Vera Rubin mass production and Vera CPU strengthen NVIDIA's CPU opportunity in AI factories

Overview

This report is Citigroup's rapid commentary on NVIDIA's Computex Day 1 announcements. The report focuses on Vera Rubin, Vera CPU, Spectrum-X Ethernet Photonics, and RTX Spark, arguing that these announcements reinforce NVIDIA's platform position in AI factories, agentic AI, and edge AI PCs.

Core views

The core view is positive: first, Vera Rubin has entered full production, reducing market concerns about delays in the product timeline; second, agentic AI makes the CPU a bottleneck for GPUs, increasing the importance and content value of Vera CPU; third, Vera CPU has already won support from early customers such as Anthropic and OpenAI; fourth, Spectrum-X and RTX Spark expand NVIDIA's ecosystem in networking and AI PCs.

Analysis framework

The report uses an event-driven commentary and product launch tracking approach, treating mass production progress, performance metrics, customer adoption, and ecosystem partners from the Computex keynote as the core evidence for judging NVIDIA's short-term catalysts and medium-term platform value, while using a P/E method to explain the basis for the target price.

Methodology notes

• Event-driven researchProduct launch catalyst analysis

  Assess the impact of an event on a company's fundamentals and market expectations through new product mass production, performance improvement, customer adoption, and ecosystem partners.

  Citigroup views Vera Rubin mass production, Vera CPU performance improvements, adoption by Anthropic and OpenAI, and the launch of RTX Spark as positive incremental developments.

• Valuation methodP/E valuation

  Multiply a target P/E ratio by forward earnings power and discount it back to the present to derive the target price.

  The report says NVDA's $300 target price is based on approximately 28x P/E and about $11.9 in C27E EPS power, with the 28x P/E in line with the three-year average.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• NVIDIA CORP (NVDA.O)

  Core covered target and beneficiary of the positive event

  Strengths

  Vera Rubin mass production, leading Vera CPU performance, strong AI factory system-level capabilities, and customer support from Anthropic and OpenAI.

  Weaknesses

  The report does not provide the current share price or quantified short-term order data, and some views rely on product launch information.

  Comparison

  Vera CPU has 3x and 6x performance advantages over AMD/Intel x86 solutions in SQL and data processing scenarios, respectively.

  Risks

  Competition in gaming platforms, slower-than-expected adoption of new platforms, volatility in automotive and data center demand, and the impact of crypto mining on gaming sales.

• COREWEAVE INC (CRWV.US)

  Early ecosystem partner and adopter of NVIDIA Spectrum-X Ethernet Photonics

  Strengths

  As an AI cloud infrastructure participant, it is positioned to benefit from networking upgrades in million-GPU AI factories.

  Weaknesses

  The report mentions it only as an adopter and does not expand on the company's fundamentals.

  Comparison

  Listed alongside Microsoft and OCI as an early ecosystem partner.

  Risks

  The pace of adoption, deployment scale, and commercialization impact are not quantified in the report.

Key data

• Report date2026-06-01The main text shows the publication time as 01 Jun 2026 02:52:29 ET.
• Target price$300Based on approximately 28x P/E and about $11.9 in C27E EPS power.
• Vera CPU relative to x86 performance3x for SQL workloads; 6x for data processingThe comparison targets are x86 competitors such as AMD/Intel.
• Vera CPU general agent tool performance1.8xCovering agentic AI tools such as Python, code analysis, and compilation.
• Cumulative Grace shipmentsNearly 2.5MThe report uses Grace's shipment base to demonstrate Vera CPU continuity.
• Vera Rubin system throughput10x improvement in scaled agent throughput versus BlackwellThe platform consists of the Vera Rubin NVL72 GPU system, Vera CPU, BlueField-4 STX storage, and Spectrum-6 networking.
• Spectrum-X Ethernet Photonics5x better power efficiency; 5x higher AI uptime; 1.3x faster deployment timeCoreWeave, Microsoft, and OCI are early ecosystem partners and adopters.
• RTX Spark capabilities1 petaflop of AI compute; up to 128GB unified memoryUsed for AI, creative, and gaming scenarios, capable of running 120B-parameter LLMs and 1M-token context.

Impact & implications

If Vera CPU becomes a key bottleneck-solving solution in agentic AI factories, NVIDIA's CPU content value could rise significantly, driving the data center platform beyond GPU sales into CPU, networking, storage, and complete system expansion. RTX Spark also shows NVIDIA is trying to extend AI capabilities to Windows PCs and local agent scenarios.

Risks

• Competition in gaming could lead to NVIDIA losing market share and pressure the stock price.
• Slower-than-expected adoption of new platforms could weigh on data center and gaming sales.
• Volatility in the automotive and data center markets could increase volatility in the stock and valuation multiples.
• Changes in crypto mining demand could affect gaming segment sales.
• Catalyst events and related stock price reactions may not materialize as expected.

What to watch

• The actual shipment pace and supply chain ramp after Vera Rubin enters full production.
• Adoption progress of Vera CPU at Anthropic, OpenAI, and other AI labs and cloud service providers.
• Whether the CPU-to-GPU content ratio continues to evolve toward nearly one-to-one.
• The deployment scale of Spectrum-X Ethernet Photonics at customers such as CoreWeave, Microsoft, and OCI.
• Market acceptance of RTX Spark and AI Windows PCs in local agent, creative, and gaming scenarios.