Apr 16, 2025 · 1h 2m · big-technology
Why China's AI Strategy Just Might Work — With Grace Shao
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of the Big Technology Podcast, host Alex Kantrowitz interviews tech analyst Grace Shao to analyze China's burgeoning AI ecosystem following DeepSeek's breakthrough. They examine how China's mission-driven founders, open-source monetization, deep manufacturing supply chains, and rapid consumer adoption position the nation as a formidable global contender despite ongoing semiconductor restrictions.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Alex holds 30.9% of the talking time here. How this is scored →
Alex as informed peer 5.2 Guest teaching 6.7 Guest disagreement 1.6 Alex pushing back 2.5
speaking balance: gold is Alex, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 5:49 Rejecting the Western Geopolitical Narrative on Open Source
Grace explicitly pushes back against the framing that China open-sources AI to disrupt US tech labs, characterizing it as a politically charged narrative detached from domestic business realities.
Hardest push from Alex ▶ 15:52 Pressing on China's True AI ParityAlex refuses to let his core question go, directly challenging Grace to substantiate how China can be globally competitive when foundational model breakthroughs remain concentrated in American frontier research labs.
Biggest teaching moment ▶ 6:19 Educating on Structural Barriers to Chinese Enterprise SaaSGrace provides a comprehensive structural breakdown of China's labor-driven economy and weak IP history to explain why Chinese AI firms are compelled to monetize via consumer apps rather than enterprise software.
Alex holds their own ▶ 10:42 Connecting Meta Strategy to Chinese Copycat DynamicsAlex demonstrates commanding industry knowledge by weaving together Kai-Fu Lee's insights on Chinese tech cloning with proprietary data from an interview with Yann LeCun regarding Meta's consumer AI assistant.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Alex as informed peer | Guest teaching | Guest disagreement | Alex pushing back | Why |
|---|---|---|---|---|---|---|
| DeepSeek's Significance and the New Generation of Mission-Driven Founders | 4 | 6 | 3 | 5 | Alex questions the claim that Chinese AI founders are mission-driven rather than profit-driven, noting DeepSeek was founded by a quantitative hedge fund. Grace reframes Liang Wenfeng's background, explaining how humble roots and patriotic motivations differentiate this generation from earlier internet entrepreneurs. | |
| Drivers of China's Open-Source Models and Consumer AI Monetization | 6 | 7 | 2 | 3 | Alex presents the US perspective that China open-sources AI models to undercut American research lab moats. Grace counters this geopolitical interpretation by detailing China's structural lack of enterprise SaaS adoption and historical reliance on consumer super-apps. | |
| Consumer Distribution, WeChat's Integration, and Meta Strategy Parallels | 7 | 6 | 1 | 1 | Alex draws sophisticated parallels between Meta's open-source strategy and Chinese platforms, citing discussions with Kai-Fu Lee and Yann LeCun. Grace validates the comparison and details Tencent's strategic choice to integrate DeepSeek into WeChat's search ecosystem. | |
| AI Productization, Physical AI, and the 'Good Enough' Model Era | 5 | 7 | 3 | 6 | Alex presses Grace on why China should be considered a global AI contender beyond DeepSeek given the concentration of frontier labs in the US. Grace explains that as models reach a 'good enough' threshold, China's hardware supply chain and mechatronics expertise give it an edge in physical robotics. | |
| Rapid Domestic AI Adoption and Cultural Receptivity to Tech | 6 | 6 | 1 | 2 | Alex cites reporting detailing rapid DeepSeek adoption across Chinese automakers, smartphones, appliances, and hospitals. Grace adds cultural and institutional context, explaining government incentives, lower privacy friction, and broad public tech enthusiasm. | |
| Entrepreneurial Evolution, Regulatory Crackdowns, and Market Rebound | 4 | 7 | 1 | 2 | Grace draws on her experience working at Alibaba during the tech crackdown to explain the shift from copycat models to genuine innovation. Alex listens as Grace outlines how AI resurgence and public tech stocks have revived investor confidence. | |
| Two Sessions AI Policy Push and the 'Top Scholar' Phenomenon | 6 | 7 | 1 | 1 | Alex cites Bloomberg coverage of the National People's Congress and New York Times reporting on DeepSeek's founder being celebrated as a 'Top Scholar'. Grace elaborates on the imperial Zhuangyuan tradition and the cultural reverence for academic meritocracy in China. | |
| Mid-Episode Transition and Part Two Topic Roadmap | 5 | 7 | 1 | 1 | Alex asks about the viral capabilities of Chinese humanoid robotics compared to Western prototypes. Grace breaks down China's manufacturing edge in mechatronics, quoting Jensen Huang and citing Unitree's commanding global market share. | |
| Hangzhou's Rise as an AI Powerhouse and the 'Six Dragons' | 4 | 8 | 1 | 1 | Alex prompts Grace to explain the rise of Hangzhou as an AI center beyond Shenzhen. Grace details Hangzhou's strategic location, academic talent pipeline from Zhejiang University, and the cluster of companies known as the Six Dragons. | |
| The Manus AI Agent and China's GPU Compute Bottleneck | 5 | 6 | 2 | 3 | Alex inquires about the viral Manus agent and whether US chip export controls create an insurmountable bottleneck for China. Grace demystifies Manus as an application wrapper rather than a fundamental breakthrough and agrees that compute shortages remain China's primary vulnerability. |
Statements from this episode (21)
Shao: DeepSeek CEO Liang Wenfeng and his team were educated in China
“Liang Wenfeng, the CEO of DeepSeek and his whole team, they're actually educated in China.”
