As 2024 comes to an end, the large language model (LLM) industry in China remains one of the most crowded and heavily invested sectors, yet it faces increasing skepticism regarding its actual commercial value and sustainability. While funding continues to pour in, tangible outputs, real monetization, and game-changing products remain elusive.
This report examines the state of China’s LLM industry, compares it with the U.S. AI landscape, and analyzes the key challenges, market shifts, and the inevitable consolidation ahead.
1. From Frenzy to Reality Check: The Rise and Fall of China’s LLM Boom
At the height of China’s AI boom in early 2023, a wave of startups and major tech companies rushed into the LLM space, eager to become the “Chinese OpenAI.” This led to an explosion of AI model releases, with companies like Baidu, Alibaba, Tencent, ByteDance, iFlytek, SenseTime, and even automakers like Li Auto unveiling their own models.
By mid-2023, there were over 200 Chinese LLM projects, and investors were pouring billions into AI startups. Startups like Zhipu AI, Moonshot AI, MiniMax, and Baichuan—collectively known as the “Six Little Dragons”—were attracting funding rounds exceeding $1 billion RMB. However, by late 2023, cracks began to appear.
The Warning Signs
- User Retention Problems
A viral image circulated in September 2023, claiming that new user retention rates for China’s top LLMs (from Tencent, ByteDance, Baidu, iFlytek, and Moonshot AI) dropped below 1% after 30 days. While the data lacked an official source, its rapid spread highlighted a growing frustration with the industry’s lack of compelling user adoption. - Investment Fatigue & Overvaluation
Investment analysts began warning that China’s LLM startups were overvalued, with companies like Zhipu AI, Moonshot AI, and Baichuan AI reaching valuations of over $2.8 billion USD without demonstrating viable business models. FOMO-driven investments of 2023 turned into cautious skepticism by 2024. - Rising Costs, Weak Monetization
- Training and running large-scale AI models is incredibly expensive.
- Analysts estimated that China’s LLM companies had spent over $10 billion USD on model training by the end of 2024, but their total revenue across the industry was less than $2 billion USD.
- Unlike OpenAI’s ChatGPT Plus ($20/month) or Microsoft’s Copilot subscriptions, few Chinese models had effective monetization strategies.
- Slowing Technological Progress
While OpenAI, Google DeepMind, and Anthropic continued pushing the boundaries of LLMs (e.g., OpenAI’s upcoming Orion model), China’s AI startups faced a plateau in model performance improvements. - Hiring Freezes & Pay Cuts
- AI job postings in China dropped 30%-40% in 2024, compared to 2023.
- Salaries, which once doubled in 2023, stagnated, and AI firms introduced stricter hiring policies.
- A leading Chinese AI company even blacklisted candidates who had switched jobs more than three times in five years.
2. The AI Winter Begins: From Expansion to Survival Mode
The Chinese AI industry has entered a survival phase, where success now depends on real-world applications rather than mere hype.
(1) The End of “Bigger is Better”
Initially, China’s AI firms followed OpenAI’s brute-force approach, focusing on building increasingly large models. However, the economic reality forced them to shift priorities:
- GPT-3 (2020) → 175 billion parameters ($1.4M training cost)
- ChatGPT-3.5 (2022) → $4.6M training cost
- GPT-4 (2023) → 1.8 trillion parameters
- GPT-5 (2025, estimated) → Over $700M in chip costs alone
Unlike OpenAI, Chinese firms cannot afford to sustain such aggressive spending without clear ROI (return on investment).
(2) Shift to Application-First Strategy
Chinese AI leaders like Robin Li (Baidu CEO) and Kai-Fu Lee (AI expert) have emphasized that “applications, not models, determine success.” Without practical business applications, even the most advanced AI models are worthless.
Areas of focus now include:
- Enterprise AI (AI-powered SaaS, customer service bots)
- Government & Public Services
- Education (AI tutors, adaptive learning)
- Healthcare AI (drug discovery, diagnostics)
- Finance & Legal AI
(3) AI Market Consolidation: The Survival of the Fittest
With hundreds of AI startups competing but only a few monetizing successfully, massive industry consolidation is inevitable.
Predictions:
- Only 1-2 of the “Six Little Dragons” will survive long-term.
- Tech giants (ByteDance, Baidu, Alibaba, Tencent) will dominate.
- Most AI startups will be acquired by major firms.
- Smaller AI firms will pivot to niche applications or shut down.
Baidu CEO Robin Li emphasized:
“A foundation model without applications is worthless.”
Investors are no longer interested in funding new AI models; instead, they demand proof of revenue and real-world adoption.
3. Comparing the U.S. & China: AI Divergence in 2024
| Factor | U.S. AI Industry (2024) | China’s AI Industry (2024) |
|---|---|---|
| Leading Companies | OpenAI, Anthropic, Google DeepMind, Microsoft, Meta, Amazon | Baidu, Alibaba, Tencent, ByteDance, iFlytek, SenseTime |
| Funding | ~$100 billion invested in 2023-2024 | ~$40 billion invested in 2023-2024 |
| Monetization | OpenAI’s ChatGPT Plus ($20/month), Microsoft Copilot | No major consumer subscription model |
| Technology Edge | LLMs trained on more powerful GPUs(NVIDIA H100) | Limited GPU supply due to U.S. sanctions |
| AI Chips | NVIDIA, AMD, Google TPUs | Huawei Ascend chips, Baidu Kunlun chips(weaker than NVIDIA) |
| Commercialization | Strong enterprise adoption (Microsoft, AWS, Google Cloud) | Mostly experimental, weak enterprise adoption |
| Hiring Trends | AI talent remains in demand | Hiring slowdowns & pay cuts |
| Industry Future | Growth continues, but concerns over profitability | Survival phase, consolidation expected |
4. Conclusion: The Harsh Reality of China’s LLM Market in 2024
China’s AI industry has gone from hype to reality, facing a funding slowdown, monetization struggles, and technological stagnation. While the country remains committed to AI leadership, the coming years will be a test of survival for most LLM companies.
Key Takeaways:
- Overcrowded Market → Only a few will survive
- Hundreds of Chinese LLM startups, but only a handful will endure.
- The AI industry is shifting from “model hype” to real applications.
- High Costs, Weak Monetization → Business Models Need to Evolve
- AI model training costs exceed revenue by billions.
- Unlike OpenAI’s ChatGPT Plus, Chinese LLMs lack sustainable monetization.
- Tech Giants Will Win
- Alibaba, Tencent, Baidu, and ByteDance will dominate the AI market.
- Most startups will either be acquired or disappear.
Final Thought: The End of the AI Gold Rush?
The “AI gold rush” in China is over, and winter has begun. As hype fades and investments dry up, only practical applications and sustainable businesses will survive.
For China’s AI industry, 2025 will be a defining year—a moment of reckoning that will separate AI survivors from failures.