Dell Technologies World 2026 Keynote | May 18–21 | Las Vegas

2026-06-05 · Source: NVIDIA · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Cloud Computing & IT Infrastructure · Depth: Advanced, medium

Summary

Nvidia CEO Jensen Huang, speaking at Dell Technologies World 2026, highlighted the advent of "useful AI" driven by agentic systems. These systems understand, reason, plan, use tools, and iterate, leading to a 100X to 1,000X increase in computational demand for tasks that can run for a week, yet deliver significant productivity gains. Huang detailed Nvidia's architecture for agents, which integrates large language models on powerful systems like the NVLink72, a secure NVIDIA Open Shell sandbox, and a NemoCloud reference harness running on a CPU. The architecture leverages Nvidia's Vera CPU, optimized for token generation with industry-leading single-threaded performance and three times the memory bandwidth. This comprehensive platform supports all frontier and open-source AI models, operates across cloud and local environments, and incorporates confidential computing. A demonstration featured a desk-side computer capable of running a one trillion-parameter AI model, a feat previously considered impossible.

Key takeaway

For AI Architects evaluating infrastructure for next-generation AI, recognize that agentic systems demand a 100X to 1,000X increase in computational resources. Your strategy should prioritize hybrid AI architectures that support both cloud and local deployment, leveraging specialized CPUs like Nvidia's Vera for efficient token generation. Consider confidential computing for sensitive models and plan for orchestrating multiple agents to maximize productivity gains in areas like software development and DevOps.

Key insights

Agentic AI drives unprecedented productivity and computational demand, requiring specialized, hybrid architectures.

Principles

• Agentic systems require 100-1000x more computation.
• AI's economic value shifts to token generation.
• Hybrid AI architectures support diverse models.

Method

Nvidia's architecture for agents involves a secure sandbox (NVIDIA Open Shell), a CPU-based harness (NemoCloud), and large language models running on high-performance systems or in the cloud.

In practice

• Deploy agentic systems for software development, DevOps, QA.
• Orchestrate multiple agents for complex tasks.
• Utilize confidential computing for sensitive models.

Topics

• Agentic AI
• NVIDIA Architecture
• Vera CPU
• LLM Inference
• Confidential Computing
• Hybrid AI

Best for: CTO, VP of Engineering/Data, MLOps Engineer, AI Architect, Director of AI/ML, AI Engineer

Related on AIssential

• Jensen Huang says he’s found a ‘brand new’ $200B market for Nvidia — Nvidia's Vera CPU targets a new \$200 billion market for agentic AI by optimizing for rapid token processing.
• X86: The Enterprise Engine to Scale AI-Factory Deployments — Enterprise agentic AI requires robust, compatible infrastructure, prioritizing software continuity and memory capacity over raw accelerator speed.
• How Dell is Scaling Enterprise AI with NVIDIA Vera Servers — Dell and NVIDIA are scaling agentic AI from pilot to production with a unified infrastructure, leveraging Vera CPUs and the Dell AI Factory.
• Meta to Spend Billions on AMD Gear, AI Scare Trade Continues — Diversification in AI chip procurement and specialized AI agents are reshaping market dynamics and software valuations.
• Nvidia CEO Jensen Huang: The idea that AI will destroy software is "ridiculous" — AI agents will integrate with and enhance existing software, driving a shift towards specialized "token factories" and extreme co-design in computing infrastructure.
• NVIDIA Vera CPU Sets a New Standard for Agentic Workloads in AI Factories — The NVIDIA Vera CPU optimizes agentic AI and reinforcement learning by making CPU execution a high-performance, critical path within the AI loop.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by NVIDIA.