---
format: "aidr-story-markdown/v1"
id: "c1ce0af332a126bafe8677badb8a28b44614dc07a7d4e35347c99d4885401177"
canonical_url: "https://aidr.today/c1ce0af3?lang=en"
title: "Nvidia Vera Rubin NVL72 Cuts GPU Needs for Large AI Model Training by 75%"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-10-26T16:22:44.000Z"
category: "Chips"
topics: ["nvidia","llm"]
source_urls: ["https://marketbrief.now/ai/nvidia-vera-rubin-nvl72-cuts-gpu-needs-for-large-ai-model-training-by-75-9891d6da","https://huggingnews.com/ai/nvidia-vera-rubin-nvl72-cuts-gpu-needs-for-large-ai-model-training-by-75-9891d6da"]
summary: "Nvidia's Vera Rubin NVL72 can train a 10T-parameter AI model on 100T tokens in one month using about 16,000 GPUs, compared to 64,000 GB200 GPUs required under the current Grace Blackwell architecture."
---

# Nvidia Vera Rubin NVL72 Cuts GPU Needs for Large AI Model Training by 75%

> [Open the canonical story](<https://aidr.today/c1ce0af3?lang=en>)

**Published:** 2026-10-26T16:22:44.000Z
**Category:** Chips
**Topics:** nvidia, llm

## Summary

Nvidia's Vera Rubin NVL72 can train a 10T\-parameter AI model on 100T tokens in one month using about 16,000 GPUs, compared to 64,000 GB200 GPUs required under the current Grace Blackwell architecture\.

## Sources

- [Story source](<https://marketbrief.now/ai/nvidia-vera-rubin-nvl72-cuts-gpu-needs-for-large-ai-model-training-by-75-9891d6da>)
- [Story source](<https://huggingnews.com/ai/nvidia-vera-rubin-nvl72-cuts-gpu-needs-for-large-ai-model-training-by-75-9891d6da>)