Defining NVIDIA’s integrated factory automation stack
NVIDIA’s new factory automation stack is a unified industrial AI platform that combines open-source physical AI agent tools, the DSX full-stack AI factory framework, and the Factory Operations Blueprint reference design to simplify how manufacturers design, simulate, and deploy autonomous factory operations at scale. Together, these pieces aim to replace fragmented industrial systems with an integrated architecture spanning chips, software, simulation, and plant control. The strategy is clear: move from isolated pilots to production-grade, AI-managed factories that can be designed in simulation, validated before build-out, and run with continuous feedback from the shop floor. For manufacturers, this promises fewer custom integrations, faster time to deployment, and a clearer path from traditional automation to plants where physical AI agents and centralized decision engines coordinate production in real time.
DSX: A full-stack platform for AI factories
DSX sits at the core of NVIDIA factory automation as a full-stack framework for building AI factories from design to operations. It aligns compute, networking, storage, facility design, power, cooling, controls, simulation, and operations into one industrial AI platform that infrastructure builders can reuse instead of reinventing. DSX blends modular, open-source software libraries and APIs with NVIDIA accelerated computing platforms and partner technologies, backed by DSX Reference Design for validated architectures and DSX Sim for high-fidelity infrastructure simulation. Operators can model an AI factory, validate performance, and then deploy with fewer surprises. As Jensen Huang said, “With the DSX platform, you can simulate the entire factory before you spend a dollar, validate performance before a single rack is installed and operate with the kind of reliability that production AI demands.”
FOX Factory Operations Blueprint: Standardizing autonomous factories
While DSX focuses on AI factory infrastructure, the Factory Operations Blueprint—codenamed FOX—targets the messy reality inside manufacturing plants. Today, PLCs, SCADA, MES, and ERP run as separate islands, which limits plant-wide intelligence and slows down quality and root-cause analysis. FOX proposes a reference design that adds a unified decision layer above these systems, turning scattered signals into an AI-driven management view of the entire factory. It ingests data from legacy PLCs and modern IoT sensors, integrates NVIDIA Metropolis for vision-based quality inspection, and uses NVIDIA Omniverse digital twins to keep a live, simulated mirror of the shop floor. The result is an architecture where AI can reroute materials, adjust machine parameters, and coordinate workflows, pushing factories from predictive monitoring toward autonomous factory operations that adapt in real time.
Physical AI agents: Pre-built skills for robotics and vision AI
NVIDIA’s open-source physical AI agent tools supply the “hands and eyes” of autonomous factory operations. Announced at NVIDIA GTC Taipei, the collection spans Omniverse, Cosmos, Isaac, Metropolis, Alpamayo, and Jetson technologies, turning tasks like simulation, synthetic data generation, training, validation, and deployment into agent-executable workflows. With the NVIDIA Agent Toolkit, developers gain pre-built skills and structured instructions that coding agents can follow to control robots, autonomous vehicles, industrial digital twins, and vision AI systems. According to NVIDIA, AI agents can now use its libraries, models, and frameworks directly, so physical AI development can move faster and at lower cost. For manufacturers, these physical AI agents reduce the need for bespoke robotics code and make it easier to plug smart workcells into the wider industrial AI platform defined by DSX and FOX.
Lowering adoption barriers for autonomous factory operations
The combination of DSX, FOX, and physical AI agents changes the adoption curve for NVIDIA factory automation. DSX gives infrastructure teams a repeatable, simulated path to build AI factories with predictable performance and reliability. FOX offers a reference blueprint to unify PLC, SCADA, MES, ERP, and vision systems into a single AI decision layer, trimming custom integration work. Open-source physical AI tools add reusable skills for robotics, digital twins, and vision, helping teams move from prototypes to scalable deployments. Together, these elements turn industrial AI from a bespoke engineering project into a product-like stack: simulation-first, reference-designed, and agent-ready. Enterprises that once saw autonomous factory operations as risky can now start with proven architectures and tools, lowering both technical and organizational barriers to building AI-managed plants.
