Jumia Cuts Another 10% of Staff, Pivots to AI for Profitability

Jumia Technologies is cutting approximately 200 jobs, representing 10% of its current 2,000-person workforce, as the African e-commerce company accelerates its push toward profitability by the end of 2026. Chief executive Francis Dufay told Bloomberg TV that the company is now implementing AI-driven workflows across operations, logistics, finance, and marketing.

The narrative mirrors what Big Tech has been telling markets for the past year, but the underlying math has been in motion much longer. Jumia’s headcount has plummeted from 4,318 employees at the end of 2022 to under 2,000 today, marking a 54% workforce reduction over four years. The November 2025 round cut 7% of staff, and this latest round adds another 10% on top of that.

During the same period, Jumia has exited South Africa and Tunisia, lost Baillie Gifford as its largest external shareholder, and watched Rocket Internet , the German incubator that launched Jumia in 2012 , divest its remaining stake. Accumulated losses reached $2.2 billion as of December 2025. AI may be the current explanation, but the restructuring predates the buzzword.

The timing of this round makes sense against Q1 performance. Jumia reported revenue of $50.6 million for the quarter, a 39% year-on-year increase. Gross merchandise value rose 31% to $211.2 million, and adjusted EBITDA loss narrowed by 32% to $10.7 million.

Nigeria, Jumia’s largest market, saw physical-goods GMV climb 42%. Kenya grew by nearly 50%. Dufay has reaffirmed guidance for adjusted EBITDA breakeven and positive cash flow in Q4, with full-year profitability targeted for 2027. The latest cuts are sized to bridge the gap between Q1 results and that Q4 target.

Where AI is doing the heavy lifting, according to the company, is in the back office and call centre. Customer service tops Jumia’s automation list, followed by marketing operations and logistics query routing. General and administrative expenses fell 7% to $17.6 million in Q3 2025 thanks to similar workflows, and the current round aims to push that figure lower. The bet, simply put, is that Jumia can run the existing platform with 1,800 people instead of 2,000, and that automation will pay for itself before customers notice any difference.

African tech labour has absorbed AI-tagged restructuring at roughly the same rate as the US sector over the past eighteen months. Flutterwave cut around half its Kenya and South Africa staff in mid-2025 ahead of a potential IPO. Sabi reduced its team by 20% and pivoted into commodities. MAX, the Nigerian mobility-financing startup, started 2025 by letting 30% of its workforce go. Jumia’s 7% cut in November and 10% now fit squarely inside that pattern, which itself sits within the broader 2026 tech-layoff wave that has already surpassed 100,000 jobs globally.

Whether AI is the genuine cause or convenient cover is a question the global sector has faced for a year. Mark Zuckerberg told Meta staff last month that the company’s own cuts were about capex rather than AI productivity , an unusually candid admission that the AI narrative had become the easiest line to use in public statements. Jumia is smaller and has less to gain from the framing, but the pattern holds: AI is the technology being adopted, cost reduction is the strategy being executed, and the marketing language is what gets quoted.

For African tech labour, the practical question is whether the AI productivity story holds up at Jumia’s scale. The continent lacks the GPU infrastructure, model talent, and inference budgets that Big Tech relies on to back up similar claims. Replacing a Lagos call-centre seat with a large language model is cheaper and simpler than replacing a Menlo Park engineer with a coding agent , which is exactly why it happens first, and exactly why the social arithmetic looks different.

The number to watch is the Q4 EBITDA print. If Jumia hits its target, the AI framing becomes the default explanation. If it falls short, the framing becomes just one of several explanations that, on the evidence, did not suffice.