Cassava Technologies announced that its AI Factory will host Nvidia’s open‑source Nemotron models as a managed service, allowing businesses to access the models through APIs rather than building their own infrastructure.

According to Cassava, Nemotron is designed for reasoning and tool use, capabilities needed for AI agents that perform multi‑step tasks and interact with corporate systems.

Nvidia has released the model’s weights, training data and methods, enabling developers to fine‑tune the models for specific applications.

The service is delivered via the Cassava AI Multi‑Model Exchange (CAIMEx), which also includes a Token Optimiser that reuses previously processed content to reduce token usage across both Cassava‑hosted and third‑party models.

Cassava is integrating Nvidia OpenShell to combine compute resources, controlled agent activity and independent security enforcement, addressing concerns about AI agents accessing business systems.

The rollout builds on Cassava’s AI Factory, first announced in March 2025 with data centres in South Africa, Egypt, Kenya, Morocco and Nigeria, all intended to keep data within Africa’s borders.

Cassava’s Group Chief Technology & AI Officer Ahmed El Beheiry said the expansion aims to provide the capacity and economics needed for businesses to build and deploy AI solutions at scale across its markets.