Invest Like the Best with Patrick O'Shaughnessy
Gavin Baker - Watts and Wafers - [Invest Like the Best, EP.473]
· Summarized
Key takeaways
- Physical constraints such as Watts (power supply) and wafers (semiconductor capacity) will heavily influence the future of AI scaling and innovation.
- Orbital compute provides a long-term solution to power shortages and scale limitations, transforming the infrastructure landscape.
- TSMC’s capacity management and industry supply discipline are critical to avoiding bubbles and ensuring sustainable growth in chip manufacturing.
- Most economic returns in AI are still captured at the frontier models, with ongoing debates about valuation, efficiency, and the future role of frontier tokens.
- Innovations in chip architecture and disaggregation (pre-fill vs. decode) are vital, with startup successes often coming from doing the hard, fundamentally different things.
- Major tech companies like Google, Meta, Amazon, and Microsoft continue to shape AI’s ecosystem with strategic decisions, alliances, and internal initiatives.
- Societal risks such as political violence and societal disruption are real, and thoughtful, safety-focused AI development and governance are necessary for long-term benefit.
Summary
Introduction and Podcast Context
The episode opens with Patrick O'Shaughnessy welcoming Gavin Baker for their sixth in-depth discussion. The central theme revolves around 'Watts and Wafers,' the two critical physical constraints that will shape the next phase of AI development. Gavin discusses the near-term power shortages expected to ease by 2027-28 with new energy sources and long-term solutions like orbital compute. On wafers, he compares current timing to previous bubbles, emphasizing TSMC's capacity decisions as a key variable. Broader industry dynamics, Elon Musk’s Terrafab initiative, GPU disaggregation, and the value flow from frontier models are also explored.
The Power and Wafers Constraints in AI Progress
Gavin Baker elaborates on the physical constraints of Watts (power supply) and wafers (semiconductor manufacturing) that will govern future AI advancements. Power shortages are seen as manageable by 2027-28, with orbital compute offering a long-term solution. Wafers, particularly TSMC's capacity decisions, are highlighted as a critical and potentially decisive factor in avoiding or incurring bubbles. He discusses how these constraints influence the scaling and economics of AI models.
Industry Strategies and Technological Disruption
The discussion covers Elon Musk's Terrafab project, disaggregation of GPUs, and the emergence of new chip startups. Gavin emphasizes that doing something fundamentally different and hard—like Cerebrus' innovative chip architecture—is vital for startup success. He explores the ongoing race in chip design, the significance of capacity expansion, and the risks of bubbles driven by supply-demand imbalances. The conversation underscores the importance of competition, supply discipline, and technical innovation.
Valuation, Market Dynamics, and Frontier Models' Economics
Gavin analyzes how AI company's valuations, especially OpenAI and Anthropic, compare with other tech firms like Snowflake and Databricks. He notes differences in capital efficiency, burn rates, and compute secured by each. The trend toward usage-based pricing and its implications for ARR growth are discussed, alongside the potential for continued economic value creation primarily at the frontier model layer. He highlights that most returns still come from frontier tokens, raising questions about their future role.
Technical Frontiers: Continual Learning, Chips, and Open Source AI
The podcast examines the evolution of AI through continual learning, memory, and the disaggregation of inference and pre-fill functions. Gavin stresses that the technological advances in memory, harness engineering, and chip design are critical, with innovative startups making significant contributions. He discusses the strategic importance of new chip architectures, the potential for open-source models in China, and the game-theoretic considerations of model release and distillation. The safety, safety risks, and societal implications of AI are also touched upon.
Major Tech Companies’ Positions and Future Outlooks
Gavin reviews the strategic positions of Google, Meta, Amazon, and Microsoft in AI. Google, previously dominant due to TPU advantage, still holds unmatched compute resources and data. Meta has made notable strides with AI-first initiatives. Amazon benefits from internal models and robotics, while Microsoft’s ongoing investments and risk-taking are highlighted as pivotal. The importance of these giants’ decisions in AI development and ecosystem engagement is emphasized, along with predictions about their future roles.
Broader Implications and Final Reflection
The conversation concludes with reflections on the broader impact of AI—its effect on value creation, geopolitical stability, and societal safety. Gavin emphasizes the potential for AI to improve human lives dramatically, including healthcare, but also warns of the increased risks, such as political violence and societal disruption. He advocates for a balanced, humble approach to AI development, emphasizing safety, trust, and societal benefit, citing historical examples of American leadership and global stability.
