Research that compounds
Deep insights on AI and frontier technology building what's next
Explore All Research on Substack
Deep Dive
August 31, 2026
Building Robot Bodies: Not a Supply Chain. A Learning Loop.
What US Robotics Teams Can Learn from China
Etna Portfolio
August 26, 2026
Our Investment in Agentrys: A Self-Improving Layer for Chip Design
We believe chip design is one of the first high-value domains where recursive self-improvement (RSI) can become practical.
Deep Dive
August 20, 2026
Data Is a Great Place to Start an AI Company, and a Dangerous Place to Stop
Inside the global suppliers capturing frontier labs’ billions, the shift from one-off datasets to trainable workflows, and the first in-depth look at China’s research-native data vendor ecosystem.
Deep Dive
August 17, 2026
Claude Tag Is Underrated
Anthropic’s Bet on the Third Era of AI Products
Etna Portfolio
August 4, 2026
Our Investment in Convex: The Backend Where Agents Build
As AI agents become software’s primary builders, we believe Convex’s unified, reactive, and correct-by-default backend will become the application runtime for the agent era.
Explore All Research on Substack
Research that compounds
Deep insights on AI and frontier technology building what's next
Explore All Research on Substack
Deep Dive
August 31, 2026
Building Robot Bodies: Not a Supply Chain. A Learning Loop.
What US Robotics Teams Can Learn from China
Etna Portfolio
August 26, 2026
Our Investment in Agentrys: A Self-Improving Layer for Chip Design
We believe chip design is one of the first high-value domains where recursive self-improvement (RSI) can become practical.
Deep Dive
August 20, 2026
Data Is a Great Place to Start an AI Company, and a Dangerous Place to Stop
Inside the global suppliers capturing frontier labs’ billions, the shift from one-off datasets to trainable workflows, and the first in-depth look at China’s research-native data vendor ecosystem.
Deep Dive
August 17, 2026
Claude Tag Is Underrated
Anthropic’s Bet on the Third Era of AI Products
Etna Portfolio
August 4, 2026
Our Investment in Convex: The Backend Where Agents Build
As AI agents become software’s primary builders, we believe Convex’s unified, reactive, and correct-by-default backend will become the application runtime for the agent era.
Explore All Research on Substack
Research that compounds
Deep insights on AI and frontier technology building what's next
Explore All Research on Substack
Deep Dive
August 31, 2026
Building Robot Bodies: Not a Supply Chain. A Learning Loop.
What US Robotics Teams Can Learn from China
Etna Portfolio
August 26, 2026
Our Investment in Agentrys: A Self-Improving Layer for Chip Design
We believe chip design is one of the first high-value domains where recursive self-improvement (RSI) can become practical.
Deep Dive
August 20, 2026
Data Is a Great Place to Start an AI Company, and a Dangerous Place to Stop
Inside the global suppliers capturing frontier labs’ billions, the shift from one-off datasets to trainable workflows, and the first in-depth look at China’s research-native data vendor ecosystem.
Deep Dive
August 17, 2026
Claude Tag Is Underrated
Anthropic’s Bet on the Third Era of AI Products
Etna Portfolio
August 4, 2026
Our Investment in Convex: The Backend Where Agents Build
As AI agents become software’s primary builders, we believe Convex’s unified, reactive, and correct-by-default backend will become the application runtime for the agent era.
Etna Chats
June 25, 2026
Fable 5: Expensive Intelligence Needs Expensive Work
Frontier models may capture 80% of model revenue, while lower-cost and open-source models may process 80% of tokens.
Explore All Research on Substack