TECH NEWS | NVIDIA targets more AI output from fixed power budgets
NVIDIA DSX platform combines compute, networking, power, cooling and operations to optimize AI factories around a fixed power budget.
NVIDIA is expanding its AI data center platform to improve computing efficiency, manage electricity demand and increase token throughput without requiring proportional increases in power capacity.
The company said its NVIDIA DSX platform combines compute, networking, power, cooling and operations to optimize AI factories around a fixed power budget. The approach includes technologies for dynamic power allocation, workload management, simulation and reference designs.
The announcement comes as power availability becomes a growing constraint for large-scale AI deployments. At the AI Infra Summit, NVIDIA Vice President of Hyperscale and High-Performance Computing Ian Buck highlighted AI factory efficiency as a key infrastructure concern.
One early validation came from GPU cloud provider Lambda, which tested NVIDIA DSX MaxLPS on a five-rack, 19-node cluster using NVIDIA HGX B200 GPU servers.
According to results released at the summit, Lambda ran 19 nodes within the same power budget used by 16 nodes operating at full power. The configuration increased cluster-wide token throughput by 24%, from about 4 million tokens per second to 5 million, while improving performance per watt by 23%.
DSX MaxLPS monitors GPU- and rack-level power consumption and reallocates available power based on workload requirements. NVIDIA said its projections indicate the technology could enable up to 40% more GPU capacity for next-generation Vera Rubin NVL72 AI factories within the same megawatt power budget in suitable deployment environments.
NVIDIA is also developing DSX Flex for grid participation. The technology is designed to respond to utility signals, including load-shedding, demand-response and pricing events, by adjusting flexible AI workloads while keeping higher-priority jobs running.
An earlier demonstration involved NVIDIA and Emerald AI at an AI factory participating in Silicon Valley Power’s Flexible Load Interconnect Program. Emerald AI’s Conductor platform automatically reduced power consumption from 4 megawatts to 3 megawatts after receiving a utility signal while maintaining higher-priority workloads.
NVIDIA said the facility has responded to more than 200 demand signals.
The first dedicated DSX Flex commercial deployment is planned for a 96-megawatt Vera Rubin AI factory at NVIDIA’s AI Factory Research Center in Manassas, Virginia.
The DSX platform also includes DSX OS, an open-source modular software layer for AI factory lifecycle management; DSX Sim for modeling and validating factory designs; and DSX Reference Designs covering compute, networking, storage and facilities.
NVIDIA is incorporating an 800-volt direct-current power architecture into its reference designs to support denser accelerated computing systems and reduce conversion complexity.
The company said the broader DSX approach is intended to optimize the entire AI factory rather than individual components, including power delivery, cooling, networking and workload operations.
NVIDIA founder and CEO Jensen Huang has said a 1-gigawatt AI factory cannot simply become a 2-gigawatt facility, making efficiency and productivity within existing power constraints increasingly important as AI computing demand grows.
Full disclosure: All news articles published on the TechSabado website are written by human journalists, unless otherwise specified. Final text editing is also performed by human editors, with artificial intelligence (AI) used only to assist with additional grammar and style guide corrections..
————————————————————————-
TEN YEARS OF TECHSABADO!
PLEASE LIKE our FACEBOOK PAGE and SUBSCRIBE to OUR YOUTUBE & SPOTIFY channels.
WATCH TECHSABADO ON OUR YOUTUBE CHANNEL:
WE ARE ALSO ON SPOTIFY
WATCH OUR OTHER YOUTUBE CHANNELS:
PLEASE LIKE our FACEBOOK PAGE and SUBSCRIBE to OUR YOUTUBE CHANNEL.
WE ARE ALSO ON SPOTIFY
WATCH OUR OTHER YOUTUBE CHANNELS:
PLEASE LIKE our FACEBOOK PAGE and SUBSCRIBE to OUR YOUTUBE CHANNEL.
