Meta has expanded its regional Sub-Saharan Africa LLaMA Accelerator across tech innovation hubs in Lagos, Nairobi, Cape Town, and Dakar, committing grant funding, cloud credits, and engineering mentorship to founders building with open-weight AI.

The accelerator's centerpiece is a concerted push into low-resource African languages. By fine-tuning LLaMA foundation weights on high-fidelity corpora of Igbo, Wolof, Zulu, and Luganda, African startups are creating automated legal advisory, voice-driven agricultural diagnostics, and micro-insurance chatbots accessible to non-English speakers.

A newly released economic assessment commissioned by Meta forecasts that open-source artificial intelligence adoption could infuse over R520 billion ($28B USD) into South Africa's economy alone over the next decade, with small and medium-sized enterprises reaping the primary benefits.

'Closed, proprietary models risk locking Africa into exorbitant licensing costs and foreign cultural biases,' said a lead researcher at the African Open Science Hardware community. 'Meta's open-weights strategy gives our engineers the agency to inspect, modify, and own our AI systems.'