A Digital Awakening Faces an Uneven Road

Jan 10 , 2026. By BEZAWIT HULUAGER ( FORTUNE STAFF WRITER )

Ethiopia’s artificial intelligence (AI) ecosystem has evolved from a fringe experiment to a slowly expanding frontier, spearheaded by a handful of pioneering tech firms and endorsed at the highest levels of government. However, the sector continues to struggle with structural constraints, talent gaps, and market fragility, reports By BEZAWIT HULUAGER, FORTUNE STAFF WRITER

Over the past years, Ethiopia’s artificial intelligence (AI) ecosystem has quietly emerged from obscurity into the public spotlight, led by pioneers such as iCog Labs and newcomers like Addis AI and Hassab.

Hiruy Tsegaye is a product manager at iCog Labs, best known for helping develop Sophia, the humanoid robot capable of speaking Amharic with a sense of humour. He began working on Artificial Intelligence (AI)-related projects 15 years ago, a time when no Ethiopian company was solely focused on the field.

Before the COVID-19 pandemic, iCog Labs reached peak operations mainly through international outsourcing. The pandemic, however, took a toll. The company’s headcount dropped from 154 employees to 55, while the number of freelancers shrank by half. Four active projects continue today, but the sustainability of this model remains uncertain. Industry experts warn that within five years, AI may be able to code independently, raising the possibility that programming itself could become a machine-driven task.

Previously, iCog collaborated with foreign firms on advanced concepts. Now, it is shifting focus toward the domestic market, with a possible strategic shift under review.

For nearly a decade and a half, a handful of local IT companies like iClog have been quietly exploring and experimenting with AI, developing systems and solutions long before AI became a global buzzword.

“Since we're early birds, we're able to work on complex projects,” Hiruy said.

One current project iCog targets is misinformation, with AI systems analysing the relationship between claims, checking facts, rating sources, and studying how information spreads. The system also detects whether content has been cited elsewhere. Data shows that 40pc of content produced in 2023 was generated by AI, a share expected to grow as both creation and reaction increasingly move to machines.

“Accuracy is becoming a currency in its own right,” Hiruy told Fortune.

Much of the early work by these companies unfolded out of the public eye, even as government officials, including Prime Minister Abiy Ahmed (PhD), began championing an ambitious Digital Ethiopia agenda. The strategy, endorsed by the Council of Ministers and presented as a linchpin for national transformation, puts technology at the centre of reform efforts.

However, their ambitions often clash with stubborn realities. Resource constraints, a shortage of technical talent, weak demand, and high operating costs remain persistent. Only a few players address daily, applied use cases. Others attempt more complex work but struggle without strong institutional support.

The pandemic’s impact sped up the shift to remote work and contract-based employment, which, while lowering costs for enterprises, has made it harder for companies like iCog to retain clients. Skilled talent is expensive, as many specialists now contract directly with global firms, and training is rarely feasible for the company’s specialised work.

“Expensiveness has become the new logo for freelancers, even though demand is low,” Hiruy said, noting that hourly rates now stand at 25 dollars. “Despite our reputation, we don’t have our own product. That will be the next move.”

As government services migrate online, the banking industry is experimenting with AI-driven credit scoring. Artificial intelligence is gradually gaining a foothold, pushing a younger generation of tech enthusiasts and established industries toward new models. Still, Ethiopia ranks among the world’s least-exposed countries to AI, according to the World Bank, a verdict that veterans of the sector describe as discouraging.

Another initiative decentralises cloud data storage, letting individuals rent out unused computing power. The AI system locates idle capacity on computers and smartphones using geolocation, enabling users to earn monthly rent. The project, built for Belgium-based Nunet, leverages the untapped potential of more than a billion smartphones and half a billion laptops worldwide.

In healthcare, iCog Labs is developing AI to track movement, meals, and other behavioural patterns to enable predictive diagnosis and preventive care.

A year ago, Addis AI entered the scene, co-founded by Biniyam Daniel and Kalkidan Demele, both software engineers. Their idea grew from a hiring challenge; foreign candidates lacked technical depth, while locals struggled with English in interviews. They chose local hires and delivered the project, which inspired a broader solution. They want to teach communication, presentation, and language skills through AI. The founders built a demo app that coached users in Amharic on interviews and CV writing.

