NVIDIA DSX Platform Helps AI Factories Overcome Power Constraints

Power availability has become the primary bottleneck for large‑scale AI deployments, with hundreds of gigawatts stuck in interconnection queues and billions of dollars of equipment on backorder.

NVIDIA DSX addresses this by integrating three layers: DSX Sim for joint grid‑facility‑rack simulation using SimReady assets, DSX Flex to turn AI factories into flexible grid assets that respond to real‑time supply conditions, and DSX MaxLPS/Exchange to extract more compute per watt and share telemetry across power, cooling, and networking domains.

Key ecosystem partners are aligning their solutions with DSX. GE Vernova contributes grid‑to‑rack power expertise, Hitachi Energy qualifies its battery storage as DSX Ready, Schneider Electric with AVEVA and ETAP delivers validated Vera Rubin reference architectures and an Omniverse‑based digital twin, Siemens Energy develops a grid‑aware reference design for generation through distribution, and Phaidra embeds DSX Flex, Max‑Q, and Exchange into self‑learning agents for autonomous cooling and power optimization.

Why it matters for GPU and AI infrastructure

By co‑designing energy supply and compute demand, DSX reduces stranded capacity, shortens time‑to‑token, and improves sustainability — critical for operators scaling GPU clusters in power‑constrained regions.