NVIDIA Launches DSX Ready Program for AI Factory Power and Cooling

NVIDIA has announced the launch of DSX Ready, a new qualification program designed to certify power and cooling components for AI data centers. The initiative targets two specific categories at its debut: battery energy storage systems and cooling distribution units. By establishing clear technical benchmarks, the company aims to help infrastructure builders select products that align with their overall architectural requirements.

The primary goal is to reduce the uncertainty and engineering risks associated with integrating third-party hardware into large-scale AI facilities. According to the NVIDIA Blog, this program connects a unified reference design for AI factories with specific partner offerings. This approach allows builders to evaluate products with greater confidence, ensuring that power and cooling systems can effectively support the computing load without introducing new bottlenecks.

Initial Partners and Testing Scope

The program begins with a select group of qualified suppliers. For battery energy storage, the initial partners include Hitachi Energy, LG Energy Solution, and Tesla. In the cooling sector, qualified solutions come from LG Electronics, LiquidStack, and Vertiv. These companies have completed specific qualification tests and submitted data for review, meeting the functional requirements defined by NVIDIA for their respective categories.

The qualification process involves rigorous testing to ensure that these components meet the necessary performance and safety standards. For cooling units, the process utilizes a self-qualification suite to verify functionality. For battery systems, partners must adhere to a defined qualification boundary. While this certification validates the product's technical fit within the reference design, it does not replace the need for site-level engineering or guarantee stability in every possible real-world configuration.

Reducing Integration Risks for Builders

Building AI infrastructure is increasingly constrained by power availability, water usage, and grid capacity. Optimizing one part of the system can inadvertently shift performance bottlenecks to another. By providing a standardized qualification mark, DSX Ready helps builders filter out incompatible options early in the design phase. This allows engineering teams to focus on how qualified offerings fit their specific site configurations and operating needs, rather than verifying basic compatibility from scratch.

The program is part of a broader strategy to treat the AI factory as a single, unified system rather than a collection of isolated parts. NVIDIA emphasizes that the value of a reference design increases when it is directly linked to specific, vetted products. This structured approach aims to streamline the path from initial design to deployment, ensuring that the final infrastructure can deliver useful AI output within existing physical and energy constraints.

Future Expansion and Trade-offs

While the launch focuses on power and cooling, NVIDIA indicates that additional categories will be rolled out over time. These future additions may cover other aspects of infrastructure and software, expanding the scope of the qualification program. For now, the trade-off for builders is that they must adhere to NVIDIA's specific reference design requirements to access this streamlined selection process.

Partners seeking to join the program must demonstrate that their specific offerings meet the category-specific requirements. This creates a defined path for suppliers to prove their technology's compatibility. For builders, the benefit is a clearer way to discover and evaluate offerings that are already aligned with the core AI factory platform, potentially accelerating project timelines and reducing the complexity of system integration.