Amidst the sustained surge in global demand for artificial intelligence computing, chip giant NVIDIA (NVDA) is solidifying its leadership position in the AI infrastructure sector through a series of strategic collaborations and product launches. At the GTC 2025 conference held recently in San Jose, NVIDIA CEO Jensen Huang announced a partnership with Peter Steinberg, the founder of the open-source AI agent project OpenClaw, to jointly launch an enterprise-grade product—NVIDIA NemoClaw. This product cleverly combines the ease of building AI agents with OpenClaw and the high-standard privacy and security control features required by NVIDIA for running these agents within enterprises.
In his keynote speech, Huang highly praised OpenClaw, calling it “the most popular open-source project in human history” and suggesting its importance rivals that of the HTML and Linux operating systems from the internet era. He stated, “OpenClaw democratizes the creation of personal agents. It can manage large language models, break down problems step-by-step, and even collaborate with other agents, featuring powerful interoperability. You could say it’s an operating system.”
However, Huang also pointed out a key challenge for enterprises applying AI agents: agents running inside an enterprise might access sensitive information and communicate with the outside world, which is not permitted from a security standpoint. It is precisely for this reason that NVIDIA chose to collaborate with OpenClaw, injecting enterprise-grade application capabilities into it, enabling seamless connection to the strategy systems of various SaaS companies, thereby ensuring the secure and compliant application of AI agents. Steinberg added, “Through collaboration with NVIDIA and the broader ecosystem, we are building ‘Claws’ and guardrails so that anyone can create powerful and secure AI assistants.” The core of NemoClaw lies in the deep integration of OpenClaw’s robust capabilities with NVIDIA’s self-developed Nemotron AI models.
Beyond software collaborations, NVIDIA also showcased its ambitious market blueprint on the hardware front. At the conference, Jensen Huang significantly raised the long-term revenue expectations for its AI chip business, projecting that the company’s next two generations of AI chips, Blackwell and Rubin, will cumulatively generate at least $1 trillion in revenue by the end of 2027. This figure far exceeds the previous forecast of approximately $500 billion by the end of 2026, fully demonstrating its confidence in the continued explosion of future AI computing demand.
To achieve this goal, NVIDIA announced several hardware innovations. The company revealed it would integrate technology from AI chip startup Groq to launch a new Groq 3 Language Processing Unit (LPU). This dedicated chip aims to accelerate inference computation for large language models and will act as a coprocessor working alongside existing AI accelerators to significantly improve system response speed. The chip will be manufactured by Samsung Electronics, with related systems expected to be launched in the second half of this year.
Simultaneously, NVIDIA showcased its new general-purpose central processing unit architecture, “Vera,” formally marking its entry into the traditional data center processor market. Huang believes the CPU business “will certainly be a multi-billion dollar market opportunity.” The Vera processor integrates the strengths of processors used in data centers, gaming PCs, and laptops, featuring high throughput, powerful computing capability, and low power consumption. NVIDIA plans to launch server systems composed entirely of Vera CPUs, usable independently or in coordination with other NVIDIA systems, adding a new form factor to its product line. The company’s next-generation flagship AI system, “Vera Rubin,” is expected to be unveiled in the second half of 2026, named in honor of the renowned astronomer Vera Rubin.
