Education & Research Software & Algorithm Provider Cloud & Data
NVIDIA Vera Rubin NVL72 Achieves 30x Efficiency Improvement for AI Workloads
Original from NvidiaNews: Up to 30x More Work Per Watt: NVIDIA Vera Rubin NVL72 Sets a New Efficiency Standard for AI Agents

NVIDIA Vera Rubin NVL72 Achieves 30x Efficiency Improvement for AI Workloads

NVIDIA's Vera Rubin NVL72 systems have demonstrated a remarkable performance increase, achieving up to 30 times higher throughput per megawatt compared to the NVIDIA GB300 NVL72 for agentic AI workloads. This significant enhancement is crucial as agentic AI applications, which require extensive token usage for tasks like investment research, continue to proliferate across various industries.

The efficiency gains from the Vera Rubin NVL72 are particularly important for power-constrained AI factories, allowing them to execute 30 times more agentic work within the same energy footprint. This leap in performance is attributed to NVIDIA's innovative GPU architecture and continuous software optimizations, which enhance the overall efficiency of both Vera Rubin NVL72 and GB300 NVL72 systems.

As agentic AI workloads evolve, the demand for efficient infrastructure will only increase. Future performance measurements will need to adapt to capture the complexities of agent workflows, which can involve hundreds of thousands of tokens. No further timeline was disclosed at the time of publication.

Editor's Note

The advancements in NVIDIA's Vera Rubin NVL72 highlight the growing need for efficient AI infrastructure as agentic workloads become more prevalent. This shift emphasizes the importance of optimizing energy consumption while maximizing performance, which is critical for enterprises looking to implement AI solutions at scale. The competitive landscape will likely see increased focus on efficiency metrics as companies strive to meet the demands of complex AI tasks.

RobotToday Initiative

Robotics needs a service framework.

RSF defines a common language for robot service capability, lifecycle operations, certification pathways, and service-provider networks.

Share

Related Articles/News

FCC Approves Ascento Guard Dori USA v1 UAS with Conditional Exemption

FCC Approves Ascento Guard Dori USA v1 UAS with Conditional Exemption

The Federal Communications Commission (FCC) has granted Conditional Approval for Ascento Inc.'s Ascento Guard Dori USA v1 Uncrewed Aircraft System, exempting it from the Covered List. This decision follows a similar exemption for Wi-Fi routers from WNC Corporation and reflects the FCC's ongoing updates to its regulations regarding UAS. This approval is si...

defense News US Government
NVIDIA Groq 3 LPX Enters Full Production, Enhancing Agentic AI Inference Speed

NVIDIA Groq 3 LPX Enters Full Production, Enhancing Agentic AI Inference Speed

NVIDIA has announced that the Groq 3 LPX, an interactive AI inference accelerator, is now in full production. This product, part of the NVIDIA Vera Rubin platform, significantly enhances AI inference by enabling rapid token generation, which is essential for agentic systems to perform complex tasks in real time. The importance of Groq 3 LPX lies in its ab...

SpaceXAI Implements NVIDIA Vera CPUs for Enhanced Agentic AI Applications at Scale

SpaceXAI Implements NVIDIA Vera CPUs for Enhanced Agentic AI Applications at Scale

NVIDIA has announced that SpaceXAI will utilize NVIDIA Vera CPUs to enhance its next generation of agentic AI applications. This deployment marks a significant step as it introduces the first CPU specifically designed for AI agents, which will help in orchestrating tools, executing code, processing data, and running simulations effectively. The integratio...

NVIDIA Launches Groq 3 LPX in Full Production, Enhancing Vera Rubin Inference for Agentic Systems

NVIDIA Launches Groq 3 LPX in Full Production, Enhancing Vera Rubin Inference for Agentic Systems

NVIDIA has announced that the Groq 3 LPX is now in full production, enhancing the Vera Rubin NVL72 platform with fast token generation capabilities. This advancement allows for 3,400 output tokens per second in complex agentic systems, significantly outperforming competitors by four times. The introduction of Groq 3 LPX is crucial as AI transitions from t...

Nvidia Plans Over 15% Price Increase for AI Servers Impacting Major Tech Firms

Nvidia Plans Over 15% Price Increase for AI Servers Impacting Major Tech Firms

Nvidia has announced a price increase of more than 15% for its AI servers, prompting warnings to Microsoft, Google, and Oracle from contract manufacturers regarding the higher costs. This decision reflects Nvidia's response to rising production expenses and market demand for AI technologies. The price hike is significant as it may affect the operational b...

Artificial Intelligence News AI infrastructure
Nvidia Makes Strategic Investment in Cloverleaf, a US Data Center Developer

Nvidia Makes Strategic Investment in Cloverleaf, a US Data Center Developer

Nvidia has announced a strategic investment in Cloverleaf, a data center developer established in 2024. Cloverleaf has successfully executed gigawatt-scale projects throughout North America, showcasing its capability in the energy sector. This investment is significant as it highlights Nvidia's commitment to expanding its infrastructure capabilities in th...

