Alpha~
2026.05.21 07:00

NVDA FY2027 Q1 Earnings Call — Key Takeaways

portai
I'm LongbridgeAI, I can summarize articles.

The Most Important Incremental Driver: CPU + Agents Open a $200 Billion New TAM

Vera defined as a CPU purpose-built for agentic AI, unlocking a $200B incremental market

  • CPU revenue visibility for this year is approaching $20 billion (referring to standalone CPUs only, excluding the CPU value embedded in Vera Rubin systems)
  • Jensen explained: the agent's harness, I/O, orchestration, memory management, tool calls, browser, compilers, etc. all run on the CPU, while actual thinking / inference runs on the GPU
  • → CPUs do not cannibalize GPUs; rather, they represent incremental infrastructure demand driven by agentic AI

Incremental Driver #2: Inference Market Share Continues to Rise

  • NVIDIA's share in frontier AI compute and inference is increasing. Key variable: Anthropic has joined the partnership ecosystem (previously coverage was "essentially zero"), with large amounts of compute going live via AWS, Azure, CoreWeave, xAI, and others
  • OpenAI, Gemini, xAI, Meta, Mistral, Microsoft AI, Perplexity, Cursor, and others are all on the NVIDIA platform
  • Jensen stated: almost all frontier model companies will adopt Rubin from the start — this was not the case during the Blackwell era → Rubin will be more successful than Blackwell

Incremental Driver #3: Strong Visibility on the Rubin Ramp

  • Mass production and shipments begin in H2 2025: Q3 kickoff, Q4 continued ramp, with Q1 FY2026 expected to be very significant
  • Customer demand and purchase orders are largely in place
  • Vera Rubin vs. Blackwell: up to 35x inference throughput, up to 10x AI factory revenue
  • Google A5X bare metal instances can support up to 960,000 Rubin GPUs across multiple sites — an important customer validation signal

Incremental Driver #4: Classifying LPUs as a Niche Product

  • Actively tempering expectations around LPUs
  • SRAM-based accelerators like LPX (including Cerebras) are suited for low-latency, high token-rate premium use cases, but are limited in throughput, model capacity, and long-context handling
  • Still a niche product in the near term, currently well below 20% market share
  • May grow in premium token scenarios in the future, but the primary platform remains Grace Blackwell / Vera Rubin

Incremental Driver #5: Meaningfully Increased Shareholder Returns

  • Quarterly dividend increased from $0.01 → $0.25 per share
  • New $80 billion share buyback authorization added, on top of the remaining $39 billion from the prior plan
  • Reaffirmed commitment to return 50% of free cash flow this year

$NVIDIA (NVDA.US)

The copyright of this article belongs to the original author/organization.

The views expressed herein are solely those of the author and do not reflect the stance of the platform. The content is intended for investment reference purposes only and shall not be considered as investment advice. Please contact us if you have any questions or suggestions regarding the content services provided by the platform.