Artificial Intelligence Primer

Objective: Understand the transformational shift driven by Generative AI and identify the winners across the value chain.

The Investment Thesis (The Why)

• Macro Trend: The shift from deterministic software (code) to probabilistic software (models) is the largest platform shift since the internet.
• The Moat: Data moats, proprietary model architectures, and massive compute scale are the primary defensible assets.
• Catalysts:
  • Release of next-gen frontier models (e.g., GPT-5, Gemini 2.0 Ultra).
  • Integration of agents into enterprise workflows.
  • Breakthroughs in energy efficiency for inference.

The Vertical Map

1. Layer 1: Compute & Infrastructure (The Pickaxes)
   • Description: The hardware and cloud platforms powering training and inference.
   • Examples: NVIDIA (NVDA), AMD (AMD), TSMC (TSM), Microsoft Azure (MSFT).
2. Layer 2: Foundation Models (The Brains)
   • Description: Companies building the core LLMs and multimodal models.
   • Examples: OpenAI (Private), Google (GOOGL), Anthropic (Private), Meta (META).
3. Layer 3: Applications & Agents (The Users)
   • Description: Software leveraging models to solve specific end-user problems.
   • Examples: Salesforce (CRM), Palantir (PLTR), Adobe (ADBE).

Company Universe

Financial Metrics (Key Performance Indicators)

• Revenue Growth: >20% for infrastructure, >30% for software applications.
• Gross Margins:
  • Hardware: >60% (NVDA standard).
  • Software: >70%.
• CapEx Intensity: High for Layer 1 & 2 (building data centers), Low for Layer 3.

Risks

• Valuation: Multiples are historically high; any growth deceleration will be punished.
• Geopolitical: US-China chip export controls could limit TAM for hardware players.
• Technological: Rapid commoditization of open-source models (e.g., Llama) could erode margins for closed-model providers.