Nvidia is introducing a modular infrastructure design intended to help businesses bring artificial intelligence systems online faster, shortening the time between receiving computing equipment and generating revenue from AI workloads.
Announced at the company’s GTC event in Berlin on Wednesday, the Nvidia DSX POD is a standardized reference design that packages essential AI data center capabilities into a unit roughly the size of a conventional shipping container. The approach aims to address one of the industry’s major challenges: the lengthy process of building and commissioning facilities capable of supporting large-scale AI operations.
A Faster Route to AI Infrastructure
Developing a data center can require months or even years of construction and preparation. During that period, operators continue to incur expenses without generating revenue from the AI processing capacity the facility is designed to provide.
Nvidia’s DSX POD is designed to reduce that waiting period through a modular, factory-built solution. Rather than requiring customers to complete a traditional facility before deploying their computing systems, the units can be delivered ready for connection to existing power, cooling and networking infrastructure.
According to Nvidia, DSX POD configurations powered by its Vera Rubin platform could reduce deployment timelines from approximately six months to just 14 days.
CEO Jensen Huang described the potential impact in an interview with Deirdre Bosa, explaining that cutting the time required to put AI factories into operation from months to weeks could fundamentally change how quickly customers expand their computing capacity.
Designed to Complement Traditional Data Centers
The company does not intend for DSX POD to replace conventional data center construction. Instead, the solution is designed to provide an earlier starting point for AI operations while permanent facilities are still being completed.
By enabling customers to deploy computing resources sooner, Nvidia hopes to help businesses begin running AI applications and generating income before their larger infrastructure projects are fully operational.
Huang said the approach brings the precision and speed of factory manufacturing to AI infrastructure, allowing customers to expand capacity and begin revenue-producing operations more quickly.
Water Efficiency Through Cooling Innovation
Alongside faster deployment, Nvidia highlighted potential water savings from the DSX POD’s cooling architecture. The design supports liquid cooling while incorporating dry coolers to reduce water consumption.
Nvidia estimates that this configuration could save as much as 54 million liters of water per day compared with a standard data center. The projected savings could make the system attractive to operators seeking to manage the environmental and resource demands associated with expanding AI computing infrastructure.
Faster Deployment Could Accelerate Chip Demand
The benefits of reducing deployment delays extend beyond Nvidia’s customers. Shorter installation timelines could help ease one of the constraints slowing the broader expansion of AI infrastructure, potentially enabling Nvidia to deliver and sell its processors to customers more quickly.
However, computing hardware is only one part of the challenge. Securing sufficient electricity remains another major obstacle for data center developers, and building new power generation facilities generally takes considerably longer than constructing data centers themselves.
Even with a faster deployment model, access to reliable electricity will remain essential to bringing additional AI computing capacity online.
Industry Partners Target a 2027 Rollout
Nvidia is collaborating with a range of technology and infrastructure companies to prepare the DSX POD for commercial deployment. Its partners include Anduril, Armada, Bull, Crusoe, Dell Technologies and Foxconn, among others.
The companies are expected to begin delivering the systems in early 2027. If the rollout achieves Nvidia’s projected installation timelines, the modular design could offer businesses a faster way to expand AI operations while longer-term data center projects continue to progress.
