SANTA CLARA, October 26, 2026 — Nvidia has shifted from a graphics card maker to an enterprise AI fortress builder by wrapping GPUs, BlueField DPUs, and Nemotron models into a full-stack platform that lets companies run AI agents behind their own firewalls.
The Shift From Gaming To Enterprise Infrastructure
Nvidia Corporation is an American multinational technology company headquartered in Santa Clara, California. The firm develops graphics processing units, systems on chips, and application programming interfaces for data science, high-performance computing, artificial intelligence, and mobile and automotive applications.
Founded in 1993 by Jensen Huang, Chris Malachowsky, and Curtis Priem, Nvidia has been described as a Big Tech company. Originally focused on GPUs for video games, the company broadened their usage into other markets, including AI, professional visualization, and supercomputing.
The first problem was who would quit first. Jensen Huang’s wife, Lori, did not want him to resign from LSI unless Malachowsky resigned from Sun at the same time, and Malachowsky’s wife, Melody, felt the same way about Huang. Priem broke that deadlock by resigning first from Sun, effective December 31, 1992.
In 2025, Nvidia became the first company to surpass US$4 trillion and US$5 trillion in market capitalization, largely driven by the AI boom and soaring demand for its AI data center hardware. This financial milestone marks the transition from a niche gaming player to a dominant force in enterprise infrastructure.
Hardware-Isolated Security With BlueField And Sentry
Nvidia Sentry is a security software framework that monitors AI agents and enforces policies governing their access to models, tools, and data. Part of the Nvidia OpenShell reference architecture announced at GTC Berlin, it is built on Nvidia DOCA, runs on Nvidia BlueField-4 DPUs, and provides host-isolated runtime protection for AI agents.
Nvidia Sentry validates that an agent operates according to its designed intent and within its defined operational boundaries. Trusted even when host resources cannot be trusted, it controls the agent’s only path to the model, providing tamper-evident traceability of agent activity while enforcing granular security controls on agent access in silicon.
Nvidia Sentry uses out-of-band observability to establish real-time situational awareness, continuously evaluating agents’ runtime security and behavior to validate their security and integrity. This post explains how Nvidia Sentry evaluates agent requests, detects threats, and enforces security policies independently of host software.
Software runtime boundaries remain essential. Nvidia OpenShell is a secure runtime for autonomous agents that governs what an agent can do. OpenShell separates policy enforcement from the agent workload, so a compromised agent cannot change its own restrictions.
DOCA Tower is Nvidia Sentry’s AI security gateway. Built on Envoy, the open source edge and service proxy, it operates as a transparent proxy for agent traffic, placing a bidirectional inspection layer between an agent and the external environment.
Open Models And On-Premise Deployment
The new suite, called Cadence Maestro, introduced on Monday, was built in part on Nvidia’s open-weight Nemotron model and secured with Nvidia’s OpenShell safety technology. Cadence said customers can run the models on its own cloud or any public cloud, or keep them entirely in-house behind their own firewalls and data centers.
Cadence customers can run models on-premise behind firewalls, contrasting with Synopsys-OpenAI cloud-hosted approach. They can also fine-tune the models on their own chip designs, and early results show up to 30% higher quality and up to five times better cost-efficiency versus general-purpose models.
Cadence SVP Paul Cunningham framed Maestro as model-agnostic, noting there are more AI model providers out there than chip design software vendors. When asked whether Cadence might strike its own deals with proprietary AI labs, Cunningham said both open- and closed-weight approaches are being pursued equally.
Enterprise Adoption And Strategic Partnerships
Cadence Design Systems launched a portfolio of AI models tuned to help design computer chips, deepening its partnership with Nvidia. Both companies make software used to design computing chips — a process that involves arranging billions of transistors precisely to meet cost and performance targets.
The first Maestro models reach select early-access customers in the fourth quarter. Cadence said they can run the models on its own cloud or any public cloud, or keep them entirely in-house behind their own firewalls and data centers.
What This Means For The Enterprise AI Market
Nvidia is positioning itself as the sole provider of a complete security stack for enterprise AI agents. By combining BlueField-4 hardware trust domains with OpenShell software runtimes, Nvidia ensures that even if an agent is compromised, it cannot bypass its own restrictions or access models without Nvidia Sentry’s independent hardware safeguards.
This architecture allows companies to deploy advanced AI agents in high-security environments like chip design firms and banks while maintaining strict control over data sovereignty. The trend underneath is a move from cloud-hosted AI to on-premise, secure-by-design infrastructure that prioritizes enterprise autonomy and security.
Watch this space: as AI agents become autonomous workforce members, the ability to trust them without trusting their host environment will define the next decade of enterprise computing. Nvidia’s strategy of hardware-isolated security and open-weight models positions it to dominate not just by selling chips, but by controlling the very rules that govern how those chips are used in critical industries.
BlueField-4 DPUs And Hardware Trust Domains
Nvidia BlueField-4 DPUs provide the hardware foundation for Nvidia Sentry’s security architecture. According to Nvidia technical documentation, BlueField-4 runs security services in a hardware trust domain independent from the host server. This design ensures that even if the host operating system or applications are compromised, the BlueField processor maintains a separate, trusted execution environment. The DPU sits on the node’s only path to the model, intercepting all agent-to-model communications before they reach the host CPU. This hardware-isolated approach creates a strict security boundary that agents cannot perceive or bypass, allowing Nvidia Sentry to enforce access policies independently of host software and provide tamper-evident traceability of all agent activity.
Nvidia OpenShell functions as a secure runtime for autonomous agents, governing what an agent can do within its operational boundaries. According to the Nvidia OpenShell reference architecture, OpenShell separates policy enforcement from the agent workload, preventing a compromised agent from changing its own restrictions. In the reference architecture, agents execute on Nvidia Vera CPUs while BlueField-4 sits on the node’s only path to the model. Nvidia Sentry inspects agent-to-model communications and enforces access policies without relying on the agent to report its own behavior. Memory introspection adds context about the agent’s execution environment, such as a container or microVM, to help detect external threats and agent misbehavior. This combination of OpenShell software runtimes and BlueField-4 hardware trust domains creates a defense-in-depth security model for enterprise AI deployment.
Sources
- Nvidia
- Inside NVIDIA Sentry: Secure AI Agents at Runtime | NVIDIA Technical Blog
- Cadence unveils Nvidia-backed AI models that chipmakers can run in-house
Photo: Anderseidesvik (CC BY-SA 4.0)
