Nvidia Unleashes Vera CPU, Eyes $200 Billion AI Market

Nvidia is making a bold play for the future of artificial intelligence, unveiling a new processor designed to power the next generation of AI systems. The company has launched its Vera CPU, a specialized chip built from the ground up to handle "agentic AI," a technology poised to revolutionize how AI interacts with the world. This strategic move positions Nvidia at the forefront of a new hardware market, estimated to be worth a staggering $200 billion, and has reportedly prompted a strategic re-evaluation from major industry players.

The official rollout of the Nvidia Vera CPU and the comprehensive Vera Rubin AI supercomputing platform took place on March 16, 2026. The highly anticipated launch was a centerpiece of the company's annual GTC 2026 conference, held in San Jose, California. Jensen Huang, Nvidia's founder and CEO, presented the Vera CPU to the world, proclaiming it the "world's first processor purpose-built for the age of agentic AI and reinforcement learning." He emphasized the chip's efficiency, stating it delivers results with twice the efficiency and 50% faster performance compared to traditional rack-scale CPUs.

The Vera Rubin platform itself was introduced as a full-stack solution, featuring seven new chips that are already in full production. This integrated system is designed to scale the world's largest AI factories, built on an "extreme co-design" philosophy where the GPU, CPU, networking, security, and cooling components are all architected together as one cohesive unit. Key components include the Vera CPU, the Rubin GPU, the NVLink 6 Switch, ConnectX-9 SuperNIC, BlueField-4 DPU, Spectrum-6 Ethernet switch, and the newly integrated Groq 3 LPU.

The journey to the Vera CPU's launch has been a multi-year effort, beginning with significant announcements from Nvidia's leadership. The initial reveal of the Rubin microarchitecture, which includes both the Rubin GPU and the Vera CPU, was made by CEO Jensen Huang at Computex in Taipei in 2024. Further details about the Vera Rubin platform, named in honor of astrophysicist Vera Rubin, were previewed at GTC 2025. This set the stage for subsequent developments, including the September 2025 announcement at the AI Infra Summit, where Nvidia introduced the Rubin CPX. This new class of GPU was specifically designed for massive-context processing, intended to seamlessly integrate with Vera CPUs and Rubin GPUs within the advanced Vera Rubin NVL144 CPX platform. At that time, Huang underscored the platform's significance, noting, "The Vera Rubin platform will mark another leap in the frontier of AI computing — introducing both the next-generation Rubin GPU and a new category of processors called CPX."

Agentic AI, the core focus of Vera, represents a significant evolution beyond traditional artificial intelligence. Unlike previous AI models that primarily answered questions, agentic AI systems are designed to autonomously execute tasks, browse systems, and manage complex workflows. This means moving from reactive models to systems capable of continuous operation, environmental interaction, and sequence-based decision-making. These advanced capabilities are what define the "agentic inflection" that Huang referenced.

By February 26, 2026, during Nvidia's quarterly earnings call, Jensen Huang publicly declared that agentic AI had reached "an inflection point." He highlighted that AI agents were "solving real problems," observing that the world had "awakened to the agentic inflection" within the preceding two or three months. Huang also shared his vision for the future, expressing his belief that the next major inflection point for AI would be "physical AI," involving the direct embedding of artificial intelligence into machinery such as self-driving cars and advanced robotics.

The Vera CPU itself boasts impressive technical specifications designed to meet the rigorous demands of agentic AI. It features 88 custom Nvidia-designed Olympus cores, equipped with Spatial Multithreading, which reportedly doubles the performance of its predecessor. The chip offers full Armv9.2 compatibility and is notable as the first CPU to support FP8 precision. Memory capabilities are robust, with up to 1.2 terabytes per second (TB/s) of memory bandwidth, providing twice the bandwidth at half the power compared to traditional CPUs, and supporting up to 1.5 TB of LPDDR5X memory. Vera is engineered to orchestrate next-generation AI systems at rack scale, efficiently managing, coordinating, and securing large-scale AI workloads, while ensuring optimal data flow between compute resources, networking fabrics, and storage. Its design is specifically optimized for continuous inference, intricate agentic AI workflows, and long-context reasoning. The accompanying Rubin GPU, a critical component of the platform, is manufactured by TSMC using a 3nm process, featuring 288 GB of HBM4 memory with 22 TB/s bandwidth, and delivering 50 PFLOPS of FP4 inference compute. The flagship Vera Rubin NVL72 rack configuration integrates 72 Rubin GPUs and 36 Vera CPUs, providing 3.6 EFLOPS of NVFP4 inference and 2.5 EFLOPS of training compute, all within a 100% liquid-cooled system engineered for hyperscale AI factories.

Following the GTC 2026 announcement, Nvidia swiftly began delivering its initial Vera CPU systems to key anchor customers. On May 19, 2026, Ian Buck, Nvidia's Vice President of Hyperscale and High-Performance Computing, personally oversaw the delivery of Vera CPUs to Anthropic in San Francisco, OpenAI in Mission Bay, and SpaceXAI in Palo Alto. A subsequent delivery to Oracle Cloud Infrastructure in Santa Clara occurred on May 20, 2026. Buck underscored the significance of these deployments, stating, "Agentic AI is creating a new CPU moment in the AI factory — as models move from answering to acting, Vera is purpose-built to keep that work moving at scale." Reports indicate that Elon Musk was present to accept the new Nvidia Vera CPU delivery for SpaceXAI, while James Bradbury, Anthropic's head of compute, was also on hand for their delivery. Karan Batta, who leads product management at Oracle Cloud Infrastructure (OCI), shared OCI's ambitious plans, commenting, "OCI plans to deploy hundreds of thousands of NVIDIA Vera CPUs beginning in 2026 because agentic AI demands sustained performance at massive scale." He further elaborated that Vera's architecture is "purpose-built for high-throughput reasoning workloads, delivering the efficiency, density and footprint OCI needs to power the next generation of enterprise AI."

Nvidia's aggressive move into this specialized CPU market, traditionally dominated by x88 and Arm server vendors, signifies a broader and critical shift within the AI industry. The orchestration layer of AI infrastructure is rapidly becoming a central battleground, with companies vying to provide the foundational hardware that can manage and accelerate these increasingly complex and autonomous AI systems. This development is not just about faster chips; it's about enabling a new paradigm of AI that can think, act, and operate with greater independence, fundamentally changing how businesses and individuals interact with artificial intelligence. The focus on "extreme co-design" also highlights the industry's recognition that optimal performance for AI at scale requires a holistic approach, where all components are developed in tandem to maximize efficiency and capability.

The introduction and immediate deployment of the Vera CPU mark a pivotal moment for Nvidia and the broader technology landscape. With major players like Anthropic, OpenAI, SpaceXAI, and Oracle Cloud Infrastructure already integrating Vera into their operations, the impact of agentic AI is poised to accelerate dramatically. The company's vision, articulated by Jensen Huang, extends beyond current applications, hinting at a future where physical AI permeates everyday machinery. As the race to build the most powerful and efficient AI factories intensifies, Nvidia's Vera CPU is positioned as a foundational technology, driving the next wave of intelligent systems.