Nvidia CEO Jensen Huang says firms cannot ignore the OpenClaw second.
“Each firm on the earth in the present day must have an OpenClaw technique, an agentic system technique,” Huang mentioned throughout Nvidia’s 2026 GTC convention in San Jose on Monday. “That is the brand new laptop.”
Huang provided deep reward for OpenClaw, beforehand often known as Clawdbot and Moltbot, an open-source AI agent that has taken Silicon Valley by storm. Although OpenAI lured away OpenClaw creator Peter Steinberger, the service will stay on as an open-source venture.
“OpenClaw has made it attainable for us to create private brokers,” he mentioned. “The implication is unbelievable.”
Huang mentioned that OpenClaw “gave the trade precisely what it wanted at precisely the time.”
OpenClaw will do for AI what Home windows did for private computing, Huang mentioned. He additionally in contrast OpenClaw to different influential applied sciences, such because the Linux working system, the Kubernetes cloud venture, and HTML.
“It made it attainable for all the trade to seize onto this open-source stack and go do one thing with it,” Huang mentioned.
Huang mentioned there’s one huge caveat about utilizing OpenClaw that Nvidia has been working to handle: safety.
Nvidia introduced its spin on OpenClaw, known as NemoClaw, which permits customers so as to add privateness and safety controls to their AI brokers, or “claws.”
“With one command, anybody can run always-on, self-evolving brokers anyplace,” Nvidia mentioned in its announcement for NemoClaw, which it known as “an open supply stack that provides privateness and safety controls to OpenClaw.”
“It has a community guardrail, it has a privateness router, and in consequence, we may defend and hold the claws from executing inside our firm, and do it safely,” Huang mentioned.
Nvidia is selling NemoClaw at its convention by internet hosting a “build-a-claw” occasion the place attendees can develop their very own custom-made AI agent.
“OpenClaw brings folks nearer to AI and helps create a world the place everybody has their very own brokers,” Steinberger mentioned in a press release launched by Nvidia. “With Nvidia and the broader ecosystem, we’re constructing the claws and guardrails that permit anybody create highly effective, safe AI assistants.”
Huang made a number of different bulletins at Nvidia GTC on Monday, together with a brand new inference system that includes expertise from Groq, an AI chip startup with which Nvidia made a $20 billion deal. He additionally projected $1 trillion in demand for its Blackwell and Rubin AI chips by way of 2027.
