ServiceNow links Nvidia for AI control

ServiceNow used its Knowledge 2026 conference on May 5 to push a specific argument about enterprise AI: companies will not just buy models and chips, they will buy control layers. The company said its AI Control Tower is now included in Nvidia’s Enterprise AI Factory validated design, extending ServiceNow’s governance software from desktop agents to data-center-scale model workloads. Nvidia said the pairing combines its infrastructure stack with ServiceNow’s oversight tools for deployment, auditability and lifecycle management. Nvidia and ServiceNow also introduced Project Arc, a desktop agent that runs inside Nvidia’s OpenShell runtime and is governed by ServiceNow AI Control Tower. ServiceNow said the agent is meant to live on employee desktops and complete complex multi-step work across enterprise tools and systems. Nvidia described OpenShell as an open-source secure runtime that puts autonomous agents in sandboxed, policy-governed environments. (newsroom.servicenow.com) ### Why does ServiceNow care about Nvidia’s AI Factory design? ServiceNow’s May 5 announcement ties its software to Nvidia’s preferred blueprint for enterprise AI infrastructure. By getting AI Control Tower into the Enterprise AI Factory validated design, ServiceNow positions its governance layer alongside the compute, models and runtime software that enterprises use to build and run agent systems at scale. (newsroom.servicenow.com) Joe Davis, ServiceNow’s executive vice president of AI Engineering & Delivery, said the companies were linking “how work gets done with how AI runs.” Kari Briski, Nvidia’s vice president of Generative AI for Enterprise, said secure deployment of “long-running, autonomous agents” requires governance that spans models, software and AI infrastructure. Those comments show both companies framing governance as part of the production stack, not an add-on after deployment. (newsroom.servicenow.com) ### What does OpenShell actually add to this setup? Nvidia’s OpenShell is the runtime layer that constrains what an agent can do once it is operating on a machine or in a broader environment. Nvidia says every session is sandboxed, permissions are verified before execution, and the system keeps a full audit trail of allow-and-deny decisions. The company also says the runtime can enforce policies across filesystem, network and process layers. (newsroom.servicenow.com) Project Arc uses that runtime so an autonomous agent can access local files, terminals and installed applications without getting unrestricted host access. Nvidia said enterprises can define what the agent can see, which tools it can use and how each action is contained. ServiceNow’s role is to add governance, workflow context and observability through AI Control Tower and Action Fabric. (build.nvidia.com) ### Why are Red Hat and Fortinet showing up in the same story? Red Hat and Fortinet announced adjacent Nvidia partnerships on May 12, widening the same enterprise pitch from governance into infrastructure security. Red Hat said its AI Factory with Nvidia is adding OpenShell integration work, infrastructure-level oversight and confidential computing through confidential containers with Nvidia Confidential Computing in technology preview. (blogs.nvidia.com) Fortinet said on May 12 that it was accelerating FortiAIGate with Nvidia AI platforms and software. Fortinet described FortiAIGate as an AI security gateway that sits between applications and models, applying guardrails to inputs and outputs to block prompt injection, data leakage and other runtime risks. The company said the integration is aimed at zero-trust security and real-time governance across deployment models. (redhat.com) ### What are these vendors selling enterprises now? The common thread is not a new foundation model. The common thread is control over autonomous systems after they are deployed. ServiceNow is selling governance and observability, Nvidia is supplying the runtime and infrastructure blueprint, Red Hat is adding hybrid-cloud and confidential-computing controls, and Fortinet is adding gateway inspection and security enforcement around model traffic. (fortinet.com) Nvidia’s own product language points in that direction. OpenShell’s documentation emphasizes unified governance, live policy updates, private inference routing and full audit trails. Red Hat said its latest work is meant to help customers move “from experimentation to production,” while Fortinet said its gateway is built to secure “AI at runtime.” ### What should readers watch next? May 12 gave the next marker in this buildout: Red Hat said confidential containers with Nvidia Confidential Computing are available in technology preview, and Fortinet said its deeper FortiAIGate integration with Nvidia is now part of its enterprise AI push. (newsroom.servicenow.com) ServiceNow, for its part, tied Project Arc, AI Control Tower and the Enterprise AI Factory design together at Knowledge 2026 in Las Vegas on May 5. The next concrete step is whether those components move from conference-stage announcements into broader customer deployments across desktops, hybrid clouds and AI factories. (build.nvidia.com)