Nvidia-backed Cassava wants to fix Africa’s network downtime with AI

Cassava Technologies, the Nvidia-backed technology infrastructure company, has launched an AI-enabled network management system, Cassava Autonomous Network, to improve network performance across Africa.

The new management system is designed to independently monitor and optimise Radio Access Networks (RAN), the infrastructure in a telecommunications network that connects mobile devices to cell towers and manages how signals travel between them. 

In many African markets, managing RAN networks requires engineers to make manual adjustments when congestion occurs. Cassava said its system replaces these manual interventions with automated optimisation.

“With this solution, we are delivering on a significant step toward intelligent, self-healing, autonomous networks that drive coverage, quality, profitability, and improve customer experience across the continent,” said Ahmed El Beheiry, Group COO and Group Chief Technology & AI Officer at Cassava Technologies.

Africa’s connectivity demand is on the rise. The number of internet users across the continent surpassed 330 million in 2024 and is expected to more than double by the end of the decade. Although 4G networks remain the dominant mobile technology, 5G deployments are expanding, with 53 operators across 29 African countries offering commercial 5G services by the end of 2025. 

Operators are now managing multiple network generations, including 2G, 3G, 4G, and 5G, while facing rising customer expectations in network quality. Network downtimes and the delay it takes to repair them can affect an operator’s revenue and user count. Cassava wants its autonomous system to reduce operational expenditure for telecom operators while improving network quality.

Cassava Autonomous Network is powered by the AI infrastructure of NVIDIA, a U.S. chipmaker that invested an undisclosed amount in Cassava Technologies in October 2025. The infrastructure also operates on NVIDIA NIM microservices, allowing the deployment of the latest AI models on any NVIDIA-accelerated infrastructure, and NVIDIA Network Configuration Blueprint, an open-source model that allows telecom operators to use their own data to train AI agents. 

Cassava’s new infrastructure is also deployed on the Cassava AI Multi-Model Exchange (CAIMEx), an African multi-model exchange it launched in November 2025, giving mobile network operators (MNOs) direct access to AI tools. 

The company claims the new infrastructure replaces manual network adjustments, increases efficiency by up to 75%, and can reduce repair time for minor issues from four days to approximately 35 minutes.

Cassava said the product is vendor-agnostic, meaning it can work across equipment from multiple telecom operators and networks, including legacy, hybrid, and cloud-native deployments.

Major telecom equipment providers like Nokia and Huawei already offer AI-driven network optimisation solutions for their RAN infrastructure. Cassava said its differentiation is that it is building an African, open architecture system that works across vendors.

With NVIDIA’s backing, Cassava is betting that the next growth phase of Africa’s connectivity sector will depend on making networks intelligent enough to manage themselves.

“In today’s multi-vendor landscape, flexibility is the ultimate currency,” El Beheiry said. “Cassava Autonomous Networks provides a truly open architecture that respects existing RAN investments while introducing advanced agentic AI capabilities.”