African AI startups are increasingly developing their own technical talent because they struggle to recruit experienced engineers in a market where international companies can offer significantly higher salaries.
Rather than wait for qualified specialists to become available, founders are hiring graduates and junior developers and training them to build and manage artificial intelligence systems. However, the approach comes with a major financial challenge: the computing resources needed to train AI engineers are often priced in US dollars, making them expensive for businesses earning revenue in local currencies.
The challenge is highlighted in AI Made in Africa: Voices from AI Businesses Across Africa, a report published by the German development agency GIZ and research firm Briter in October 2026. The study maps 311 AI businesses across the continent and draws on a survey of 85 companies conducted between April and July.
Experienced AI professionals remain difficult to find
The report identifies a shortage of specialised workers as one of the biggest obstacles facing African AI businesses.
More than half of the surveyed companies, 51%, said senior AI research positions were the hardest roles to fill. Machine learning operations and infrastructure specialists followed at 33%, while 23% identified data engineering positions as difficult to recruit for.
The report estimates that Africa accounts for only 1% to 3% of the world's AI talent. It links the gap partly to limited access to higher education and the relatively small number of universities offering specialised AI programmes.
Sub-Saharan Africa has a university enrolment rate of about 9%, compared with a global average of more than 40%, according to the report.
Competition from international employers makes recruitment even harder. Companies outside the continent can offer salaries that many local startups cannot match, particularly for experienced professionals who can work remotely.
One example is Danaya, a compliance technology startup in Côte d’Ivoire, which reportedly had to recruit a Cameroonian engineer who had built his career in France to fill its chief technology officer position because it could not find an equivalent candidate locally.
Startups are turning junior developers into AI specialists
With experienced professionals difficult to recruit, many founders are investing in training instead of relying entirely on external hiring.
The survey found that 47% of AI businesses have established formal training programmes, through which they recruit junior engineers and develop their skills. Another 37% work with local AI bootcamps, while an equal proportion recruit talent from universities.
Some companies are already applying this approach.
Kenya's Acre Insights recruits graduates with backgrounds in mapping, computer science and data modelling before providing intensive on-the-job training.
In Nigeria, MyItura takes a similar approach by hiring local junior developers and providing them with continuous, structured training.
The strategy can give startups employees who understand their products and systems more deeply than outside recruits might. However, it also means businesses are taking on part of the responsibility for developing a skilled workforce that universities and governments would ordinarily help provide.
Training costs are rising with cloud computing expenses
Developing AI expertise requires more than classroom instruction. Engineers need access to computing infrastructure to train models, run experiments, and test applications.
This creates another challenge for startups operating with limited budgets.
About 65% of surveyed businesses identified cloud and computing resources as their biggest development expense, while 61.1% described high cloud costs as their leading infrastructure challenge.
The financial pressure is particularly significant because most cloud services are billed in dollars.
The report illustrates the effect of currency depreciation in Nigeria: a cloud service costing $1,000, which was worth approximately ₦471,000 in early 2023, had risen to around ₦1.53 million by mid-2025.
At the same time, 56% of the surveyed businesses charge less than $2,500 annually per customer. This creates a difficult imbalance for companies that earn relatively modest revenue but must pay international prices for the infrastructure needed to develop their products.
Some startups are using partnerships and cloud credits to reduce the burden. MyItura, for instance, says it has secured partnerships that provide access to cloud credits for training and running its AI models. Skilladder has also sought partners to help subsidise its cloud subscriptions.
However, infrastructure costs can increase as an AI company attracts more users and expands its services, making it difficult to keep expenses under control as the business grows.
Renting AI models offers a shortcut, but creates another expense
Startups that lack the expertise or resources to develop their own models can use third-party AI services instead.
This allows businesses to build products using existing models without immediately hiring large teams of AI researchers or investing heavily in their own computing infrastructure.
However, relying on external models can become expensive as usage increases.
The report cites estimates of between $20,000 and $50,000 in monthly fees for companies serving thousands of daily users. These figures come from a US source rather than data collected specifically from African businesses, so they should not be treated as a direct measure of costs across the continent.
Moving away from rented models can also be difficult for startups that have built their products around external services.
Founders call for computing subsidies and research support
The report suggests that developing Africa's AI workforce will require more than encouraging startups to train graduates.
Around 64% of surveyed businesses said they need research and computing subsidies to support their operations.
Such assistance could reduce the cost of experimenting with AI models, provide more affordable access to computing infrastructure and help startups develop technical expertise locally.
The report also highlights the importance of retaining experienced professionals. Competitive salaries matter, but skilled workers also want opportunities to build products, solve difficult problems and see their work reach a wider market.
Some initiatives are already providing support. Google has committed $2.1 million to AI talent programmes in Nigeria, while MTN has announced plans to offer cloud services priced in naira at discounts of 15% to 20% compared with international providers.
Nevertheless, limited electricity supply and the high cost of developing local computing infrastructure remain significant obstacles.
Africa's AI ambitions depend on developing local talent
The findings show that African AI startups are not simply waiting for the workforce they need to emerge. Many are recruiting inexperienced workers and investing in the training required to turn them into specialists.
But the approach puts additional pressure on businesses already dealing with expensive infrastructure, limited funding and competition from international employers.
The report does not provide a separate breakdown of how many Nigerian startups are training their own engineers or the average monthly cost of these programmes. It also leaves unanswered how often trained employees leave for better-paying opportunities once they gain experience.
Source: Giz