NVIDIA Launches DSX Platform to Standardize AI Factory Infrastructure and Grid Interaction

At the GTC Taipei event on May 31, 2026, NVIDIA debuted the DSX platform, an integrated architecture intended to assist developers in building and operating AI factories. The platform aligns various layers of the technology stack—including silicon, software, and facility infrastructure—to minimize the cost per token and improve the reliability of large-scale AI production.

A primary feature of the platform is DSX MaxLPS, a software suite designed to maximize intelligence output within fixed power limits. By utilizing 45-degree-Celsius liquid cooling and specialized in-rack optimizations, the technology allows operators to increase GPU density by up to 40% while maintaining peak energy efficiency. This is paired with DSX OS, an open-source modular operating system that handles lifecycle management, multi-tenant operations, and automated health monitoring.

To manage the energy demands of these facilities, NVIDIA introduced DSX Flex, which facilitates communication between AI factories and the electrical grid. This tool enables dynamic workload adjustments in response to utility signals such as demand response, pricing changes, and load shedding. NVIDIA is currently testing these capabilities through a multi-megawatt pilot program in collaboration with Silicon Valley Power and Emerald AI.

The platform also includes DSX Sim, a high-fidelity simulation tool that allows builders to model and validate infrastructure designs before physical installation. This digital twin approach is supported by partners like Quanta Cloud Technology and Pegatron, who are working with Dassault Systèmes to automate the design process from individual racks to entire facilities.

A wide range of industry partners are adopting the DSX framework. Major system manufacturers, including Dell Technologies, HPE, Lenovo, and Supermicro, are developing DSX-ready hardware. Simultaneously, cloud service providers such as CoreWeave (NASDAQ: CRWV), IREN, and Lambda are implementing DSX software components to accelerate the deployment of AI cloud capacity and improve hardware utilization rates.