Nvidia builds a ‘new stronghold’ after success with AI chips

“Every company in the world should have a strategy for agent systems. This is the new computer today,” Nvidia CEO Jensen Huang said at the annual GTC 2026 event in mid-March, when introducing NemoClaw – an agent platform based on OpenClaw but integrating enterprise-level security and privacy features.

Theo CNBCthe announcement of NemoClaw is an important strategic shift that shows that Nvidia is really making a new move, wanting to clearly shape its true plan.

The chip manufacturing model is not enough

Nvidia is considered to be winning in the AI ​​era by “holding tightly” to users with its chips and software ecosystem. The company’s products have had a profound impact on how AI models are built, making it nearly impossible for those who want to change to a competitor in the same industry.

Theo Business InsiderNvidia holds the advantage thanks to special know-how, such as the Cuda parallel computing platform – a software layer that allows developers to exploit the power of GPUs for general-purpose tasks. The company’s AI chips also account for the majority of the market. At GTC 2026, Mr. Huang predicted that Nvidia would reach $1,000 billion worth of chip orders by the end of this year, double his estimate last year and nearly 5 times the company’s 2025 revenue.

However, the AI ​​industry is shifting from model building and training to operations. According to experts, the workload of those models does not require the same level of deep dependence as before. Google, Amazon and Broadcom are all developing their own optimized chips.

The “stronghold” that once made Nvidia the most valuable enterprise in the world is starting to weaken. The barriers to entry that have made the company the most valuable name in the world are gradually disappearing. Selling chips, even the best chips, is ultimately a cyclical sale. Meanwhile, owning a platform to create space for new operating chips is a more sustainable model in many aspects: closely linked, high profit margin, and difficult to replace. That is the vision of CEO Nvidia with the “card” NemoClaw.

Platform strategy with NemoClaw

NemoClaw is built on OpenClaw, an agent platform developed by engineer Peter Steinberger in November 2025. According to Mr. Huang, Nvidia cooperated with Steinberger to develop NemoClaw, although this person is currently working for OpenAI.

Nvidia’s version adds “safeguards” like security tools, privacy routing, and data controls. Talk to New StackKari Briski, Nvidia’s Vice President of Generative AI Software, affirmed that this is a “safe version for AI agents”, and “applies security mechanisms based on network and privacy policies”.

Nvidia still emphasizes the “open” element of NemoClaw. According to experts, it sounds generous, but that is the new strategy. The company is broadcasting its message for free, while also profiting from what’s behind it, the chips and computing power that every AI agent needs to function.

The same thing happened: Microsoft did not charge for Internet Explorer, Google did not charge for Android, but they expanded the ecosystem to make money in Windows and the Google Search search engine. Huang learned exactly this “move” by not charging NemoClaw.

Putting customers in a “commodity” position

Theo CNBCthe bold thing about Jensen Huang’s strategy is to directly create a “threat” to some of the company’s top customers. Nvidia currently depends on a few artificial intelligence companies such as OpenAI, Anthropic, Google or Meta. If one of them has the upper hand, they can force prices back against Nvidia.

“NemoClaw was born to prevent that situation from happening,” an anonymous CEO in the AI ​​field told CNBC. “This is the classic ‘commercialization of supplements’ strategy.”

Also according to this CEO, when businesses can deploy AI agents for free through NemoClaw, companies like OpenAI or Anthropic will find it difficult to charge high fees. Open source keeps the model layer fragmented, creating the opportunity for hundreds of companies to participate and operate their own models, with no one in a sole position to dictate the conditions. Nvidia will play a central role, and GPU demand will explode.

Besides, Nvidia is considered to be entering a gap that no competitor has filled, at least in the US. The open source ecosystem in the US is declining in quality compared to the early stages of the AI ​​wave. Meta pioneered open source with the Llama, but the next-generation model is expected to be closed source. Google and OpenAI have long maintained the best models as proprietary, and Anthropic has never published model weights.

Theo TechCrunchopen source is now stronger in China as a series of businesses in this country join the race. DeepSeek has demonstrated that advanced models can be built at low cost. Alibaba, ByteDance and many other companies also followed with their own products. Data from OpenRouter (a platform that provides unified APIs to access all major LLMs on the market) shows that 4 of the 5 most popular models on the platform in February were open source, most of them from China.

However, as Nvidia’s strategy becomes clearer, the question is: Is it possible for a chipmaker to become a platform company?

In fact, history records previous attempts by Intel and IBM, but they all failed. But Jensen Huang may be a different case. He successfully pivoted Nvidia from developing equipment for the gaming industry to cryptocurrency, cloud computing, and most recently AI training. The network infrastructure segment that was almost unknown three years ago has now become a billion-dollar business for the company. According to Reutersup to now, no CEO in the semiconductor industry has had better early trend vision and timely pivots than Huang.

However, the key issue is that NemoClaw needs to be accepted by businesses. The space that Nvidia is taking advantage of could also quickly disappear if Meta changes direction or Google expands their model.

“So the question is not whether NemoClaw will go live tomorrow, but whether the company will maintain its position as a chip company or shape its platform development.” CNBC comment. “One direction sells according to the cycle, the other direction creates cumulative growth. If Huang is successful, Nvidia should be valued in the second direction, not just a chip company.”