BUILD AGENTS

MAR 23, 2026Lex Fridman

#494 – Jensen Huang: NVIDIA – The $4 Trillion Company & the AI Revolution

•
Optimize the Stack - NVIDIA has shifted from being a chip designer to a systems company, utilizing 'extreme co-design' to treat the entire data center as a single, integrated computer.
“The computer of the future is the data center, and the data center is the computer.”
— Jensen Huang
•
Flatten the Organization - Jensen maintains 60 direct reports and avoids 1-on-1s to ensure information transparency and rapid execution, believing that hierarchy is the enemy of agility.
•
Master Natural Language - The future of programming is no longer C++ or Java; natural language is becoming the primary interface, effectively turning every human into a potential developer via AI.
“The computer of the future is the data center, and the data center is the computer.”
— Jensen Huang

MAR 23, 2026Lex Fridman

•
Optimize the Stack - NVIDIA has shifted from being a chip designer to a systems company, utilizing 'extreme co-design' to treat the entire data center as a single, integrated computer.
“The computer of the future is the data center, and the data center is the computer.”
— Jensen Huang
•
Flatten the Organization - Jensen maintains 60 direct reports and avoids 1-on-1s to ensure information transparency and rapid execution, believing that hierarchy is the enemy of agility.
•
Master Natural Language - The future of programming is no longer C++ or Java; natural language is becoming the primary interface, effectively turning every human into a potential developer via AI.
“The computer of the future is the data center, and the data center is the computer.”
— Jensen Huang

MAR 20, 2026Conviction

•
The Rise of AutoResearch - AI is transitioning from passive assistants to closed-loop agents capable of designing, executing, and optimizing their own experiments without human intervention.
“We’re moving into the 'loopy' era where agents aren't just helping you write code, they are closing the loop on the entire scientific process of discovery.”
— Andrej Karpathy
•
Software Engineering as Orchestration - Coding is shifting from manual syntax writing to high-level system architecture, where natural language becomes the primary interface for building complex software.
•
The Loopy Era of AI - We are entering a phase where models improve through autonomous feedback loops and self-generated data, moving beyond the constraints of static human datasets.
“We’re moving into the 'loopy' era where agents aren't just helping you write code, they are closing the loop on the entire scientific process of discovery.”
— Andrej Karpathy

MAR 20, 2026Conviction

•
The Rise of AutoResearch - AI is transitioning from passive assistants to closed-loop agents capable of designing, executing, and optimizing their own experiments without human intervention.
“We’re moving into the 'loopy' era where agents aren't just helping you write code, they are closing the loop on the entire scientific process of discovery.”
— Andrej Karpathy
•
Software Engineering as Orchestration - Coding is shifting from manual syntax writing to high-level system architecture, where natural language becomes the primary interface for building complex software.
•
The Loopy Era of AI - We are entering a phase where models improve through autonomous feedback loops and self-generated data, moving beyond the constraints of static human datasets.
“We’re moving into the 'loopy' era where agents aren't just helping you write code, they are closing the loop on the entire scientific process of discovery.”
— Andrej Karpathy

MAR 19, 2026Bankless

•
AI agents require native financial rails - Existing payment systems are built for human interaction, but agentic commerce necessitates high-frequency, low-latency machine-to-machine transactions that traditional banks cannot support.
“The original sin of the internet was the lack of a native payment layer.”
— Georgios Konstantopoulos
•
MPP solves the internet’s 'original sin' - The Machine Payment Protocol provides a standardized layer for agents to negotiate and settle payments autonomously, effectively adding the native payment layer the early internet lacked.
•
Performance dictates the sovereign L1 design - Tempo launched as a standalone L1 rather than an Ethereum L2 to optimize specifically for the performance, latency, and security requirements of automated agent stacks.
“The original sin of the internet was the lack of a native payment layer.”
— Georgios Konstantopoulos