NVIDIA Delivers First Vera CPUs to Anthropic, OpenAI, SpaceX and Oracle

At a glance

NVIDIA has begun delivery of its first Vera CPUs, custom-designed processors for agentic AI systems, to partners Anthropic, OpenAI, SpaceX and Oracle Cloud.

What changed

NVIDIA’s Ian Buck personally delivered the initial Vera CPUs to the four organizations. Vera represents NVIDIA’s first custom CPU developed specifically for the requirements of agentic AI workloads. The company indicated this marks the start of broader Vera-powered system deployments.

Why it matters

Operationally, organizations can expect reduced latency and improved throughput for agentic workflows, lowering infrastructure costs and simplifying orchestration compared with general-purpose CPUs. Commercially, the availability of purpose-built silicon from NVIDIA strengthens its position in the AI infrastructure market and accelerates partner product roadmaps. For compliance-aware teams, the move supports more predictable performance and auditability of AI agent execution environments.

Key details

• Vera CPUs are optimized for agentic AI rather than traditional training or inference alone.
• Recipients include leading AI labs and cloud providers already operating at scale.
• Deliveries are the first in what NVIDIA describes as a longer roadmap of Vera-based systems.

Sources

• https://x.com/nvidia/status/2056494241904271780
• https://x.com/karpathy/status/2056753169888334312
• https://x.com/NVIDIAAI/status/2056887241432014959

Notes for citation

This article is based on official NVIDIA announcements and related statements from May 2026. Publication date should reflect the delivery timeline of 18–19 May 2026. All claims are drawn directly from the referenced posts.

Want to discuss how this affects your workflows? Book a call →

AI-assisted analysis by Skirr AI