Episode

Jerry Chen | The New New Moats

Podcast: Greymatter
Published: Jun 22, 2023
Duration seconds: 1217
Processing state: processed
Canonical source: https://pdst.fm/e/traffic.megaphone.fm/GRL5791819795.mp3?updated=1687454568
Audio: https://pdst.fm/e/traffic.megaphone.fm/GRL5791819795.mp3?updated=1687454568
JSON: /v1/public/podcasts/greymatter/episodes/jerry-chen-the-new-new-moats
Markdown: /podcast/greymatter/jerry-chen-the-new-new-moats.md

Actions

POST https://stenobird.com/v1/public/podcasts/greymatter/episodes/jerry-chen-the-new-new-moats/transcription-requests
Idempotently request low-priority transcript generation for this episode.
GET https://stenobird.com/podcast/greymatter/jerry-chen-the-new-new-moats.md
Read the agent-friendly Markdown representation of this episode resource.

Summary

Jerry Chen revisits his 2017 thesis on 'Systems of Intelligence' to evaluate how generative AI has changed the landscape of defensible business models. He explores whether startups can still build moats against big cloud providers using the new AI application stack.

Topics

Artificial Intelligence
Venture Capital
SaaS
Large Language Models
Cloud Computing
Business Strategy
Open Source
Software Infrastructure

Highlights

Main idea: The new AI stack is emerging as a bridge between foundation models and private enterprise data via frameworks like LlamaIndex
Failure mode: Startups face high risk because underlying technologies like retrieval and memory are evolving too rapidly for long-term stability
Practical takeaway: As AI becomes a ubiquitous ingredient like mobile, defensibility shifts back to classic moats like network effects, switching costs, and workflow integration
Main idea: Open source models like Llama provide a viable way for startups to challenge big cloud incumbents by building specialized systems of intelligence
Practical takeaway: Success in the AI era requires agility and a focus on building deep integration and trust rather than just layering an API call onto a model

Chapters

1:00 Revisiting the 2017 Thesis: The motivation behind analyzing how startups can build defensive value against big cloud providers.
2:30 The Dual Nature of AI Usage: Distinguishing between user-facing chat interfaces and the underlying LLM integration in applications.
4:00 The Emerging AI Application Stack: How new frameworks are connecting foundation models to enterprise systems of record and engagement.
5:20 Systems of Record vs. Engagement: The evolution of databases and SaaS platforms into the new AI-driven infrastructure.
6:50 Risks in the Rapidly Evolving Stack: The dangers of building on top of technologies that are changing too quickly to predict.
9:50 The Future of Vector Databases: Analyzing whether vector embeddings will create a new category or simply enhance existing databases.
12:40 Open Source as a Competitive Weapon: How open source models allow startups to attack big cloud providers through better distribution and specialization.