Sponsors
- Ramp · Ramp is the only platform built to make your finance team leaner, faster, and better, saving businesses 5% annually on average so you can stay focused on growth. Ramp customers grew revenue 3.2 times faster than the average American business. Visa, Fersel, Cursor, Stripe, Notion, 11-Mibes, Shopify, and 70,000 other businesses all run on Ramp. Learn more at ramp.com slash invest.
- Vanta · Vanta automates security and compliance for over 16,000 fast-moving companies, helping them prepare for audits and manage vendor risk around the clock. It acts as a 24-7 GRC engineer, finding issues, drafting fixes, and reducing assessment times by up to 50%. Special offer for Invest Like the Best listeners: $1,000 off at vanta.com slash invest.
- Ridgeline · Ridgeline has built a complete, real-time operating system for investment managers, integrating trading, portfolio management, compliance, and reporting in one modern cloud platform. Learn more at ridgelineapps.com.
- Rogo · Rogo is the AI platform for finance, building agents trained to understand how bankers and investors work, from diligence and modeling to analysis and deliverables. Visit rogo.ai/invest to learn more.
Show notes
My guest today is Gavin Baker, founding partner and CIO of Atreides Management, and this is our sixth conversation.
The central theme is watts and wafers, the two physical constraints that in Gavin's view will dictate the next phase of AI.
On power, he thinks the near-term shortage starts to ease in 2027 and 2028 as new sources of energy come online, and that orbital compute solves it in the long term.
On wafers, he explains what is different this time from the dotcom bubble and why TSMC’s capacity decisions may be the single most important variable to watch.
We also discuss Elon’s Terrafab, the disaggregation of GPUs, the role of new chip companies, and whether the economic value of AI will keep accruing to frontier models.
For the full show notes, transcript, and links to mentioned content, check out the episode page here.
-----
Become a Colossus member to get our quarterly print magazine and private audio experience, including exclusive profiles and early access to select episodes. Subscribe at colossus.com/subscribe.
-----
Ramp’s mission is to help companies manage their spend in a way that reduces expenses and frees up time for teams to work on more valuable projects. Go to ramp.com/invest to sign up for free and get a $250 welcome bonus.
-----
Trusted by thousands of businesses, Vanta continuously monitors your security posture and streamlines audits so you can win enterprise deals and build customer trust without the traditional overhead. Invest Like the Best listeners get a special offer of $1,000 off Vanta when you go to vanta.com/invest.
-----
WorkOS is the infrastructure B2B and AI-native companies use to sell to enterprise. It covers everything enterprise security requires: SSO, SCIM, RBAC, Audit Logs, AI governance, and more. Trusted by 2,000+ fast-growing companies, including OpenAI, Anthropic, Cursor, and Vercel.
-----
Rogo is the AI platform for finance. They're building agents for Wall Street that are trained to understand how bankers and investors actually do work: from diligence and modeling, to turning analysis into deliverables. To learn more, visit rogo.ai/invest.
-----
Ridgeline has built a complete, real-time, modern operating system for investment managers. It handles trading, portfolio management, compliance, customer reporting, and much more through an all-in-one real-time cloud platform. Visit ridgelineapps.com.
-----
Editing and post-production work for this episode was provided by The Podcast Consultant (https://thepodcastconsultant.com).
Timestamps:
(00:00:00) Welcome to Invest Like The Best
(00:02:29) Gavin Baker Intro
(00:03:32) Anthropic's Record ARR Growth
(00:11:49) Should OpenAI and Anthropic Raise at a Much Higher Valuation?
(00:13:23) How Elon Preserves Investor Trust
(00:14:00) Watts & Wafers
(00:15:45) Data Centers in Space Explained
(00:20:51) Orbital Compute’s Impact on Terrestrial Data Centers
(00:26:24) TSMC Supply Discipline & Bubble Risk
(00:30:50) Demand for Frontier Tokens & The Bitter Lesson
(00:35:33) Continual Learning & Memory
(00:40:01) New Chip Companies & Startups
(00:42:49) Prefill vs. Decode Disaggregation
(00:48:40) AI-Native Founders: Different & Hard
(00:51:27) Token Path & Application Layer
(00:56:13) How Gavin Uses AI in Atreides
(01:00:06) Signs of a Diversity Breakdown
(01:05:42) Google, Meta, Amazon, Microsoft
(01:11:42) Broader Knock-On Effects of AI
Follow this show on AudioBrevity and new episodes get summarized for you. Create a free account