OpenClaw: NVIDIA's New Agentic AI Software Is Changing the Game

Jensen Huang, NVIDIA's CEO, has called the new AI software, OpenClaw, one of the most important software releases in technology history. He mentioned during a conference organized by Morgan Stanley that the software has achieved astonishing adoption. In just three weeks, it reached a level of adoption that took Linux about 30 years to achieve. OpenClaw represents a new phase of AI known as Agentic AI, where the system isn't just limited to answering questions; it can perform tasks independently. This framework allows users to issue commands like "create," "develop," or "execute," and the agent will carry them out automatically. This includes advanced tasks such as browsing the internet independently, fixing software errors, and managing complex workflows within companies. Huang pointed out that this shift creates what he described as a "computational vacuum," as these systems require massive processing capabilities compared to traditional applications. For example, an agent working to solve a single software problem might consume a thousand times more data and processing power than a regular conversation. Continuous system operation to monitor and optimize workflows could even reach a million times more. This huge demand is a massive opportunity for hardware manufacturers, but it also poses a challenge for current infrastructure. Processors like NVIDIA Hopper and Blackwell were designed to train large models, but the industry is now shifting towards systems capable of handling long, complex tasks executed independently by AI agents. According to Huang, OpenClaw's adoption rate is accelerating at an unprecedented pace, to the point where the growth curve looks like a vertical line on a graph. Within NVIDIA itself, these agents have already started developing software and writing software tools, which could represent the new business model for tech companies in the future. To keep up with this increasing demand, NVIDIA is developing a new generation of hardware codenamed "Vera Rubin," focusing on increasing memory capacity and improving systems' ability to handle the long-context tasks required by Agentic AI applications. Huang concludes that organizations adopting these technologies early could achieve a significant leap in productivity, while companies that lag might find it difficult to catch up. As OpenClaw continues to develop, it seems the future of software will be completely reshaped around agents capable of working independently, opening new horizons for the technology and AI industries.