NVIDIA AI Infrastructure(@NVIDIAAIInfra):How can AI factories get more from the power they have while helping keep the grid flexible? NVIDIA DSX MaxLPS optimizes power within an AI factory's existing budget. At @nscale, it boosted aggregate inference throughput by 49% without adding provisioned power. NVIDIA DSX Flex lets that same facility respond to the grid. In a simulated grid-demand event, DSX Flex and MaxLPS together cut GPU power consumption by 25% for 30 minutes. See how Nscale put dynamic power management to work at its Iceland AI factory ➡️

2026.10.23 16:56

How can AI factories get more from the power they have while helping keep the grid flexible? NVIDIA DSX MaxLPS optimizes power within an AI factory's existing budget. At @nscale, it boosted aggregate inference throughput by 49% without adding provisioned power. NVIDIA DSX Flex lets that same facility respond to the grid. In a simulated grid-demand event, DSX Flex and MaxLPS together cut GPU power consumption by 25% for 30 minutes. See how Nscale put dynamic power management to work at its Iceland AI factory ➡️