Organized by Jin10 Data
NVIDIA (NVDA.O) unveiled the NVIDIA DSX platform at the NVIDIA GTC Taipei conference in Taipei, further expanding its footprint into AI factory infrastructure.
Unlike in the past, when sales of GPUs were the core focus, DSX aims to provide enterprises with a complete AI factory solution encompassing design, simulation, deployment, and operational management.
As AI models continue to grow in scale, the challenges facing data centers extend beyond chip performance to include power supply, cooling capacity, resource allocation, and overall operational efficiency. NVIDIA believes that the key competitive metric for the future AI industry will gradually shift from individual chip performance to overall infrastructure efficiency—how to generate more computing power and intelligent services within limited power, space, and resources.
To this end, the DSX platform integrates NVIDIA’s chips, systems, software, reference architectures, and partner technologies to cover the entire lifecycle of AI factory construction and operation. By unifying computing, software, and infrastructure stacks, the platform helps customers accelerate deployment, enhance reliability and operational efficiency, and reduce the cost of generating tokens during AI inference.
Jensen Huang said:
We’re not just delivering chips—we’re providing every infrastructure builder with a complete methodology to build an AI factory. With the DSX platform, you can simulate the entire factory at no cost, validate performance before installing your first rack, and operate with the reliability required for production-grade AI.
The software system released this time primarily includes DSX MaxLPS and DSX OS.
Among them, DSX MaxLPS leverages 45°C liquid cooling and rack-level power optimization technologies to increase token output per megawatt of electricity. NVIDIA states that this technology enables up to 40% additional GPU deployment with minimal impact on performance, further reducing computing costs within a fixed power budget.
DSX OS is an open-source software platform designed for AI factory operations, supporting lifecycle management, intelligent scheduling, health automation, multi-tenant operations, and platform services. NVIDIA will also open-source modular software libraries, APIs, reference designs, and an accelerated computing platform to build a unified software architecture.
In addition to its core software, DSX integrates several existing capabilities. The DSX Reference Design provides a reference architecture covering computing, networking, storage, power, and cooling systems; DSX Sim enables digital twin simulation and optimization throughout the entire process from planning to operations; DSX Flex dynamically adjusts workloads based on grid load and electricity pricing changes; and DSX Exchange enables data coordination among computing, networking, energy, and cooling systems.
In terms of commercial deployment, cloud service providers such as CoreWeave, Crusoe, IREN, and Lambda have implemented DSX core components to enhance GPU utilization and accelerate the time-to-market for AI cloud services.
The hardware ecosystem is also expanding in tandem. Companies such as Dell Technologies (DELL.N), Hewlett Packard Enterprise (HPE.N), Lenovo (0992.HK), Supermicro (SMCI.O), ASUS, Foxconn, Gigabyte, Pegatron, and Quanta Cloud Technology are developing NVIDIA DSX-ready systems to help customers build full-stack AI factories.
Meanwhile, DSX Flex has launched commercial pilot projects with Emerald AI and Silicon Valley Power to validate the AI factory’s ability to dynamically adjust power consumption based on grid demand.
From a strategic perspective, the DSX marks NVIDIA’s continued transition from an AI chip supplier to an AI infrastructure platform provider. By integrating chips, software, data center architecture, operations management, and energy scheduling into a unified system, NVIDIA aims to establish industry standards covering the entire lifecycle of AI factories and further solidify its leading position in the global AI infrastructure market.