NVIDIA DSX AI Factory Megawatt & Throughput Optimizer

Power & Grid Controls

Megawatt Capacity Budget 50.0 MW

Factory scale up to 100MW campus footprint

Grid Profile Condition Fluctuating

Simulates renewables variability & utility curtailment

NVIDIA DSX Platform Mode Adaptive

Real-time dynamic clocking, power capping & phase load steering

Liquid Cooling Thermal Efficiency 94.5%

Direct-to-chip CDU liquid loops impact parasitic PUE

Target Workload Architecture Workload communication & memory bandwidth profiles

Export generated successfully.

AI Output Throughput4,285,000
tokens/sec across active clusters

Energy Efficiency (PUE)

1.14

Power Usage Effectiveness

Grid Adaptation Score

96.4%

Frequency & curtailment resilience

Megawatt Utilized

48.2 MW

of 50.0 MW budget allocation

Operational Telemetry: Power Allocation vs AI Token Throughput 10-Hour Adaptive Profile

Subsystem Power & Performance Distribution DSX Platform Telemetry Bus v4.2

Subsystem Domain	Allocated MW	% of Total	Efficiency Ratio	Operational State

Source Grounding: Based on official announcement by @nvidia at #AIInfraSummit: "Every megawatt matters. The NVIDIA DSX AI Factory Platform helps AI factories maximize AI output, improve energy efficiency, and intelligently adapt to changing grid conditions."