Jensen fans don’t need a new chip. He needs a new moat.

Nvidia Owned the first era of AI — CEO Jensen Huang is making sure to own the next one. He is transforming Nvidia from a chipmaker that helps drive market cycles to an operating system for the future of artificial intelligence.

This change has gone largely unnoticed and has not yet been priced in by investors. But this week saw the clearest sign yet.

At Nvidia’s annual developer conference GTC, Huang announced NemoClaw, an open-source, chip-agnostic platform for building and deploying AI agents, autonomous software programs at the heart of the industry’s latest advances.

“Every company in the world should have an agent system strategy,” Huang says. “This is my new computer now.”

While the new chip announcements got the most attention at GTC, the NemoClaw announcement is a more important strategic shift and shows what Nvidia is really becoming.

Nvidia won the AI ​​training era by locking in users. The company’s chip and software ecosystem is so deeply embedded in how it builds its AI models that it would have been nearly impossible to switch to a competitor.

However, as the industry moves from building and training models to running them, inference workloads no longer require the same lock-in. google, Amazon and broadcom Everyone is building their own chips for inference. The moat that made Nvidia the world’s most valuable company is fading.

Selling chips, even the best chips, ultimately means selling them in cycles. Owning the platforms these chips run on will be a more permanent business. It’s more sticky, has higher margins, and doesn’t shift easily. So Juan goes on the offensive using Nimocrow.

NemoClaw is built on OpenClaw, an open source agent created by an individual developer that went viral earlier this year and became the fastest growing open source project in history. Open source means anyone can download, modify, and run the software locally on their own server. While this is powerful, it is also dangerous because no company controls what the agent can access on the machine.

Companies banned OpenClaw due to increased security risks. Nvidia’s version adds guardrails such as security tools, privacy routing, and data controls.

“Open” sounds generous, but it’s strategic for Nvidia. Nvidia transfers the layer that drives adoption and monetizes what lies underneath: the chips and computing power that all AI agents actually need to run. microsoft We didn’t charge for Internet Explorer, and Google didn’t charge for Android, but we enabled the introduction of Windows and search that we could monetize.

Huang is following that strategy and is not charging NemoClaw. The product is the platform. Mark Zuckerberg spent years and tens of billions of dollars on the Metaverse, trying to escape dependence on platforms owned by Apple and Google. Huang is trying to make sure Nvidia doesn’t end up in that position.

The most aggressive part of Huang’s strategy is that it is a direct threat to some of its top customers. Today, Nvidia relies on a handful of companies that build the most powerful AI models, including OpenAI, Anthropic, Google, and Meta. If either of them becomes dominant enough, it will gain leverage to squeeze Nvidia in pricing.

NemoClaw, named after Nvidia’s existing NeMo AI framework, prevents that from happening. One AI CEO, who requested anonymity to speak candidly about the issue, called this a classic “commercialization of complements” strategy. If companies can deploy AI agents for free through NemoClaw, it will be very difficult for OpenAI and Anthropic to charge a premium price for their versions. In open source, the model layer is fragmented, with hundreds of companies building and running their own models, but no one large enough to dictate the terminology. Nvidia will be in the middle, and demand for GPUs will skyrocket.

Nvidia is also stepping into a gap that no one else has filled, at least in the United States. Meta pioneered open source AI with its Llama model, but the next frontier model could reportedly be closed. Google and OpenAI keep their best models exclusive, and Anthropic has never released an open weight. The U.S. open source bench is thinner than it has been since the AI ​​boom began.

Meanwhile, Chinese research labs are accelerating their open source efforts. DeepSeek has proven that frontier models can be built for a fraction of the amount spent by American labs. alibabaByteDance, and others followed suit.

Data from OpenRouter, which tracks real-world model usage, shows that four out of the five most popular models on its platform this month are open source, and most are made in China. OpenRouter’s rankings are limited to its own customer base, and developers contracted by companies typically use API tools from model companies.

Can a chip maker really become an operating system?

History would suggest otherwise. past attempts intel and IBM Didn’t go anywhere. But Huang has made successful platform transitions before, pivoting Nvidia away from gaming and into crypto, cloud, and AI training. Nvidia just posted a 73% revenue increase last quarter. The latest forecast of nearly $80 billion for the fiscal first quarter shattered expectations.

Networking alone is now a multibillion-dollar business for Nvidia, which barely existed three years ago. No CEO in the semiconductor industry has a better track record of sensing this change early and proactively repositioning to take advantage of it.

For NemoClaw to matter, it needs to be adopted by businesses. Nvidia’s open source model is free, but so far unproven compared to what is being shipped by Chinese labs. And if Meta changes direction or Google releases its model, the vacuum Huang is stepping into could quickly close.

Representatives for Meta and Google did not respond to requests for comment.

The question investors should be asking is not whether NemoClaw will still work tomorrow. It’s whether Nvidia is still a chipmaker or an operating system. One is selling into cycles and the other is selling compound. The market has priced in the former, but if Hwang achieves this, it should price in the latter.

Watch: Interview with Jim Cramer and Jensen Huang