
Welcome to Deep Dive Panda! We've spent the week decoding the most important conversations from over 18 hours of Chinese tech, business, and AI podcasts.
The AI conversation in China has totally changed. The first wave of productivity tools? That's seen as a solved problem. They're not trying to make us more efficient anymore. They're using AI to invent new ways for people to connect and have fun.
The Unifying Theme
The first story we all bought into about generative AI was simple. It was our super-smart intern. It wrote our emails, fixed our code, and summed up our meetings. It was a tool to make work faster and cheaper, and that's a huge market. For the last couple of years, that's what everyone was building.
But in China, that's old news. The most interesting people there aren't talking about AI as a tool. They're talking about it as a medium. A tool makes an old job easier. A medium creates something brand new. The big question isn't "How can AI make us more efficient?" It's "What new human experiences can AI unlock?"
The best example is the reaction to Sora. In the West, it's all about disrupting Hollywood. In China, it's about disrupting your friend group. Founders aren't pumped to make cheaper movies. They're excited to build apps that let you and your friends star in a short video together. The goal isn't a perfect copy of reality. It's about creating a new reality you share with your friends. Who cares if the videos look a little janky? That's not the point. The point is connection. You care because your friend is in the video, not because it looks perfect.
This shows a bigger shift. As AI gets better and cheaper, the real value isn't the tech itself. The tech is becoming a commodity. The only things you can really defend are taste, community, and knowing your user. It's a move away from perfect, soulless videos and towards messy, fun, human experiences you create with people you know.
Why It Matters
For founders and investors in the West, this is a huge signal. If your AI pitch is still just "we make X 10% faster," you're already behind. The real wins are in answering a different question: "What totally new thing can people do with our product?" The companies that last won't just use AI to do old things better. They'll build things that were totally impossible before AI.
This means we need to look for opportunities in different places. The productivity gold rush is ending. The next billion-dollar companies are coming from social, gaming, and entertainment. In these markets, AI doesn't just make things more efficient—it creates entirely new kinds of fun. The game is shifting. It's no longer a battle over who has the best model. It's a battle over who best understands their users. Can you use AI to create a new way for people to connect and feel seen?
While the West obsesses over bigger models for big companies, China's builders are sprinting after a massive consumer prize. This is where the next TikTok is going to come from.
Key Patterns & Strategic Takeaways
The AI-Native Test: Here's a simple test. If you took the AI out of your product, would it just get more expensive, or would the whole thing fall apart? If it just costs more, you built a feature. If it falls apart, you've built something new.
Your Advantage Isn't the Tech: The core tech is getting commoditized. Your real advantage isn't the AI model. It's the unique data you have, the community you've built, or the way you understand your users.
Content: Product or Marketing? A big question for creators. Is your content the thing you sell? Or is it free marketing to build a huge audience for something else, like merch? You have to pick a side.
From Copycat to Trendsetter: China's tech scene isn't just copying anymore. They're creating. Smart players like TikTok are even open-sourcing their internal tools, like their custom font, to make their style the industry standard.
Anxiety is a Feature: One founder put it perfectly. That constant anxiety you feel about falling behind? It's not a bug, it's a feature. It's the fuel that keeps you moving fast in a world that's changing every day. You can't fight it, so you might as well use it.
Pivotal Quotes
"Making a video with Sam Altman... Could this have happened before? Yes... but the cost would be over 10,000 times higher. So high that it was practically impossible... Now, it can happen anytime. This is an experience that didn't exist before... This is an AI-native experience."
"The AI era shatters a core belief: that effort brings rewards. It doesn't. It doesn't. AI is omniscient. Your learning ability can't keep up with the speed of its model's evolution."
"Sora can generate a perfect video of lovers embracing, but it can never replace the feeling of a real, clumsy hug. Because technology's logic is simulation, while humanity's logic is experience."
"AI will fundamentally re-architect the entire software industry... It creates a new kind of connection, even a new relationship of production."
