Editor’s note: African-Startups is a sister publication of EU-Startups, bringing trusted coverage of startups, venture capital, and innovation across Africa.
Cassava Technologies announced that it has added NVIDIA Nemotron open models to its AI Factory. Businesses and service providers can now use them through a managed service hosted locally by Cassava.
The models are available as APIs on the Cassava AI Multi-Model Exchange (CAIMEx). This lets companies pick the mix of models they need without building or running their own model-serving infrastructure. Cassava says the aim is to make AI cheaper and easier for enterprises to adopt.
“Expanding our AI Factory across our markets is about creating the capacity and economics needed for businesses to build and deploy AI solutions at scale. By bringing NVIDIA Nemotron into CAIMEx, improving the efficiency of every GPU cycle and introducing flexible token offerings, we are giving enterprises and service providers more choice in how they access and deliver these capabilities to their customers,” said Ahmed El Beheiry, Group Chief Technology & AI Officer, Cassava Technologies.
Cassava Technologies runs an AI Factory, which provides locally hosted infrastructure, models and services for businesses.
Based in the UK, it operates in Africa, the Middle East, Latin America, and the USA. The company consists of several units, including Cassava AI, Liquid Intelligent Technologies, Liquid Cloud and Cyber Security, Africa Data Centres, and Sasai Fintech, offering products and services in 94 countries.
The latest expansion is part of its plan to build an integrated AI ecosystem that gives companies more trust in how their data is secured and controlled.
NVIDIA Nemotron is built for reasoning and tool use. Customers can use it to build AI agents that handle complex tasks and connect to business tools. Its model weights, training data, and recipes are open, so developers can fine-tune it to build specialised agents for their own use.
Additionally, Cassava is using NVIDIA OpenShell to add security to agent workloads on its AI Factory.
As agents take on longer tasks and access business systems, security and guardrails need to be built into the infrastructure. OpenShell brings together model compute, controlled agent execution and independent security enforcement.
The expansion also includes the CAIMEx Token Optimiser, which aims to lower the cost of generative and agentic AI. It spots content that has already been processed and, where suitable, reuses it instead of processing it again. This cuts computing needs and helps customers get more from their AI tokens. It works with both Cassava-hosted and third-party models.
The announcement comes a month after Vodafone Business and Cassava Technologies announced a partnership to develop Egypt’s first AI factory.
