Agentic AI has always called for a different kind of CPU. NVIDIA CEO and founder Jensen Huang introduced the answer — the standalone Vera CPU — at GTC San Jose in March as NVIDIA’s next multi-billion dollar business.
On Friday, that CPU went from NVIDIA’s labs into customer hands.
The first NVIDIA Vera CPUs arrived at three of the world’s leading AI labs on Friday — Anthropic in San Francisco, OpenAI in Mission Bay, SpaceXAI in Palo Alto — followed by a delivery to Oracle Cloud Infrastructure in Santa Clara on Monday. NVIDIA Vice President of Hyperscale and High-Performance Computing Ian Buck hand-delivered them.
“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,” Buck said.
NVIDIA’s Ian Buck hand-delivered the first-ever NVIDIA Vera CPUs to our partners @AnthropicAI, @OpenAI, @SpaceX, and @OracleCloud. 🎉
Vera is NVIDIA’s first custom CPU, purpose-built for the age of agentic AI. This is just the beginning. The road to Vera-powered systems starts… pic.twitter.com/Ep5PLqcqIa
— NVIDIA (@nvidia) May 18, 2026
The big idea: imagine you could work 10x faster. Could your computer keep up? Agentic AI puts more demand than ever on the infrastructure we use to do all kinds of work — from building slides to compiling and testing software, analyzing data, searching files or even running simulations.
AI agents don’t run on GPUs alone. Every agentic sandbox, every tool call, every orchestration layer, every long-context retrieval operation — that’s CPU work. Vera is a new class of CPU designed with that reality as its starting point.
This gauntlet of concurrent, real-time tasks puts pressure on CPUs in ways traditional core-density focused designs were never built to prioritize. Vera packs 88 custom NVIDIA-designed Olympus cores, 1.2 TB/s of memory bandwidth and 50% faster per-core performance. Under constant load, work completes more quickly — increasing the efficiency of the entire AI factory and helping users get their work done with faster responses.
Vera Heads to San Francisco and Anthropic
The first delivery landed at Anthropic’s sleek SoMa offices in San Francisco.
James Bradbury, Anthropic’s head of compute, took the handoff from their conference room near the Bay.
Buck, aided by a bare NVIDIA Vera CPU motherboard he carried as a guide, walked Bradbury through the server built around the new CPU, talking through the features that make Vera different.
“Scaling compute is an important accelerant for the growth of models,” Bradbury said. “We’re excited to see Vera emerge as a promising part of the ecosystem when solving for agentic workloads.”
Feeding OpenAI’s Workloads
At OpenAI’s Mission Bay headquarters, the handoff moved outside — to an open-air balcony off the main offices.
The famously moody San Francisco weather cooperated on this day as Sachin Katti, head of compute infrastructure at OpenAI, thanked Buck for bringing the server over.
Buck walked through Vera’s features and, at one point — retrieving a screwdriver from his pocket — pulled off the lid to reveal the system’s insides.
On the Peninsula With SpaceXAI
The day’s final delivery took place at SpaceXAI’s offices in Palo Alto.
NVIDIA’s team walked Elon Musk through the system’s interior. Musk listened, then started asking questions — about cores, about memory layout, about cooling.
SpaceXAI is evaluating Vera for reinforcement learning workloads and the agent-based simulation pipelines that drive its training stack.
Vera Comes Home to the South Bay and OCI
On Monday inside the Oracle AI Customer Excellence Center, a team from OCI, including Karan Batta, who leads overall product management, and Gary Miller, chief customer and partner success officer, took a tour of the unboxed Vera CPU system. In the background, an NVIDIA GPU rack spun through OCI customer workloads from around the globe. The center is where Oracle customers come to kick the tires on a variety of AI workloads.
Buck explained how Vera will help.
“When AI models are posed a question, the answer, often, isn’t already prepped and ready to go. “The models actually have to generate some Python code to arrive at the correct answer,” Buck said. A task at which the Vera CPU excels. “That’s why we are seeing the demand for CPUs skyrocket,” Buck continued.
A trend the OCI team was also witnessing.
“OCI plans to deploy hundreds of thousands of NVIDIA Vera CPUs beginning in 2026 because agentic AI demands sustained performance at massive scale,” said Batta. “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.”
OCI is the first cloud provider to deploy Vera at hyperscale. For enterprise customers, that means production-grade agentic AI infrastructure at a scale no other cloud provider can match today.
The OCI team was eager to put Vera to work, offering their customers another system to customize and validate their agentic AIs and workloads, Miller said. “I am really looking forward to the reaction of people who come through here, and working together to get the most from Vera,” he said.
What Vera Delivers
Vera is part of NVIDIA’s extreme co-design story, alongside the NVIDIA Rubin GPU, BlueField 4 DPU, Spectrum-X and MGX rack architecture.
In addition to powering standalone CPU systems, Vera is the host processor for Vera Rubin NVL72 where it pairs via second-generation NVIDIA NVLink-C2C to a pair of Rubin GPUs.
In these systems, Vera and Rubin share a unified memory architecture that keeps accelerated compute highly utilized.
Vera’s fast CPU cores and interconnect handle orchestration, control, and data movement needed to feed GPUs at 2x the energy efficiency of traditional infrastructure.
The age of agentic AI has a purpose-built CPU, and its name is Vera.
Learn more about the NVIDIA Vera CPU.
Categories:
Corporate
Related News
