← Back

Wire · technology

NVIDIA: Vera Rubin NVL72 Delivers Up To 30x More Agentic AI Throughput Per Megawatt ...

Published

24 August 2026

Topic

technology

◆ Sectors

AI InfrastructureSemiconductors

◆ Geography

United States

◆ Source

Read at pulse2.com →

◆ Verified

Fusion42 · 24 August 2026 · Fusion42 review

NVIDIA's Vera Rubin NVL72 platform delivers up to 30 times more agentic AI inference throughput per megawatt and reduces token costs by up to 35 times compared to previous models, addressing power constraints in AI data centers. The platform incorporates innovations in GPUs, networking, memory, and power management to optimise AI agent workloads with extended context and multi-step reasoning.

This Wire brief sits within Fusion42's coverage of AI Infrastructure and Semiconductors.

◆ ◆ The Wire takeaway

NVIDIA has shifted AI infrastructure focus to throughput per megawatt and token cost, making it viable to run more agentic AI workloads cost-effectively. You face a new scale test where hardware efficiency can unlock market access or crush margins if lagging.

◆ Coverage

1 source · 24 Aug 2026

◆ Related on Wire

◆ Topics

AI InfrastructureSemiconductorsnvidiaagentic-aigputhroughputenergy-efficiencytoken-cost