Kenyan AI Startup AfriLang Raises $15M to Build Language Models for 50+ African Languages

AfriLang AI, a Nairobi-based artificial intelligence startup, has closed a $15 million Series A funding round led by Andreessen Horowitz (a16z), with participation from TLcom Capital, Novastar Ventures, and several African venture capital firms. The company is building large language models (LLMs) specifically designed for over 50 African languages.

Founded 18 months ago by Wanjiku Kamau, a former Google research engineer who spent six years working on multilingual AI at Google Brain in Mountain View, AfriLang has assembled a team of 45 researchers and engineers across offices in Nairobi, Lagos, and Addis Ababa.

The startup's core product, AfriLang-7B, is a 7-billion parameter language model trained on a proprietary dataset of 2.3 trillion tokens collected from African-language sources including news outlets, government records, academic publications, radio transcripts, and community-contributed text. The model currently supports 54 languages including Swahili, Yoruba, Amharic, Wolof, Hausa, Zulu, Igbo, and Lingala.

"The problem with existing AI models is that they treat African languages as an afterthought — a translation layer bolted onto English-centric architecture," Kamau explained during a demo at the Africa Tech Summit in Nairobi last week. "We built our system from the ground up with African languages as the foundation. The result is 3-4x better performance on comprehension, generation, and cultural context compared to GPT-4 and Claude on African language benchmarks."

Independent evaluations conducted by researchers at the University of Cape Town's Natural Language Processing Group confirmed AfriLang's claims. In tests across 12 African languages, AfriLang-7B outperformed OpenAI's GPT-4o by an average of 47% on cultural context understanding and 38% on idiomatic expression generation.

The company plans to use the new funding to expand language coverage to 75 languages by mid-2027, hire 30 additional machine learning engineers, and launch commercial APIs for businesses operating across Africa. Early enterprise customers include Safaricom (for customer service automation), the African Union (for document translation), and M-Pesa (for voice-based financial services).

"Africa represents 2,000+ languages and 1.4 billion people, yet receives less than 1% of global NLP research investment," said Maureen Wanjiku, partner at TLcom Capital. "AfriLang is building the foundational AI infrastructure that the continent needs to participate meaningfully in the AI economy."

The Series A brings AfriLang's total funding to $18.5 million, following a $3.5 million seed round in 2025 led by Founders Factory Africa. The company is targeting profitability by Q3 2027.

Source: AfriLang AI press release; Africa Tech Summit Nairobi 2026.