SUSE pre-validates AI stack

SUSE has launched an on-premises artificial intelligence package with Nvidia that is meant to arrive pre-checked for enterprise deployment. (suse.com) The product, called SUSE AI Factory with NVIDIA, was announced at SUSECON in Prague on April 21 and combines SUSE AI with NVIDIA AI Enterprise in a single reference stack. SUSE said it is aimed at moving customers from local testing to production across data centers, edge sites and cloud environments. (suse.com) SUSE said the stack includes NVIDIA NIM microservices for model serving, open Nemotron models, NVIDIA NeMo tools for building agents, Run:ai for graphics processor orchestration, Kubernetes operators, OpenShell secure runtime and NemoClaw for agent deployment. SUSE said those pieces use its K3s Kubernetes distribution as part of the reference design. (suse.com) Artificial intelligence “models” are the systems that generate answers, while “agent” software lets those models take actions such as calling tools or handling multi-step tasks. Nvidia describes NemoClaw as an open-source stack that installs Nemotron models and the OpenShell runtime together so those agents can run with policy controls and local data handling. (nvidia.com) SUSE and Nvidia are pitching that setup to companies that do not want sensitive data sent to a public cloud. SUSE’s launch materials say the package is built for “strict data sovereignty requirements,” and review coverage described it as a fit for regulated industries that need tighter control over infrastructure and information. (suse.com; thenewstack.io) That pitch lands as more companies try to move generative artificial intelligence from pilots into governed production systems. Compare the Cloud, citing the SUSECON announcement, described the bundle as a prescriptive platform for building and running production workloads rather than a stand-alone model release. (comparethecloud.net) SUSE has been building that message for months. In March, the company said it was working with Nvidia on autonomous agents for enterprise artificial intelligence and highlighted NeMo Agent Toolkit for dialogue management and safety, plus NemoClaw for safer deployment of always-on assistants. (suse.com) Nvidia’s own documentation says NemoClaw is still alpha software and has been available as an early preview since March 16, 2026. That means part of the newly packaged stack includes components that Nvidia is still labeling as early-stage, even as SUSE presents the overall bundle as pre-validated for enterprise use. (nvidia.com) SUSE’s argument is that packaging, support and policy controls will matter as much as model performance for buyers in finance, government and other tightly regulated sectors. The company opened with a stack for customers that want artificial intelligence kept inside systems they already control. (suse.com)