The AI Policy Podcast
Episode Title: Why is China’s AI Sector Booming?
Date: September 12, 2025
Host: Center for Strategic and International Studies (CSIS), Gregory C. Allen and Sadie McCullough
Overview
In this episode, Gregory C. Allen, Senior Adviser with the Wadhwani Center for AI and Advanced Technologies, and first-time co-host Sadie McCullough explore why China's AI sector is experiencing such rapid growth. The conversation analyzes the catalysts of China’s “AI boom,” the impact of US–China competition, China’s unique policy focus on AI application, and the wider implications for global AI geopolitics, hardware development, investment, and national security.
The "Deep Seek" Moment and Ongoing Optimism
(00:54 - 04:21)
-
The Deep Seek Event:
- In January 2025, China's Deep Seek (“Deep Sea”) R1 reasoning model gained global attention and stirred both media and political reactions. US and Chinese stock markets fluctuated following the event, with Nvidia’s value briefly plummeting.
- “That Deep Sea moment was like a crazy media circus and political firestorm... President Trump had to talk about it... The CEO of Deep Seek was prominently featured there. So that was a huge moment in the global AI world.” (Greg, 01:12)
-
Global Response:
- Though some expected this to signal American decline, the tech sector in the US rebounded, and American AI companies continue to grow.
- The Deep Seek model reaffirmed China’s world-class AI talent and capability, as recognized by leading Western AI labs, despite ongoing export controls.
-
Talent Concentration:
- “More than 47% of all the AI engineers on planet Earth are currently in China. So this is still a big market, there’s still a big opportunity.” (Greg, 02:54)
Focus Shift: AI Applications vs. Superintelligence
(04:21 - 09:15)
-
Western vs. Chinese AI Priorities:
- The US and Europe are captivated by “superintelligence”—the idea that AI could far surpass human intelligence.
- In China, directives from Xi Jinping and the Politburo push for “strongly oriented towards applications.”
- “Those government muscle movements that China knows how to do very well...That is our advantage in the AI race.” (Greg, 06:21)
-
Strategic Framing:
- Greg uses analogies to explain the distinction: Is the Industrial Revolution won by the best engine, or by the best adoption? China’s bet is the latter.
- “China is saying, just because we may not have the best AI in the world from an absolute peak performance perspective, doesn’t mean that we might not have the overall best economic benefit if we are better at driving adoption.” (Greg, 08:31)
Policy Focus: The “AI Plus” Directive
(09:15 - 14:50)
-
New Policy Document (“AI Plus Action”):
- Released by China’s State Council in August 2025, the “AI Plus” plan echoes the earlier “Internet Plus” revolution.
- The document sets ambitious integration and adoption targets across six key sectors: science and technology, industrial development, consumption upgrading, livelihood, governance, and global cooperation.
-
Pragmatism Over Hype:
- “Notably, the document doesn’t cite AGI as the driver...It’s focused on how the AI of today-ish can be leveraged to achieve the Communist Party’s economic, social and political goals.” (Citing Matt Shen, 10:19)
-
Adoption Pace in China:
- The plan targets 70% AI adoption by 2027 and 90% by 2030—numbers that, given China’s precedent with technologies like mobile payments, are ambitious but plausible.
- Memorable Moment: “The transition [from paper money to mobile payments] was like a handful of years... a jarring pace of technological change is very familiar in China.” (Greg, 12:46)
Economic Impact: Investment Boom, Market Optimism
(14:50 - 20:20)
-
Investment and Economic Size:
- Morgan Stanley projects the Chinese AI industry to grow from $3.2 billion (last year) to $140 billion by 2030; AI infrastructure could reach $1.4 trillion.
- Rapid increase in generative AI adoption: from 8% to 43% of firms using it in a single year.
-
Corporate and State Spending:
- “Chinese tech companies are preparing to spend jaw dropping amounts of money.” (Greg, 16:44)
- Top six cloud providers’ AI expenditure will reach $53 billion in 2025, up 60% from 2024.
-
Government Influence:
- Massive investments not just from private firms like Alibaba, Tencent, Baidu, but also from state-owned enterprises, e.g., telecoms.
Hardware Winners: Nvidia, Cambricon, Huawei, SMIC
(20:20 - 31:40)
-
Nvidia’s Role:
- Despite export controls, Nvidia’s sales in China have grown, especially with chips like H20.
- Yet, government-mandated avoidance of H20 in favor of potential future chips is causing buyers to hold off.
-
Market Shares and Domestic Players:
- Chinese GPU designers (notably Cambricon) have surged in market share, from 11% in 2021 to 34% in 2024.