Artificial Intelligence Investments News
HoverAir's Versa Drone Faces US Shipping Ban Just Days After Launch

HoverAir's Versa Drone Faces US Shipping Ban Just Days After Launch

HoverAir's Versa, a modular drone that transforms from a camera with snap-on wings, has halted US orders shortly after its Indiegogo launch. The company announced that it can only ship the camera component due to unresolved logistics with the Flight Kit, which may indicate a federal ban. The FCC's database no longer lists the HoverAir Versa, suggesting it...

Drones News Tech
BrainChip Introduces Symphony Community Akida Bundle for IBM Workload Management

BrainChip Introduces Symphony Community Akida Bundle for IBM Workload Management

BrainChip has launched the Symphony Community Akida Bundle, an open-source software package that integrates its Akida neuromorphic processors with IBM Spectrum Symphony Community Edition. This bundle allows developers to efficiently manage workloads by routing tasks to the most suitable processors, including Akida, which is designed for lightweight, event...

Artificial Intelligence Computing Design
Waymo Unveils Nvidia-Powered Computing System for Enhanced Robotaxi Performance

Waymo Unveils Nvidia-Powered Computing System for Enhanced Robotaxi Performance

Waymo has disclosed new technical specifications of its onboard computing system that powers its autonomous driving technology. This system features a custom-built 5 nm ASIC that delivers over 1,000 TOPS of machine learning performance, designed to operate under challenging conditions encountered by its autonomous vehicles. The significance of this develo...

Autonomous Vehicles Computing Features
Dai Weijin Discusses Collaborative Efficiency as Key to Robot Chip Development

Dai Weijin Discusses Collaborative Efficiency as Key to Robot Chip Development

Dai Weijin, Chief Strategy Officer of Chipone Technology, highlighted the critical issue of collaborative efficiency in robot chip development. While the industry focuses on TOPS performance metrics, he emphasized that the real challenge lies in ensuring system efficiency aligns with computational power. He pointed out that while automotive chips may have...

Robot Chips Collaborative Efficiency Layered Computing
ROBOTTODAY Weekly August 10 – 14, 2026

ROBOTTODAY Weekly August 10 – 14, 2026

Your weekly robotics briefing: Unitree's IPO is swamped by retail demand, the US puts 100% tariffs on imported drones, LG and Nvidia commit to a 2027 humanoid, and Uber and Pony.ai bring 2,000 robotaxis to Europe.

RobotToday Weekly Market and Business News
Chinese LLMs Dominate Global Token Usage for Fifteen Consecutive Weeks with DeepSeek-V4-Flash Leading

Chinese LLMs Dominate Global Token Usage for Fifteen Consecutive Weeks with DeepSeek-V4-Flash Leading

Chinese large language models (LLMs) have achieved a significant milestone by surpassing 34.25 trillion weekly tokens for the first time, according to OpenRouter data. This marks the fifteenth consecutive week that Chinese LLMs have led global token usage, with the top four positions occupied by Chinese models. The rise of DeepSeek-V4-Flash is particularl...

Related Suppliers

NVIDIA Robotics

NVIDIA robotics division delivering Isaac simulation and training tools, Jetson edge compute modules, and Omniverse digital twin frameworks for physical AI.

Core Component Supplier Software & Algorithm Provider

Deci AI (Acquired by NVIDIA)

Israeli deep learning optimization startup acquired by NVIDIA in May 2024 for ~$300M; known for AutoNAC neural architecture search technology.

Nscale

London-based Nscale runs a full-stack AI/GPU cloud platform and on Aug 1, 2026 agreed to acquire Anyscale for $1.65 billion.

Super Micro Computer, Inc. 超微电脑股份有限公司

San Jose AI server and edge computing leader; NVIDIA-Certified Systems from data center to industrial edge enabling robotics inference and automation AI.

Crusoe

Denver-based AI infrastructure company operating energy-first GPU cloud data centers, including a nuclear-powered facility built with Aalo Atomics.

Weldon Solutions

FANUC-certified robotic system integrator and CNC cylindrical grinder manufacturer based in York, PA.

Fireworks AI

Redwood City AI infra startup (2022) serving open-source LLMs like Kimi K3 via API, used by Cursor, Vercel and Notion.

MetaX 沐曦

Chinese fabless GPU company developing full-stack high-performance GPGPU chips for AI training, inference and graphics rendering.

Swissbit AG

Swiss industrial-grade storage and security MCU manufacturer serving robotics, automation, and edge IoT with certified flash and HSM solutions since 2001.

Polestar

Swedish electric performance car brand deploying ADAS and Level-3-capable autonomous driving tech in its EVs.

Tauro Technologies

US embedded systems design firm offering full turnkey development for robotics, autonomous vehicles, defense, and medical platforms.

Stellar Turing

US company offering Tamdrea, a cloud-native smart manufacturing digitalization platform with OEE monitoring, industrial AI chatbot, and production analytics.

inJoin the RobotToday community on LinkedIn

Daily robotics news, in-depth analysis, conference highlights, and discussions with professionals worldwide.