Undeniably, over the past two years as we have fully entered the AI era, the technology world has revolved almost entirely around Nvidia. Once a gaming graphics card manufacturer, Nvidia has grown into the world's most influential AI infrastructure company. Its hardware has become the key engine driving its market value to surpass multiple trillion US dollars. The "Blackwell" chip family. The powerhouse behind advanced AI workloads worldwide.
However, today investors are focusing on something potentially even greater than Blackwell — a new AI platform called "Vera Rubin." Jensen Huang, Nvidia's CEO, has hailed this as the company's new hope for the Agentic AI era going forward.
Vera Rubin is not just a new chip; it is a massive AI Infrastructure Platform that integrates everything, including:
All are designed to work together like an "AI Factory" — in other words, Nvidia will no longer sell just chips but will offer components that assemble into a full AI platform within data centers.
Previously, the world knew Blackwell chips during the race to build "Foundation Models" or basic AI models. Vera Rubin represents the era of Agentic AI, which actively performs tasks and requires a far more complex infrastructure, not just powerful GPUs.
These needs motivated Nvidia to build Vera Rubin to control and manage large AI workloads, especially Agentic AI tasks that require enormous memory to handle data and rapid analysis occurring simultaneously across multiple sectors.
This new hardware milestone has excited investors, especially after Nvidia stated that Vera and Rubin offer ten times the performance per watt compared to Blackwell, even though total power consumption nearly doubles. In the new AI era, the crucial question is not how powerful the chip is but how many tokens it can generate per unit of electricity used, since new AI models consume massive amounts of energy.
This is particularly true for reasoning AI, AI agents, long-context inference, and multi-agent orchestration. In other words, data center electricity costs are becoming the "largest constraint" in the AI industry, and Vera Rubin is designed specifically to address this problem.
After launching the "Vera CPU," a new AI data center CPU built on a superchip architecture with 88 custom high-performance cores called Olympus Cores and supporting the Armv9.2 instruction set, Nvidia claims it is the world's first CPU supporting FP8 precision. It can work alongside next-generation GPUs like Rubin GPU via high-speed NVLink-C2C connections, allowing CPU and GPU to share memory seamlessly.
This move places Nvidia in direct competition with Intel and Advanced Micro Devices (AMD) in the data center CPU market. However, Nvidia's difference lies in selling not just standalone CPUs but entire super AI compute systems.
In the first quarter 2027 earnings announcement, the company revealed that the new AI rack system featuring "Vera CPU" will outperform Grace Blackwell and become a new "Growth Driver" for Nvidia as the industry moves into the Agentic AI phase, meaning the AI market is expanding.
Nvidia believes AI is evolving from chatbots or intelligent assistants into operating systems capable of autonomous thinking, planning, tool usage, and decision-making. "The future of AI will include not only ChatGPT, Gemini, or Claude but also the integration of multiple AI branches, especially in business and manufacturing sectors.
Examples include AI customer service, AI software engineers, AI factories, AI robotics, and the emerging concept of Autonomous Enterprises driven by AI agents. This means the AI infrastructure market in the Agentic AI era will be many times larger than the Generative AI era, and Vera Rubin is designed to dominate this market — a point the market is beginning to recognize."
Some investors are betting that the Agentic AI era will surpass the ChatGPT era because current customers who have announced or are preparing to deploy Vera Rubin systems include hyperscalers, AI labs, and global cloud providers, with confirmed clients such as:
At this point, it is clear that "Vera Rubin" is not a single chip but an AI supercomputer platform composed of GPUs, CPUs, networking, cooling, and rack architecture, all designed collaboratively. This means customers who previously purchased only GPUs are now likely to buy Nvidia's full-stack ecosystem platform, leading to expectations that Vera Rubin could generate greater revenue and profit than Blackwell from clients building Agentic Systems or integrating AI stacks into their businesses worldwide.
Source: Nvidia [1] , [2] , [3] , Yahoo Finance , CNBC
Follow the Facebook page: Thairath Money at this link -