NVIDIA CEO Claims OpenClaw Achieved in 3 Weeks What Linux Accomplished in 30 Years; Demonstrating the Power of Agentic AI

NVIDIA’s CEO, Jensen Huang, recently spoke at the Morgan Stanley conference, highlighting a pivotal moment in artificial intelligence known as ‘agentic AI.’ He characterized the launch of OpenClaw as a landmark event in software development, describing it as potentially the most significant software release of our time.

Agentic AI’s Massive Shift in Token Demand

Huang explained that AI could be likened to a “5-layer cake, ”with the applications layer being particularly profitable for hyperscale companies and advanced research labs. OpenClaw exemplifies how AI, when applied in hyper-personalized settings, can achieve results that closely mimic human productivity. In light of rising enterprise requirements for AI, Huang explored several key industry turning points, notably mentioning OpenClaw’s rapid rise, which he claims has outpaced Linux in terms of adoption.

Probably the biggest phenomenon that’s happening, and if you’re paying attention to it, I’m sure you are, OpenClaw is probably the single most important release of software, you know, probably ever. If you look at OpenClaw and the adoption of it, you know, Linux took, right, some 30 years to reach this level. OpenClaw in, what is it, 3 weeks, has now surpassed Linux. It is now the single most downloaded open source software in history, and it took 3 weeks.

– NVIDIA’s CEO Jensen Huang

Huang expressed his fascination with the capability of agents like OpenClaw to perform complex tasks that typically necessitate specialized knowledge and extensive time investment. The ascendance of OpenClaw is not merely due to its advanced implementation; rather, it has starkly illustrated the practical applications of AI, directly enhancing the daily lives of consumers by simplifying repetitive tasks. For industry leaders such as NVIDIA, the introduction of agents like OpenClaw has sparked an unprecedented demand for computational resources.

One of the most striking revelations from Huang’s address is the soaring token consumption associated with agentic AI, which he noted has surged by an astounding 1, 000 times. This surge has resulted in a phenomenon described as a ‘compute vacuum’—a scenario where even vast hardware advancements will struggle to meet the ever-increasing token requirements as agentic AI continues to integrate into various human tasks. These AI agents can conduct extensive web searches, generate images, execute complex analyses, and tackle numerous workloads, thereby drastically increasing token usage, which signals a significant rise in compute demand for NVIDIA and similar companies.

Animated avatars labeled as various AI agents, including 'AI Research Assistant Agent, ' 'Virtual Lab AI Agent' — Image Credits: NVIDIA

In discussing the specialized compute architectures that correspond to the previously mentioned ‘5-layer cake, ’ Huang noted that the Hopper and Blackwell designs are centered on training workloads. However, with technologies like Vera Rubin, NVIDIA aims to tackle the constraints of agentic AI specifically by enhancing long-context workloads through increased onboard memory and platforms like ICMS. Given the substantial imbalance between compute resources and token demand, the anticipated need for the Rubin architecture is expected to be enormous.

Source & Images