The Full Picture of the "AI Society" China Envisions: Exploring the Background of China's Multi-Billion Dollar Investments

Prologue: New Trends in the "AI Cold War" Reflected by NYT Report

"Will China Surpass the U.S. in AI?"—The headline of the long-form report released yesterday by the New York Times instantly stirred the tech world globally. The article delves into the fact that Beijing established an 8.5 billion-dollar fund in April to intensively support over 400 "Little Giant" AI companies. While the U.S. adheres to a private sector-led approach, China aims for dominance through a typical "state capital × industrial policy" model.Techmeme

1. What is the 8.5 Billion-Dollar "Startup Acceleration Fund"?

The core of the fund is hardware support. Amid a severe GPU shortage, local governments and state-owned enterprises pool funds to create a system that "shares" computational resources and data center facilities with startups. This allows even early-stage companies to train models with 50B parameters. The NYT estimates that with "three researchers and a release fund of 2 million dollars," a practical-level dialogue model can be produced in six months.Techmeme

2. The Lineage of the Industrial Policy Approach

This is not the first time China has applied industrial policy to AI. In "Made in China 2025" and the 14th Five-Year Plan, AI is positioned as the "No.1 Frontier Area," and the State Council has advocated for "dual-wheel drive of safety and development." A recent RAND report analyzes that "while there is much waste, human and capital support is elevating private innovation."RAND

3. The Flexibility Indicated by the "Interim Measures for Generative AI"

The Interim Measures for Generative AI Services implemented in 2023 may appear to strengthen regulations, but the text explicitly mentions "classification and grading supervision" and "inclusive prudence." Administrative bodies set different review standards based on model size and usage, and while companies are required to submit transparency reports, rapid market entry is guaranteed. According to White & Case, this is a "dual-track of regulation × support" functioning as an investment signal.White & Case LLPWikipedia

4. The Open Source Wave: The Impact of DeepSeek

While American companies are strengthening closed routes citing "safety," Chinese companies, centered around DeepSeek, Alibaba, and Tencent, are aggressively advancing globally with open source LLM. An Asia Society webinar explained this as a strategy to "circumvent chip regulations and use lightweight, efficient models as a weapon."TS2 Space

5. The Chip War and the H20 "Reopening" Shock

On the night of July 14, NVIDIA CEO Jensen Huang announced that he had obtained a "special license" from President Trump to resume sales of H20 chips to China. VOX commented that "a huge hole has been made in the three-year ban on high-performance chips." Depending on computational efficiency, the H20 can surpass the flagship H100, and its price is half. China has effectively gained an "open loophole."Vox

6. "Data Villages" and Grassroots Innovation

The NYT article highlighted women working as data annotators in rural Shaanxi and Guizhou provinces. Although their average monthly income is one-third of urban areas, they are in high demand for advanced labeling tasks using smartphones and the cloud, allowing them to balance work and household duties. This is a form of Chinese-style social innovation that simultaneously advances the "common prosperity" policy and AI industry development.TS2 Space

7. Brain Drain or Circulation: The U.S.-China "Brain Battle"

The news that 7 out of 11 people hired by Meta's new "superintelligence" research institute were Chinese nationals sparked debates on domestic social media about "patriotism or career." Meanwhile, NVIDIA's CEO Huang received applause in Beijing for stating that "half of the world's AI researchers are Chinese." The Chinese government is strengthening return incentives, aiming for a brain circulation model.TS2 SpaceTechmeme

8. What Did Social Media Say?

• Techmeme (@Techmeme): "The 8.5B fund quickly narrows the 'U.S.-China gap.'"

• @EvanKirstel: "Does state-led innovation kill innovation?"

• @kyleichan: "The Chinese model is 'AI Neo-Mercantilism.'"

• Bluesky @patrickmccray: "Contrasting reduction in U.S. public investment"
  Analyzing the distribution of opinions, U.S. accounts tend to express concerns about "market distortion" and "safety risks," while Chinese accounts are more likely to praise the "strength of national will."X (formerly Twitter)Techmeme

9. Ethics, Safety, and Electricity: Three Bottlenecks

The explosion in computational power also amplifies ethical risks. The operation of the Chinese version of the "whitelist" avoids generating discriminatory, false, or violent content, but it is criticized for leading to enhanced censorship. Furthermore, training large models requires enormous electricity, with a mix of renewable and coal-fired power in desert data centers in Inner Mongolia and Gansu. This contrasts with the U.S. energy strategy of distributing subsidies under the CHIPS Act.Wikipedia

10. International Comparison: Options for the EU, India, and Japan

The EU is leading with "risk-based regulation" under the AI Act, demanding transparency as a card for market access. India is exploring a "public data + private implementation" strategy similar to the Chinese model. Japan aligns with the U.S. in high-performance chip export restrictions while maintaining Sino-Japanese joint research in a "dual-track diplomacy." The "international rental market for computational resources" could become a new battleground in the future.

11. 2030 Scenarios: Hegemony or Multipolarity?

• Scenario A: Intensified Decoupling—The U.S. strengthens export restrictions again, dividing the AI ecosystem into East and West.

• Scenario B: Rebuilding Interdependence—Foreign fabs support China's computational needs, and technical standards are shared.

• Scenario C: Multipolarity—India, the EU, and Gulf countries become a third pole, with AI platforms vying for dominance.

Experts agree that "any of these could occur depending on policy swings," but the common factor is that computational power and talent are the decisive factors.

Conclusion: A Small Step or a Decisive Turning Point?

8.5 billion dollars is smaller than a single large round in Silicon Valley. However, the method where the state acts as a command center, mobilizing regulations, funds, and data as one holds destructive power beyond mere figures. The NYT report indicates that this Chinese model has reached a "critical point" that could rewrite the global AI power map. Whether the hegemony will be held by the state or the market—the answer is still in calculation.