英伟达起步：从教科书学习

Nvidia's Start: Learning from Textbooks

黄仁勋 Jensen Huang · Y Combinator · 2026-07-26 · 约 49 分钟 · 原视频 ↗

本期速览 · Overview

黄仁勋分享英伟达如何从错误技术起步，通过教科书学习重新发明 3D 图形。

Jensen Huang shares how Nvidia started with wrong technology and learned from textbooks to reinvent 3D graphics.

英伟达的成功源于在最初技术失败后从教科书中学习。
Nvidia's success came from learning from textbooks after initial technical failure.
向世嘉承认失败并保持诚实使英伟达免于破产。
Admitting failure and honesty with Sega saved Nvidia from bankruptcy.
作为通用函数逼近器的 AI 将自动化任务但创造更多就业。
AI as a universal function approximator will automate tasks but create more jobs.

Nvidia 的核心洞见是加速算法领域，而不仅仅是构建芯片。
Nvidia's core insight is accelerating algorithm domains, not just building chips.
创始人必须面对现实、保持诚实，并拥抱持续学习。
Founders must confront reality, be honest, and embrace continuous learning.
AI 是一种通用函数逼近器，将重塑整个计算栈。
AI is a universal function approximator that will reinvent the computing stack.
自动化消灭的是任务而非工作，反而带来就业增长。
Automation eliminates tasks, not jobs, leading to employment growth.
系统思维和硬科学将是始终珍贵的技能。
Systems thinking and hard sciences will remain invaluable skills.
韧性是创业者最重要的特质。
Resilience is the single most important entrepreneurial trait.

Nvidia 初始技术完全错误；成功来自从教科书学习。
Nvidia's initial technology was completely wrong; success came from learning from textbooks.
向世嘉承认失败并要求付款反而让公司存活，违背常理。
Admitting failure to Sega and asking for payment secured survival, against conventional wisdom.
AI 创造而非摧毁就业，放射学和软件工程领域即是证明。
AI creates jobs, not destroys them, as seen in radiology and software engineering.
机器人领域的 ChatGPT 时刻数年前就已到来；物理 AI 比多数人想象的更近。
The ChatGPT moment for robotics occurred years ago; physical AI is closer than many think.
简单编码将自动化，但硬科学和系统思维永不淘汰。
Simple coding will be automated, but hard science and systems thinking will never be obsolete.
CEO 应让公司适应自己，而非迎合传统管理方式。
CEOs should mold the company to themselves, not adapt to conventional management.