Shao: Chinese AI founders are driven by recognition and vision over money
“Chinese entrepreneurs now no longer are just driven by money and making money. They're driven by recognition, influence, and power. And a lot of them are like visionary start off with and not businessmen to start off with.”
Shao: US GDP is 60% knowledge-worker driven unlike China's labor-based economy
“China's still not a very heavily depend is not heavily dependent on knowledge workers, Compared to the U S where it's like, 60% of GDP is driven by knowledge workers.”
Shao: Enterprise software is a minor business for Alibaba and ByteDance
“Even the companies that have enterprise softwares like Alibaba or ByteDance, those products are their minor businesses. So it's the whole internet economy is so dependent and reliant on consumer facing applications.”
Shao: Chinese AI monetization strategy is concentrated entirely on applications
“So I think really the monetization strategy is all an application and they're hoping that one day, whether it's your plugging ads or whatever, you know, charging at friction points, that's where it's going to make the money.”
Kantrowitz: Meta AI has 600M users compared to OpenAI's 400M
“And I was just speaking with Jan LeCun, the chief AI scientist at Meta, and he said there's Six hundred million people using their AI assistant within their apps compared to four hundred million of open AI's chat GPT.”
Shao: Tencent was first to integrate DeepSeek into WeChat
“What's interesting is I do want to point out is Tencent actually has not really been leading in their LLM development at all. In fact, they're the only and first one that have actually integrated Deep Seek into their products into their WeChat, which obviously…”
Shao: DeepSeek on WeChat was 'game over' for ByteDance's Doubao
“Once Tencent integrated Deepseek into their WeChat product, it was like game over for them.”
Shao: US LLMs decisively led China until DeepSeek breakthrough
“In terms of frontier models, the U S companies, like you mentioned, the open eyes and the tropics of the world have definitely led China's LLM development by far for a long time until deep sea made a breakthrough.”
Shao: Once LLMs become good enough, AI competition shifts to products
“If we've reached a point where everything is kind of just good enough, then I think commercialization is also going to be in product layer.”
Kantrowitz: US has seen very little consumer AI deployment compared to China
“We've seen very little consumer AI in the U S very little. And so they're actually putting it into play, which is interesting.”
Kantrowitz: 20 Chinese automakers announce plans to integrate DeepSeek models
“20 Chinese automobile brands have announced plans to embed deep seek models.”
Kantrowitz: China's top 5 smartphone makers adopt DeepSeek in updates
“China's top five smartphone sellers, I think you mentioned this, have all adopted DeepSeek in their system updates.”
Kantrowitz: Nearly 100 Chinese hospitals announce plans to adopt DeepSeek
“Nearly, nearly 100 hospitals around China have announced they will adopt DeepSeq in their operations.”
Kantrowitz: Many GPUs are transshipped through Singapore to China
“Clearly there's a lot of GPUs that have gone through Singapore and then go right to China.”
Shao: Public tech equities are the best route to Chinese AI exposure
“Given certain restrictions, US private equity firms and VC firms still cannot invest in Chinese AI companies directly. The best way to get exposure to this innovation is through public investment into these Chinese big tech right now.”
Shao: Chinese manufacturing provinces are aggressively backing AI and robotics
“There's been a lot of like municipal level or provincial level policies throughout the country. That's really been boosting AI per se, and like I said, physical AI, especially in areas like Zhejiang province or Guangdong province, where they're strong on manuf…”
Shao: Unitree controls 70% of global humanoid robot sales
“Unitree Robotics currently account for about 70% of global sales of humanoid robots, 40% of market share of quadruped robots, and that's just like a solid leader in the industry, and I think their cheapest model goes for something like 16,000 dollars, so it's …”
Shao: Hangzhou has surpassed Shenzhen as China's primary startup hub
“10 years ago, if you ask where is China's startup hub, it would have been Shenzhen, which I mentioned was like the hub of the headquarter of Tencent, DJI, SenseTime. Now it's actually up there in Hangzhou.”
Shao: Manus AI agent is not a foundational DeepSeek moment
“I don't think it's a deep seek moment for China. Because I think it did not fundamentally change how we Believe AI is being developed, right? Like in the sense that Deepsea completely challenged the scaling law and what we believed in more capital is better tr…”
Shao: Chip shortages are the biggest bottleneck facing Chinese AI development
“It's definitely also hindering Chinese, like, you know, AI development. For sure, that's probably Biggest bottleneck because we talked about it. It has innovation. It has data. It has energy. It doesn't have chips. So unless domestically, you know, we see like…”