{"version":1,"type":"story","url":"https://digestai.news/story/vera-rubin-nvl72-delivers-67-more-tokens-per-dollar-in-agentic-inference","json":"https://digestai.news/story/vera-rubin-nvl72-delivers-67-more-tokens-per-dollar-in-agentic-inference.json","markdown":"https://digestai.news/story/vera-rubin-nvl72-delivers-67-more-tokens-per-dollar-in-agentic-inference.md","slug":"vera-rubin-nvl72-delivers-67-more-tokens-per-dollar-in-agentic-inference","headline":"NVIDIA Vera Rubin NVL72 claims 30x throughput gain for agentic AI workloads","summary":"NVIDIA announced significant efficiency advancements for its Vera Rubin NVL72 platform at the AI Infra Summit, emphasizing a shift from peak performance metrics to validated tokens per megawatt. The company highlighted that the new system delivers up to 30x higher throughput per megawatt compared to the previous GB300 NVL72 when running agentic coding sessions on the DeepSeek V4 Pro model. This improvement is driven by extreme codesign involving sixth-generation NVLink interconnects and NVFP4 precision on fifth-generation Tensor Cores.\n\nKey partners demonstrated tangible results using NVIDIA’s DSX MaxLPS software, which optimizes power allocation across mixed workloads. Lambda reported a 23% improvement in performance per watt, achieving 5 million tokens per second on 19 nodes within the power budget of 16. Additionally, Emerald AI and Silicon Valley Power showcased a flexible-load program that allows AI factories to dynamically adjust energy consumption based on real-time grid signals, protecting priority workloads while supporting grid stability.\n\nThe event also featured benchmarks for the NVIDIA Vera CPU, with startups like Perplexity and ClickHouse reporting substantial latency and throughput gains for data-intensive and agentic tasks. NVIDIA positioned these updates as essential for scaling agentic AI, where long context windows and complex tool calls require infrastructure that maximizes output within strict power constraints.","keyPoints":["Vera Rubin NVL72 achieves 30x higher throughput per megawatt than GB300 NVL72 on agentic workloads.","Lambda improved performance per watt by 23% using NVIDIA DSX MaxLPS power optimization software.","Emerald AI demonstrated grid-responsive power management, allowing AI factories to throttle load based on demand."],"whyItMatters":"As agentic AI drives massive token consumption, energy efficiency becomes the primary economic constraint. NVIDIA's focus on tokens per megawatt directly impacts the cost of inference and the scalability of AI factories, influencing how operators manage power budgets and grid integration.","category":{"slug":"hardware","name":"Hardware & Compute","url":"https://digestai.news/category/hardware"},"entities":{"companies":["NVIDIA","Lambda","Emerald AI","Silicon Valley Power","Amazon","Perplexity"],"models":["DeepSeek V4 Pro","Qwen 3.8 27B"],"people":["Ian Buck"]},"firstPublishedAt":"2026-09-14T22:08:56Z","updatedAt":"2026-09-16T15:24:02Z","sourceCount":6,"hasPrimarySource":true,"sources":[{"outlet":"NVIDIA Blog","title":"AI Infra Summit: NVIDIA Vera Rubin and DSX Platform Advancements Showcase Energy Efficiencies of Optimizing Tokens Per Watt for AI Factories","url":"https://blogs.nvidia.com/blog/ai-infra-summit-vera-rubin-dsx-energy-efficiencies-tokens-per-watt-ai-factories","publishedAt":"2026-09-15T16:55:40Z","type":"primary","primary":true,"lead":true},{"outlet":"Unite.AI","title":"NVIDIA Vera Rubin NVL72 Posts First MLPerf Inference Preview Results","url":"https://unite.ai/nvidia-vera-rubin-nvl72-posts-first-mlperf-inference-preview-results","publishedAt":"2026-09-16T15:24:02Z","type":"press","primary":false,"lead":false},{"outlet":"NVIDIA Blog","title":"NVIDIA Vera Rubin NVL72 Delivers Leading Performance in MLPerf Inference v6.1 Debut","url":"https://blogs.nvidia.com/blog/vera-rubin-nvl72-mlperf-inference","publishedAt":"2026-09-16T15:00:48Z","type":"primary","primary":true,"lead":false},{"outlet":"Unite.AI","title":"CoreWeave Connects Hundreds of Rubin GPUs in One Multi-Rack Cluster","url":"https://unite.ai/coreweave-connects-hundreds-of-rubin-gpus-in-one-multi-rack-cluster","publishedAt":"2026-09-16T12:20:01Z","type":"press","primary":false,"lead":false},{"outlet":"finance.yahoo.com","title":"CoreWeave Brings Multi-Rack Nvidia Vera Rubin NVL72 Cluster Online — To Give Agentic AI Customers Greater Scale, Faster Iteration","url":"https://finance.yahoo.com/technology/ai/articles/coreweave-brings-multi-rack-nvidia-131852656.html","publishedAt":"2026-09-16T06:18:00Z","type":"press","primary":false,"lead":false},{"outlet":"SemiAnalysis","title":"Vera Rubin NVL72 Agentic Inference: 67x better Performance per Dollar","url":"https://newsletter.semianalysis.com/p/vera-rubin-nvl72-agentic-inference","publishedAt":"2026-09-14T22:08:56Z","type":"newsletter","primary":false,"lead":false}],"sourceNotes":null,"discussions":[],"thread":{"title":"The Race for Cheaper AI Inference Hardware","url":"https://digestai.news/thread/3-8b-llm-trained-to-0-384-core-score-for-under-1-000-using-consumer-gpus","storyCount":3},"cite":{"text":"Digest AI, \"NVIDIA Vera Rubin NVL72 claims 30x throughput gain for agentic AI workloads\", 14 September 2026, https://digestai.news/story/vera-rubin-nvl72-delivers-67-more-tokens-per-dollar-in-agentic-inference","publisher":"Digest AI","title":"NVIDIA Vera Rubin NVL72 claims 30x throughput gain for agentic AI workloads","datePublished":"2026-09-14T22:08:56Z","url":"https://digestai.news/story/vera-rubin-nvl72-delivers-67-more-tokens-per-dollar-in-agentic-inference"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}