Jensen Huang: Why companies need open agent systems

NVIDIA founder and CEO Jensen Huang sits down with Harrison Chase to discuss why the last six months finally made AI useful, and what it takes to turn a large language model into a real, deployable product. The path there, Jensen says, is building your own "super agents": domain-specific systems wrapped in an open harness, grounded in your data, and improved over time. NVIDIA and LangChain also announce a new blueprint for running Deep Agents with Nemotron 3 Ultra inside OpenShell, a secure, open runtime, giving every enterprise the building blocks to create and deploy super agents anywhere.

We also discuss:
• Why companies of the future are built on harnesses, not business processes
• The difference between open vs. frontier models
• How Nemotron 3 Ultra reaches near-frontier performance at a low cost
• Why cheaper, faster intelligence finds better answers
• When to start with frontier models and when to specialize
• Why more AI means more jobs

Timestamps:
00:00 Introduction
00:34 Why NVIDIA invests in an open agent ecosystem
04:14 How to specialize agentic systems
05:54 Nemotron 3 Ultra hits frontier performance in Deep Agents
06:48 How cost changes the equation for builders
08:50 Frontier vs. open models: when to use each
09:50 Building specialized super sub-agents
11:16 Advice for enterprises: when to specialize
13:10 Companies built on harnesses, not business processes
14:48 Why open stacks empower enterprises
17:25 Announcing the Deep Agents + OpenShell blueprint
18:53 Runtime, security, and access control
20:21 How much should we anthropomorphize agents?
22:12 Why more AI means more jobs
24:17 The missing pieces of the agentic stack

Referenced:
• Anthropic: https://www.anthropic.com/
• Claude Code: https://www.anthropic.com/claude-code
• Claude Opus: https://www.anthropic.com/claude/opus
• Codex: https://openai.com/codex/
• Deep Agents: https://docs.langchain.com/oss/python/deepagents/overview
• Deep Agents + OpenShell blueprint: https://www.langchain.com/blog/nvidia-enterprise
• Deep Agents + OpenShell blueprint (GitHub): https://github.com/langchain-ai/openshell-deepagent
• DeepSeek: https://www.deepseek.com/
• Google DeepMind: https://deepmind.google/
• MiniMax: https://www.minimax.io/
• NVIDIA: http://nvidia.com/
• NVIDIA DGX Spark: https://www.nvidia.com/en-us/products/workstations/dgx-spark/
• NVIDIA Nemoclaw: https://www.nvidia.com/en-au/ai/nemoclaw/
• NVIDIA Nemotron: https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/
• OpenAI: https://openai.com/
• OpenClaw: https://openclaw.ai/

Where to find Jensen:
• LinkedIn: https://www.linkedin.com/in/jenhsunhuang/

Where to find Harrison:
• LinkedIn: https://www.linkedin.com/in/harrison-chase-961287118/
• Twitter/X: https://x.com/hwchase17

Where to find LangChain:
• Website: http://langchain.com
• Docs: https://docs.langchain.com/