20VC: Sequoia's David Cahn on The Winners and Losers in AI | The $0-$100M Revenue Club: Is Triple, Triple, Double, Double Dead? | The Future of Defence: Who Wins and Who Loses | How to Analyse Margins and Growth Rates in a World of AI
Timestamps are as accurate as they can be but may be slightly off. We encourage you to listen to the full context.
Podcast Summary
In this comprehensive discussion, David Cahn, Partner at Sequoia Capital, provides a deep dive into the current state of the AI bubble and what lies ahead. Cahn confirms we are indeed in an AI bubble but emphasizes that the more critical question is who will survive it. He explores the shift from abstract thinking about AI to understanding its physical reality - the massive infrastructure build-out requiring unprecedented capital and resources. (04:58)
- Main themes include the transition from dollars to gigawatts as the primary AI metric, the fragility of circular deals in the ecosystem, and the importance of being a consumer rather than producer of compute.
Speakers
David Cahn
David Cahn is a Partner at Sequoia Capital and one of the world's leading AI investors. At Sequoia, David has led investments in Clay, Juicebox, Sesame, Kela, and Stark. Before joining Sequoia, David was a General Partner at Coatue where he led investments in Notion and Hugging Face, demonstrating his early recognition of transformative AI companies before they became mainstream successes.
Key Takeaways
Physical Infrastructure is the Real AI Moat
Cahn predicted that AI's physicality would become paramount, and this has proven accurate. The ability to build data centers and secure power infrastructure has become a fundamental competitive advantage. (05:04) The market has shifted from thinking in dollars to thinking in gigawatts, with Sam Altman now discussing power requirements rather than just capital needs. Construction and infrastructure build-out capabilities will separate winners from losers, as these physical constraints cannot be easily solved through software or financial engineering alone.
Consumers of Compute Will Outperform Producers
In bubble environments, consumers of compute benefit while producers suffer from commodity pricing pressures. (17:58) When compute is overproduced, prices decrease, reducing costs of goods sold and improving gross margins for companies that consume rather than produce compute. This framework suggests investing in companies that use AI infrastructure to create intelligence and value, rather than those building the underlying infrastructure itself.
Vertical Integration Between Data Centers and Models is Essential
The coupling of data center operations with model development teams has become critical for competitive success. (12:47) OpenAI and Anthropic are increasingly becoming "steel servers and power companies," moving vertically down the supply chain. This integration allows for better optimization of the entire stack and provides more control over the fundamental resources needed for AI development.
Hire 23-Year-Olds as Your Most Important Strategic Decision
Young talent represents the most underestimated asset in AI companies today. (51:53) Since ChatGPT has only existed for five years, no one has more than five years of experience in generative AI, creating a level playing field. Young professionals who grew up with these tools possess native fluency that older, more experienced workers lack. The dynamism and learning ability of 23-24 year olds becomes more valuable than traditional experience in rapidly evolving markets.
Market Concentration Creates Systemic Risk
The concentration of 40% of the S&P 500 in seven companies creates unprecedented systemic risk. (30:56) Unlike previous bubbles that unwound through credit mechanisms, this bubble is equity-funded and will likely unwind through stock price corrections. Since Americans hold more of their net worth in equities than ever before, any AI bubble correction will be felt broadly across personal portfolios rather than contained to institutional banking systems.
Statistics & Facts
- 40% of the S&P 500 is represented by seven big tech companies, all essentially trading on AI prospects. (28:00) This represents an unprecedented level of market concentration around a single technological theme.
- A 100 gigawatts of power build-out represents approximately $8 trillion in required investment, while 250 gigawatts would require $20 trillion. (27:27) These numbers illustrate the massive scale of capital required for the AI infrastructure build-out currently being planned.
- Only 1% of global GDP represents economic profit above the cost of capital, according to a McKinsey report cited by Cahn. (32:48) This statistic challenges assumptions about the monopolistic profit margins that AI companies might achieve.