- Cambricon’s Meteoric Rise:
- Stock price up 5500% in a year, revenue up 4000% in first half of 2025.
- Key advantage: improved compatibility with Nvidia’s CUDA ecosystem.
- “[Cambricon’s] stock is trading at an eye watering, trailing twelve month price to earnings ratio above 4,000 compared to 60 for Nvidia.” (Greg quoting SCMP, 29:10)
-
Huawei and Overcapacity:
- Huawei’s AI accelerators have high sales but low utilization due to inferior design for LLMs.
- “...people had bought a lot of Huawei chips...10,000 chip GPU clusters...where it’s actually really impossible to do anything interesting or useful...” (Greg, 27:23)
-
SMIC and Chinese Manufacturing:
- Local chipmaker SMIC (partnered with Cambricon, Huawei) is also a winner, along with equipment firms like Naura.
National Security & Export Controls: Effectiveness and Consequences
(31:40 - 41:19)
-
Export Controls' Impact:
- The controls have had significant but less-than-intended effects; they didn’t fully cripple China’s advanced AI development.
- “President Biden directed them...to, quote, go for the jugular... That’s what they thought they were doing with the October 7th export controls. The reality is they didn’t.” (Greg, 32:10)
- Allowances for products like H800 chips and manufacturing equipment blunted enforcement.
-
Domestic Substitution Is Accelerating:
- Government mandates now require at least 50% localization for chips in public computing hubs, with some cities (e.g., Beijing) aiming for 100% by 2027.
- “That [quota] basically means that all government AI data centers and potentially all state-owned enterprises are facing a huge buy local mandate...” (Greg, 36:50)
-
Data Localization and Autonomy:
- Increasing mandates for domestic training and data center operation (e.g., for autonomous vehicles), with localization often required by law.
Geopolitical and Industry Implications Post-H20 Ruling
(41:29 - 43:35)
-
Decision to Allow H20 Chip Sales:
- Has not substantially changed China’s push for self-reliance.
- China is determined: “I don’t think there’s any apology that we’re going to make or any deal that we’re going to cut that’s going to persuade them that they should not pursue self reliance as quickly as they possibly can.” (Greg, 41:41)
-
American Exporters’ Dilemma:
- If US companies halt chip sales, China will have a smaller but more autonomous AI sector; globally, US firms may dominate AI application services.
US Tech Execs’ Perspective & Future Competition
(43:35 - 45:55)
-
Sam Altman’s View:
- “I'm worried about China. There's inference capacity where China can probably build faster... I don't think it will be as simple as is the US or China ahead.” (Sam Altman, quoted by Greg, 43:45)
- China excels at infrastructure; inference bottleneck is still chip supply.
- Research: China leads in papers and patents, while breakthrough products still mainly come from the US.
- Adoption: US advantage in platforms (e.g., OpenAI, Microsoft), but not guaranteed to persist.
-
Greg’s Warning:
- “Victory is not assured here. We could absolutely lose. And that’s why it’s so important that we do what it takes to race ahead and be competitive like China. They’re not chumps.” (Greg, 45:37)
Notable Quotes and Moments
-
On Export Controls:
- “We did not have a maximalist export control strategy. We tried to show restraint...Once all the lobbyists descended and tried to water down the policy...” (Greg, 33:41)
-
On Chinese Policy Execution:
- “When they say they're going to drive... transformation across the economy... you can bet that every local government is paying attention and every business is paying attention.” (Greg, 13:29)
-
On US–China Competition:
- “Do we want to help build a bridge to the Chinese future where they can supply more and more of this capacity locally?” (Greg, 41:50)
Key Timestamps
- The Deep Seek moment: 00:54–04:21
- Application vs. Superintelligence approach: 04:21–09:15
- "AI Plus" policy discussion: 09:15–14:50
- Markets, investment, and adoption stats: 14:50–20:20
- Who wins in chips/hardware supply: 20:20–31:40
- National security and export controls: 31:40–41:19
- Geopolitical implications after H20 licensing: 41:19–43:35
- US tech execs’ views and final thoughts: 43:35–45:55
Conclusion
This episode presents an in-depth look at China’s AI sector boom and the nuanced dynamics of US–China competition. The key takeaway is that while export controls and US policy have impacted China’s trajectory, they have not halted its rise. Concerted policies, relentless application focus, vast talent, and massive investments are driving China's AI ambitions, forcing the US to race harder to maintain its edge. As Greg Allen puts it, “Victory is not assured here.”
For more insights and expert analysis, follow the CSIS Wadhwani Center’s future episodes and referenced sources such as Matt Sheehan’s and Patrick Zhang’s Substacks, and the cited industry reports.
