Particle.news
Download on the App Store

Nvidia Bets on Vera CPU With $20 Billion Revenue Target

The company says Vera will orchestrate agentic AI workloads by pairing a purpose‑built Arm CPU with Rubin GPUs in a planned H2 2026 platform rollout.

Overview

  • Nvidia has publicly signaled strong near‑term CPU ambitions, projecting nearly $20 billion in standalone Vera sales and sizing a roughly $200 billion addressable market for agentic AI orchestration.
  • In mid‑May Nvidia began delivering initial Vera production systems to major AI customers, with Ian Buck hand‑delivering units to OpenAI, Anthropic, SpaceX AI and Oracle and Dell reporting early PowerEdge shipments.
  • Vera is an Arm‑based design that Nvidia says uses 88 custom Olympus cores and about 1.2 TB/s of memory bandwidth and is tuned for orchestration, retrieval and long‑context inference in agentic AI workloads.
  • Analysts have forecast that Vera could outperform x86 rivals and drive sharp unit growth, but those projections are preliminary and hinge on supply constraints, hyperscaler demand and export limits.
  • If Nvidia completes the Vera Rubin platform rollout in H2 2026, data centers may shift CPU/GPU ratios toward purpose‑built CPUs for agentic tasks, a change that could affect cloud economics, server designs and vendor share.