Their initial target was modest, though. Initially, 5,000 university students were planned to test the app. Instead, more than 10,000 signed up in a week, crashing the servers. Addis AI has since evolved into a start-up providing Amharic-language AI assistance similar to Apple’s Siri or Google’s Gemini, with features such as image recognition, document scanning, note-taking, and voice-to-text conversion. Now open source, Addis AI connects with 450 developers and supports 10,000 daily users, reaching a total of 95,000.

The company has since shifted its focus to rural areas and farmers, teaming with the Agricultural Transformation Institute (ATI) to deliver Amharic-language support for health, agriculture, and education. Between 12pc and 14pc of Ethiopians are estimated to speak English as a second language. Hence, Addis AI wants to close the gap, offering advice in local languages. Afaan Oromo integration is underway, and expansion into Swahili, Tigrinya, and other African languages is planned. Dialect differences were a challenge, but after a year of research, accuracy now reaches 97pc for Amharic and 93pc percent for Afaan Oromo.

The platform remains self-funded, and skilled labour shortages continue, even after government-backed coding initiatives. Addis AI now targets customer service, ride-hailing, and general users, with new users getting 500 Br in free credit and language tokens priced at 30 cents per thousand.

Biniyam intends to make Addis AI fully open source.

“There’s no competition when you can count the companies on one hand,” he told Fortune.

The sector's trajectory remains one of promise, entangled with persistent structural impediments. According to the World Bank’s Digital Progress & Trends Report for last year, Ethiopia is in the lowest quartile for AI readiness, citing infrastructure gaps and limited adoption. Problems such as unreliable electricity, talent shortages, fragmented data, and dependence on foreign technology remain major obstacles and could deepen inequality.

Internet access remains a barrier. In 2024, internet penetration was around 35pc, serving roughly 40 million of Ethiopia’s over 100 million people. Smartphone ownership was expected to reach 58pc by late 2025, but access to advanced digital tools is largely restricted to urban populations.

While Ethiopia’s inclusion among supported countries has eliminated the need for VPNs, user numbers likely remain in the low millions to use international platforms, such as ChatGPT, which holds a 67pc share among AI tools in mid-2025, followed by Microsoft Copilot, Perplexity, and Google Gemini.

Hassab AI, run by Kidus Yared, a software engineer, began with a more focused goal. He plans to generate accurate Amharic subtitles. By May 2025, the company had moved from testing to enterprise-level deployment, serving five major clients with speech-to-text tools for media, call centres, and audio professionals. Hassab supports Amharic, Afaan Oromo, and Tigrinya (in beta), with plans to add Somali and Afar. The team spent over 3,000 hours training the system across three years, growing from two to 10 staff.

Human capital is a limiting factor, though negotiations with potential investors are ongoing.

Elsewhere, Mikael Hiruy, founder of Andalem AI, sees opportunity in customised AI agents, particularly for capital market regulations and legal advisory tools. He views AI as “hallucinative,” prone to inaccuracies without strict constraints. Data localisation rules and the lack of domestic data centres complicate deployment, though Amazon has provided 100,000 dollars in cloud credits. Prototypes are expected in the coming months.

Having a stated ambition is to make Ethiopia Africa’s centre of excellence for AI by 2030, the federal government’s Digital Ethiopia 2030 strategy formally endorses AI, setting out plans to upgrade infrastructure and boost internet penetration to 50pc this year. Investments include GPU clusters and data centres to support startups, students, and public institutions. The Artificial Intelligence Institute has launched digital court systems, “Mesob” service centres, AI-powered medical diagnostics, telehealth, crop disease identification, and smart city tools. It partners with more than 200 start-ups, has completed nearly 60 research projects, and operates a 16,000-terabyte data centre.

Tesfaye Zewde, the institute’s communications chief, summed up the mission as “AI for All.”

According to Kal Kassa, consultant and founder of Bitbirr, data sharing and national data centres are the bedrock of future innovation.

“Talent pipelines matter, but so do cybersecurity and data sovereignty,” he told Fortune.

Although the possibilities are clear, the path, for now, remains uneven.

PUBLISHED ON Jan 10,2026 [ VOL 26 , NO 1341]