Nvidia losing share as rivals sign deals: Seaport

Key Concepts

• AI Infrastructure: The foundational hardware and software stack required to support AI training and inference.
• Inference: The process of running a trained AI model to make predictions or generate content; identified as the next major growth phase.
• Hyperscalers: Large-scale cloud providers (e.g., Amazon, Microsoft, Google, Meta) that are the primary customers for AI compute.
• Sovereign AI: The trend of nations and large organizations building their own localized AI infrastructure.
• Supply Chain Constraints: Bottlenecks ranging from specialized chemical inputs to electrical infrastructure and manufacturing capacity (e.g., TSMC allocation).
• Vera Rubin: Nvidia’s next-generation architecture, noted for its efficiency in token pricing.
• CUDA: Nvidia’s proprietary software platform that creates a "moat" by integrating hardware and software.

---

1. Financial Performance Overview

Nvidia’s Q1 results exceeded market expectations, demonstrating significant year-over-year growth.

• Revenue: Reported at $81.62 billion, surpassing the estimated $79.19 billion.
• Data Center Revenue: Reported at $75.2 billion, beating the $73.48 billion estimate.
• Gross Margin: Adjusted gross margin reached 75%, a figure described as "incredible" for a company of Nvidia's scale.

2. Perspectives on Market Position

The discussion highlighted a divide between bullish and cautious outlooks regarding Nvidia’s future:

• The Bullish View (Hatem Diab):
  • Nvidia is the dominant player in the AI boom, successfully scaling its hardware and software ecosystem.
  • The company is effectively managing supply chain constraints by coordinating with memory providers and other partners.
  • The customer base is diversifying beyond hyperscalers to include sovereign AI initiatives and various large language model (LLM) developers.
• The Cautious View (Jay Goldberg):
  • While Nvidia executes well, the company faces extreme pressure to consistently beat high expectations.
  • Supply chain bottlenecks (e.g., potential labor strikes at Samsung, TSMC allocation limits) are forcing Nvidia to "leave money on the table."
  • The inference market is a "jump ball" where competition from Google (TPUs), AMD (MI450 series), and Amazon (Trainium) is intensifying.

3. Strategic Challenges and Industry Constraints

A significant portion of the debate focused on the physical and logistical limitations of the AI build-out:

• Industry Constraints: Jay Goldberg noted a list of 25–26 distinct constraints, including raw chemical inputs, electrical transformers, and a shortage of skilled electricians.
• Competition: The inference market is viewed as more fragmented than the training market. Competitors are gaining traction because Nvidia cannot meet the total global demand for compute.
• Execution Risks: Even with strong management, factors outside Nvidia’s control—such as geopolitical issues or manufacturing disruptions—pose risks to their deployment timelines.

4. Key Relationships and Future Outlook

• OpenAI Relationship: It was noted that OpenAI delayed a data center build-out with Oracle specifically to wait for Nvidia’s Vera Rubin architecture, citing its superior efficiency for frontier models.
• Inference as the Next Frontier: Both analysts agreed that inference is the critical next phase. While Nvidia currently leads, the market is expected to demand "true agentic CPUs" and diverse compute flavors, which may open doors for competitors.
• Synthesis: The consensus is that the demand for compute is massive and likely underestimated. While Nvidia remains the primary architect of the current AI infrastructure, the company’s long-term dominance will depend on its ability to navigate supply chain bottlenecks and maintain its technological lead against a growing list of specialized competitors.

Notable Quotes

• Hatem Diab: "The only brain that's making money in the economy is really AI infrastructure build out."
• Jay Goldberg: "It's not easy to be the king... there's so many headwinds now. There's so many constraints across the industry."