NVIDIA Details Vera Rubin and DSX Platform Energy Efficiencies

At the AI Infra Summit on September 15, 2026, NVIDIA announced several performance results and collaborations for its Vera Rubin and DSX platforms, focusing on energy efficiency and token throughput. The event drew more than 8,000 attendees this year, up from 3,500 last year.

NVIDIA introduced the DSX MaxLPS power-optimization software. The platform can deliver up to 1.4x more tokens per megawatt. In testing, AI cloud provider Lambda used DSX MaxLPS on NVIDIA Blackwell servers to run 19 nodes within the power budget typically allocated to 16 full-power nodes. This increased cluster-wide token throughput by 24%, rising from about 4 million to 5 million tokens per second, while improving performance per watt by 23%. For next-generation Vera Rubin NVL72 AI factories, DSX MaxLPS can enable up to 40% more GPU capacity within the same megawatt budget.

The Vera Rubin NVL72 platform also showed gains under other benchmarks. It delivers up to 30x higher throughput per megawatt than the NVIDIA GB300 NVL72 on the DeepSeek V4 Pro model, according to the SemiAnalysis AgentX dashboard. The results also showed up to 45x lower cost per million tokens. When combined with Groq 3 LPX, the system delivers up to 35X higher token throughput per megawatt than the GB200 NVL72 for 2-trillion-plus-parameter models at long context. On a 100K-context Qwen 3.8 27B workload, Groq 3 LPX reached 2,529 output tokens per second per user.

Startups also reported figures for the NVIDIA Vera CPU. Perplexity measured 1.9x faster sandbox starts for its SPACE platform. DeepInfra recorded 2.2x faster orchestration step latency in its benchmarks.