This week's Deep Dive Panda covers 15 specific podcast episodes, including:
Zhang Xiaojun Jùn | Business Interviews — 116. Wu Minghui's 19-year history: long ups and downs, painful turns, enterprise-level agentic models, real-world numerical games, and IPOs
Original: 张小珺Jùn|商业访谈录 — 116. 吴明辉口述19年史:漫长的沉浮、痛苦急转、企业级Agentic Model、现实世界的数值游戏、IPO
What's Next | Tech First — After Sora2, how far are we from "3D TikTok"? Chatting with the founder of VAST about the future of 3D creation | S9E32
Original: What's Next|科技早知道 — Sora2 之后,我们离「3D 抖音」还有多远?与 VAST 创始人聊聊 3D 创作的未来 | S9E32
Zhixing Bistro — E207 Dialogue with Chao Jun: Learning to make peace with yourself is a long-term practice🎾
Original: 知行小酒馆 — E207 对话朝珺:学会和自己握手言和,是一场长期练习🎾
That's how business works — Vol.227 2025 "Golden Signboard" Special: Emotional Touch
Original: 商业就是这样 — Vol.227 2025“金字招牌”特辑:情绪点金
Original: 狗熊有话说 — 515 / 我试了“1000个铁杆粉丝”模式,结果…(给创作者和创业者的启示)
Crazy Investment Circle — 124. Business Opportunities in Second-Tier Cities
Original: 疯投圈 — 124. 二线城市的商机
Original: 贝望录 — 东观西望丨4. 她势东西:高市早苗与英国铁娘子的平行线
Watching the Rise and Fall of Great Buildings — 143. How to determine if a market trend is a correction or the end? | Q3 Investment Account Review
Original: 起朱楼宴宾客 — 143.如何判断一段行情是回调还是结束?| 三季度投资账复盘
Talk about words — #266: Global Font News
Original: 字谈字畅 — #266:全球字体旧闻联播
Luo Yonghao's Crossroads — Episode 6: Tim, the Film and Television Hurricane | The Son of Yuantong #LuoYonghao'sCrossroads #Tim #LuoYonghao #PanTianhong #Film and Television Hurricane #Podcast #DeepDialogue #MakeFriends #LuoYonghaoandHisFriends
Original: 罗永浩的十字路口 — 6️⃣第六期:影视飓风Tim同学 | 圆通之子 #罗永浩的十字路口 #Tim #罗永浩 #潘天鸿 #影视飓风 #播客 #深度对话 #交个朋友 #罗永浩和他的朋友们
People's Park talks AI — Drama | Dividing Intel: Becoming a US state-owned enterprise, replicating Wintel hegemony, and creating a new "axis core" era?
Original: 人民公园说AI — 大戏|瓜分英特尔:沦为美国国企、复刻Wintel霸权,打造新”轴芯“时代?
Original: 晚点聊 — 136: Sora新世界 & Lovart 4个月复盘 | 与陈冕聊怎么做垂类Agent
AI Cocoa AI Life — [Everyone Can Understand] From Information Compression, Wisdom Forgetting to Data Synesthesia
Original: AI可可AI生活 — [人人能懂] 从信息压缩、智慧遗忘到数据通感
AI Cocoa AI Life — [AI Comment] After Sora reached the top, why do we long for a hug even more?
Original: AI可可AI生活 — [AI评论] Sora登顶之后,我们为何更渴望一个拥抱?
LinkStart — Vol.72|Chatting with Lovart’s Chen Mian and Sand.ai’s Cao Yue: Where are the new opportunities for video generation after Sora 2?
Original: 开始连接 — Vol.72|和 Lovart 陈冕、Sand.ai 曹越一起聊聊:Sora 2 之后,视频生成的新机会在哪里?
Here are the key lessons & mental models from each episode:
1. Zhang Xiaojun Jùn | Business Interviews - 116. Wu Minghui's 19-year history: long ups and downs, painful turns, enterprise-level agentic models, real-world numerical games, and IPOs
Original: 张小珺Jùn|商业访谈录 — 116. 吴明辉口述19年史:漫长的沉浮、痛苦急转、企业级Agentic Model、现实世界的数值游戏、IPO
What You Need to Know:
Your Data Is Your Moat. AI models are starting to feel like a basic utility, like electricity. What really sets you apart is the unique data you feed them. The winning companies won't be the ones with the fanciest model. They'll be the ones who use their own special industry data to build an AI that just works better for their customers. The real fight isn't over the AI model, it's over the data you put in it.
Agents Are the Real Breakthrough. The big deal for AI in business isn't another chatbot. It's AI agents that can actually use your old, clunky software just like a person would, by clicking around on the screen. This totally sidesteps the nightmare of trying to get different systems to talk to each other. It’s a whole new way for all your software to finally work together.
There's a Limit to Bigger AI. This idea that we can just keep making AI models bigger and bigger forever is probably wrong. Just like buildings can only get so tall, AI models have limits. The future isn't one giant, all-knowing AI. Instead, we'll see a bunch of smaller, specialized AIs that team up to solve problems. This means there's room for lots of different companies to win, not just a few big ones.
Surviving a Crisis Changes You. Nothing transforms a founder like a near-death experience for their company. Running on fumes with only a month of cash left forces you to stop dreaming and start executing. When you have almost nothing, you have to get focused and efficient, fast. It’s a brutal way to learn, but it builds the grit you need to last.
AI's "Black Box" Is a Huge Plus in China. In China's B2B software world, clients often demand the source code, which basically gives away your secret sauce. But AI models are a game-changer. You can give a client the model, but the real magic—the unique data and training that made it smart—stays with you. It’s a "black box," and it makes AI a much more defensible business over there.
"AI will fundamentally re-architect the entire software industry, enterprise services, and even the entire industrial internet. It creates a new kind of connection, even a new relationship of production."
Subscribe to Deep Dive Panda to unlock the rest.
Become a paying subscriber of Deep Dive Panda to get access to this post and other subscriber-only content.
Upgrade to paid

