143. 对何小鹏的第二次访谈：更大赌注、人形机器人Iron诞生、那场意外、技术剧变下CEO、GX和缝合怪
Listen
问这期播客
会先在本集摘要、章节、转录和笔记里找答案。
TL;DR · AI Summary
何小鹏在访谈中透露了小鹏汽车在2025年的重大赌注，包括人形机器人Iron的开发，并表达了对技术变革下CEO角色的思考。
Key Takeaways
- 何小鹏认为造车和开发人形机器人都是高风险高回报的挑战。
- 小鹏汽车计划在2025年投入大量资源开发人形机器人Iron。
- 何小鹏强调CEO需要保持理性，专注于战略方向而非陷入细节。
Outline
Jump quickly between sections.
Mindmap
See how the topics connect at a glance.
查看大纲文本（无障碍 / 无 JS 友好）
- 专注战略方向
Highlights
Key sentences worth saving and sharing.
造车就像在‘在血海里游泳’，而造人形机器人的胜率更低。
我们有两成胜率，但这是值得去尝试的事情。
CEO需要保持理性，专注于战略方向而非陷入细节。
Chapters
CEO使用AI小调研
CEO使用AI小调研
不能做缝合怪
不能做缝合怪
2025年下的赌注
2025年下的赌注
IRON机器人的诞生故事
IRON机器人的诞生故事
去年那场意外
去年那场意外
我们有两成胜率
我们有两成胜率
新车GX
新车GX
没有一家游出血海
没有一家游出血海
深入细节并不等于陷入细节。深入细节是用细节校正模型，提升认知模型的精度，比如马斯克、黄仁勋都是如此。至少在对AI coding的判断上，小鹏这句‘对初级程序员的
深入细节并不等于陷入细节。深入细节是用细节校正模型，提升认知模型的精度，比如马斯克、黄仁勋都是如此。至少在对AI coding的判断上，小鹏这句‘对初级程序员的辅助”，已经是上个 时代的理解，说明他的认知没有被AI coding的真实能力和边界校准过。
明白何小鹏干嘛不用现在所谓的ai，因为去年又大改了一次自动驾驶，觉得现在ai其实都很初级，ceo只需要有所涉及了解就行，让手下去试错就行，时间对于ceo太宝贵了
明白何小鹏干嘛不用现在所谓的ai，因为去年又大改了一次自动驾驶，觉得现在ai其实都很初级，ceo只需要有所涉及了解就行，让手下去试错就行，时间对于ceo太宝贵了
不提token消耗吗？微软都烧不起了😂
不提token消耗吗？微软都烧不起了😂
Transcript
CEO使用AI小调研
不能做缝合怪
2025年下的赌注
IRON机器人的诞生故事
去年那场意外
我们有两成胜率
新车GX
没有一家游出血海
深入细节并不等于陷入细节。深入细节是用细节校正模型，提升认知模型的精度，比如马斯克、黄仁勋都是如此。至少在对AI coding的判断上，小鹏这句‘对初级程序员的辅助”，已经是上个 时代的理解，说明他的认知没有被AI coding的真实能力和边界校准过。
明白何小鹏干嘛不用现在所谓的ai，因为去年又大改了一次自动驾驶，觉得现在ai其实都很初级，ceo只需要有所涉及了解就行，让手下去试错就行，时间对于ceo太宝贵了
不提token消耗吗？微软都烧不起了😂
Show notes
143. Second Interview with He Xiaopeng: Bigger Bets, Birth of Humanoid Robot Iron, That Incident, CEO in the Age of Technological Revolution, GX, and Frankenstein's Monster
143. Second Interview with He Xiaopeng: Bigger Bets, Birth of Humanoid Robot Iron, That Incident, CEO in the Age of Technological Revolution, GX, and Frankenstein's Monster - Zhang Xiaojun | Business Interviews | Xiao Yuzhou FM - Listen to Podcasts on Xiao Yuzhou FM
Morning Routine: Morning Hygiene and Grooming
Listen to podcasts on Xiao Yuzhou FM! Click to download
143. Second Interview with He Xiaopeng: Bigger Bets, Birth of Humanoid Robot Iron, That Incident, CEO in the Age of Technological Revolution, GX, and Frankenstein's Monster
Zhang Xiaojun | Business Interviews
88 minutes·2 hours ago
2116
·
7
This episode features a return interview with He Xiaopeng, Chairman and CEO of Xpeng Motors.
In our previous interview (Episode 70), He Xiaopeng described car manufacturing as "swimming in a sea of blood."
Over the past year, in addition to this bloody battle in the automotive industry, Xpeng Motors has also faced the future battle of AI and robotics—where the battlefield is longer, and the odds of success for creating humanoid robots are much lower than for car manufacturing.
For this reason, He Xiaopeng placed a bigger bet in 2025.
We start our conversation with this "bet," and he also discusses the birth of the Iron humanoid robot, the mindset of a CEO in the age of technological change, the new GX model, and more. However, there are some concerns and secrets he cannot yet reveal about what is happening.
Next, here is my interview with He Xiaopeng.
OUTLINE:
00:02:00 CEO using AI for small surveys
00:11:24 Avoid being a Frankenstein's monster
00:18:23 The bet placed in 2025
00:41:15 The story behind the birth of the IRON robot
00:49:51 The incident last year
00:57:43 We have a 20% chance of success
01:09:20 The new GX model
01:17:17 No one has bled out in the sea
LINKS:
Our podcast is available on Xiao Yuzhou FM, Apple Podcast, Spotify, and other audio platforms;
Our video podcast is available on Bilibili, Xiaohongshu, Douyin, Weibo Video, and other video platforms;
If you prefer the text version, search for our studio's WeChat official account: Language is World (语言即世界).
DETAILS: This episode is jointly presented by Language is World Studio, Weibo, and Sina Finance.
DISCLAIMER: This content is not intended as investment advice.
CONTACT: xiaojunzhang@lisw.ai
Join us in exploring the new world! 😉
Expand Show Notes
2 hours ago
He Xiaopeng is really strong!!!
2 hours ago
First, like then watch
11 minutes ago
05:32 Delving into details does not mean getting stuck in them. Delving into details is about using details to calibrate models and improve the precision of cognitive models, such as Musk and Huang Renxun. At least in terms of judging AI coding, Xpeng's statement of "assisting junior programmers" is already an outdated understanding, indicating that their cognition has not been calibrated by the true capabilities and boundaries of AI coding.
1 hour ago
18:43 Understand why He Xiaopeng doesn't use current AI, because they made a major overhaul of autonomous driving last year and think that current AI is still quite primitive. A CEO only needs to be somewhat involved and let their team make mistakes; time is too valuable for a CEO.
2 hours ago
Personal feeling: Comparing interviews with Li Xiang and He Xiaopeng, I prefer Li Xiang's style of expression. Li Xiang's expression is more passionate and inspiring. He Xiaopeng is more rational and steady. Both are brilliant entrepreneurs in contemporary China.
2 hours ago
Frankenstein's monster! Can't wait! Zhang Xiaojun, He Xiaopeng.
1 hour ago
05:14 Don't mention token consumption? Microsoft can't afford it 😂
00:00
00:00