing capacity, teacher training and applications in government, healthcare and agriculture. Asked how much of that computing capacity had been built or deployed, Bareyi pointed to skills programmes such as the 3 Million Technical Talent programme, known as 3MTT. He gave no figure for capacity already in place.

One model for such cooperation is the programme that took Kedir to Beijing. The AI Compute Faculty Development Initiative, run by the Beijing Academy of Artificial Intelligence, or BAAI, with Peking University and the AU’s African Scientific, Research and Innovation Council, known as ASRIC, combines training with access to computing resources so that lecturers can teach AI at their own universities. Its first cohort, in April, drew lecturers from ten African countries, according to ASRIC. No Nigerian university was among the beneficiaries.

In Beijing, the lecturers studied processor architecture, compiler optimisation and distributed training, with hands-on work on FlagOS, BAAI’s open-source software stack built to run AI across different kinds of chips. They are expected to develop courses for students and colleagues once home.

Kedir’s sessions in Gondar drew on that curriculum and ran on FlagOS computing resources. Some first-cohort teachers had already begun teaching what they learned, ASRIC vice-chair Khaled Ghedira said at the World Artificial Intelligence Conference in Shanghai in July.

His research points in the same direction: making more of what is already there. A paper on running AI in older data centres, Efficient AI Deployment on Legacy Data Centers in the Global South, won a best paper award in the lightning-talk session of the GlobalSouthML workshop held alongside ICML 2026 in Seoul, South Korea, the organisers said.

That matters to universities with limited budgets, and it is part of the programme’s design. Lin Yonghua, BAAI’s vice-president and chief engineer, told the Shanghai conference that the aim was to bring idle or underused hardware together under a common software stack and make it available for education and research in developing countries.

Access to machines brings work of its own. Fitsum Assamnew Andargie, who leads the RESONANCE AI4D Lab at Addis Ababa University, told the same forum that his students had started out on laptops and services such as Google Colab. When the lab gave them access to its own servers, demand rose quickly, and the lab needed people and procedures to allocate computing time and coordinate users.

Deployment was another gap, Fitsum said. Researchers spent more time on experiments than on turning results into services that other people could use.

The training should not stop when participants go home, Ghedira said. Sustaining it would take ownership by African institutions and long-term partnerships, and agreements between BAAI and several African universities had been signed or were in progress.

In Gondar, the teaching has already spread beyond the group trained in Beijing. The longer test is whether local teams can keep teaching, adapting software and maintaining services. That is what Nigeria will need too, as it tries to put AI to work in government and industry.