Invest Like the Best with Patrick O'Shaughnessy

Gavin Baker - Watts and Wafers - [Invest Like the Best, EP.473]

Invest Like the Best with Patrick O'Shaughnessy·May 21, 2026

INVESTMENT OPPORTUNITIES

  • AI-related assets/companies: Described as being at "really attractive valuations" in March/April. Anthropic is highlighted for adding $11 billion ARR in one month, growing at "ridiculous rates," with 500% NDR, and burning 80% less capital than competitors. Hypothetically, if unconstrained by compute, Anthropic could achieve $100-200 billion in revenue, making its current valuation ~5x unconstrained revenue. This is presented as a strong buying opportunity.
  • Astera Labs: Mentioned as a miscategorized company in the copper loser baskets. Its biggest product, a switch, connects copper and optics to accelerators. The argument is that it's on the "other side of that connection," implying it's not a "copper loser" and has a unique, hard-to-replicate position. It's suggested to be a "good bet" for venture.

OVERVIEW

This episode provides a deep dive into the current state and future of AI, focusing on the two physical constraints: watts (energy) and wafers (chips). Host Patrick O'Shaughnessy and guest Gavin Baker discuss the unprecedented growth of AI companies like Anthropic, the geopolitical implications of AI infrastructure, and the potential for new market dynamics driven by technological breakthroughs and changing economic incentives.

KEY TOPICS

  • Current AI market dynamics and unprecedented growth of frontier models
  • Comparison of current market conditions to historical economic bubbles
  • The "Watts" constraint: energy supply for AI and its future resolution
  • The "Wafers" constraint: chip manufacturing, TSMC's role, and new entrants
  • Orbital compute as a long-term solution for AI infrastructure
  • Elon Musk's companies (SpaceX, Tesla) and their impact on AI and manufacturing
  • Geopolitical implications of energy independence and AI leadership
  • The "Bitter Lesson" and value accrual to frontier models vs. open-source
  • The shift to usage-based pricing in AI and its bullish implications
  • The strategy for new chip companies: "different and hard"
  • AI's impact on personal safety and cybersecurity
  • The role of AI in geopolitics and warfare, specifically in Ukraine
  • Historical parallels for economic bubbles and technological revolutions
  • Analysis of major tech companies (Google, Meta, Amazon, Microsoft) and their AI strategies
  • AI's impact on the application layer, including value destruction and creation
  • The "token path" concept for AI companies
  • The investment landscape for AI startups and the challenges they face

MAIN TAKEAWAYS

  • The current AI market, particularly the growth of companies like Anthropic, is unprecedented and offers significant investment opportunities at historically attractive valuations, despite market drawdowns.
  • The "Watts" (energy) constraint for AI compute is expected to ease by 2027-2028, with orbital compute offering a long-term solution, while the "Wafers" (chips) constraint, controlled by a small group in Taiwan (TSMC), is the more immediate and unique bottleneck preventing an AI bubble.
  • Elon Musk's approach of consistently delivering returns and not being "greedy on valuation" has given him a unique capital-raising "superpower," contrasting with the more cautious fundraising strategies of AI companies facing uncertainty.
  • The unexpected phenomenon of "frontier models" capturing the overwhelming majority of economic value in AI, rather than open-source or commoditized models, is a critical factor for investors to understand.
  • AI is fundamentally changing market pricing models (shifting from "all you can eat" to "pay by the drink") and extending the useful life of hardware (like GPUs to 10-15 years), which has significant, positive implications for the private credit industry.

NOTABLE QUOTES

"All you had to do in March was simply observe what was happening to Anthropic."
"You get a chance to buy AI at really attractive valuation."
"If I were to watch one thing to understand whether there's a bubble, it's Taiwan Semi's capacity decisions."
"My single most useful agent is a really good summary of the points that would be interesting to me from podcasts."

Summarized with DriftNote — AI-powered podcast summaries

Try it free