AI Factory Infrastructure
NVIDIA DSX POD
The fastest path to productive intelligence at scale.
Overview
Accelerate the Path From Site Readiness to Capacity
Large-scale AI factories can be impacted by integration complexity and the onsite work required to install, cable, connect, test, and bring capacity online. The NVIDIA DSX™ POD reference design, built for the Vera Rubin platform, offers a repeatable, validated architecture that lets partners preintegrate compute and supporting infrastructure in the factory, creating a faster and more predictable path to productive intelligence.
Benefits
Build and Scale With Greater Confidence
The NVIDIA reference design for factory-integrated infrastructure containers helps customers reduce onsite complexity, improve deployment consistency, and scale through experienced delivery partners.
Faster Time to First Token
Reduce construction and installation time by moving cabling, wiring, piping, and functional validation to the factory, reducing the work required at the deployment site.
Deploy and Scale With Modular Containers
Use a common, validated design foundation to make infrastructure deployments more consistent and easier to replicate across sites, simplifying capacity planning.
Maximize Productive Use of Power
Use NVIDIA DSX MaxLPS for policy-driven power control, dynamic workload orchestration, and high-fidelity telemetry to maximize throughput within defined power limits.
Deliver Through Experienced Partners
Work with NVIDIA’s robust ecosystem of partners, who engineer, manufacture, integrate, customize, service, support, and deliver customer solutions.
Partners
Explore the NVIDIA DSX POD Ecosystem
Leading ODM partners and container solution providers leverage the DSX POD reference design to deliver standardized containers with factory-integrated infrastructure.
-
ODMs
-
Container Solution Providers
Learn What’s New From GTC Berlin
The NVIDIA DSX POD reference design is among the AI infrastructure announcements from GTC Berlin. See how it fits into the broader vision for the next generation of AI factories.
Next Steps
Ready to Plan a Modular AI Factory?
Work with NVIDIA and delivery partners to evaluate a repeatable approach for turning prepared site capacity into productive AI infrastructure.
Learn More About NVIDIA DSX POD
Contact your NVIDIA representative to discuss your site requirements and learn how the DSX POD reference design and delivery partners can support your AI factory deployment.
FAQ
It’s a partner-delivered, factory-integrated modular data center design built on a validated NVIDIA cluster-system foundation. It’s intended to help large-scale AI-factory operators reduce onsite integration complexity and accelerate the path to productive AI capacity.
The initial focus is large-scale, greenfield, liquid-cooled AI infrastructure, including NVIDIA Cloud Partners (NCPs) and other organizations building capacity at scale. Conventional and other modular construction approaches remain relevant for different deployment requirements.
No. It’s designed to move more installation, cabling, wiring, piping, and functional validation upstream. Customers still need a prepared site and connections to site power, facility water, cooling, and network infrastructure.
NVIDIA develops, integrates, tests, and validates the cluster-system design foundation. NVIDIA ecosystem partners engineer, manufacture, integrate, customize, service, support, and deliver customer solutions.