How DayOne’s Modular Platforms Support NVIDIA DSX POD

The rapid growth of artificial intelligence is reshaping not only demand for data center capacity, but how that capacity is designed, built and delivered. As AI infrastructure grows more complex, customers need power, cooling and connectivity systems that can support demanding compute environments while bringing capacity online reliably and at speed. 

Most recently, NVIDIA named DayOne as a partner supporting the new DSX POD reference design. DayOne’s modular platforms enable AI-ready data center capacity to be deployed faster and more consistently on a global scale. By standardizing designs, prefabricating key systems and moving suitable work into controlled factory environments, we reduce onsite complexity and create a more repeatable path from design to deployment.  

Modular infrastructure is already part of DayOne’s approach

At DayOne, modularization and prefabrication are part of a broader approach to making data center infrastructure more standardized, scalable and efficient to deploy. Rather than engineering every component separately for each project, DayOne has developed a common ecosystem of prefabricated products that can be configured into different platforms to meet specific customer requirements. 

A data center is more than the building that secures the compute. Around it sits an intricate physical infrastructure of power and cooling systems that keeps increasingly dense computing environments running. DayOne’s modular approach starts by turning more of this supporting infrastructure into standardized and scalable products, which can then be configured within broader platforms for different deployment needs. 

That extends across multiple layers of the data center: from power distribution to back power generation modules, from primary cooling modules to air or liquid cooling skids to the data hall, and through to IT skids that bring together racks, busways, cable management, monitoring and other elements. 

Modularization and prefabrication have been integral to our engineering and delivery approach from the outset. Beyond conventional colocation capacity, we have developed modular capabilities across multiple layers of the data center to enable more standardized design, integrated delivery and scalable deployment. 

Together, DayOne’s products become building blocks. How they are integrated at the data-center level is where DayOne’s three infrastructure platforms come in. Where flexibility matters, there is ONEShell™. Its configurable design accommodates single- and double-storey designs requirements, as well as air- and liquid-cooled solutions. 

"Where the challenge is AI at much greater scale, ONEFactory™ compresses that timeline further. Its long-span design is native to AI factory requirements, with higher space utilisation and power-to-token efficiency at its core."

Push prefabrication further and you arrive at ONEPod. With an all-prefabricated, plug-and-play approach, that takes integration away from the construction site.  

The aim of standardization is not to build the same data center everywhere. It is to create scalable products and platforms that can be configured around different infrastructure requirements.  

As more of the infrastructure becomes standardized and prefabricated, less is left to resolve at site, helping shorten the path from design to deployment. As a result, DayOne can deliver data center projects in eight to twelve months in Southeast Asia, significantly accelerating the deployment of digital infrastructure capacity for customers. That becomes increasingly relevant as AI infrastructure places greater demands on power and cooling.  

Building for the next generation of AI

At GTC Berlin, NVIDIA introduced DSX POD, a reference design for complete NVIDIA Vera Rubin AI factories built by partners inside modular containers, integrated and validated before delivery. DayOne is joining the global ecosystem to support the infrastructure requirements of NVIDIA DSX POD deployments.  

Our modular infrastructure designs have been reviewed with NVIDIA against its reference design requirements. For customers, designing known infrastructure requirements earlier in the process can help reduce unnecessary redesign and deployment complexity. 

NVIDIA DSX POD addresses factory integration of AI infrastructure through its reference design. DayOne’s modular delivery approach addresses the surrounding data center infrastructure and how it can be deployed at scale, highlighting why standardization and earlier integration matter as AI capacity requirements grow. 

As Charlie Boyle, Vice President of DGX Systems at NVIDIA, said, “DayOne’s support for this reference design, combined with its modular infrastructure and integrated delivery capabilities, can help customers bring capacity online sooner and shorten the time to first token.” 

The future of AI infrastructure will depend not only on access to compute, but on the ability to deliver the systems that support it with consistency, reliability and speed. DayOne will continue advancing its modular infrastructure approach to help customers bring increasingly complex digital infrastructure into operation.