Summary in One Sentence:
The AI industry in both China and the US has reached a fascinating stage: the number of users on both sides is comparable—China’s native AI apps have nearly 500 million monthly active users, while ChatGPT alone has 1 billion weekly active users in the US. However, the profitability gap between the two is enormous. The fundamental reason isn’t technical superiority but the different ways AI has been defined as a product. In China, AI is integrated as an additional feature within various apps; in the US, it’s offered as a paid service. By 2026, the industry will face a critical turning point, where the competition will no longer be about which model is smarter but which can become a sustainable, profitable business without relying on massive funding.
---
Detailed Analysis:
1. The Core Difference: The Products Being Sold by AI Companies in China and the US
Don’t be fooled by the similar names; the actual products offered are vastly different. There are four main approaches to selling AI services globally:
- “selling capabilities”: This is like buying a video website subscription, where you pay more for higher-quality content and ad-free viewing, but you can’t guarantee the quality of each piece of content. Examples include ChatGPT and DouBao. You pay for access to a “more powerful model,” but you can’t verify exactly how much better it is, making it easy for users to compare prices. For instance, when DouBao’s 68-yuan subscription was criticized, it was because users felt the AI wasn’t significantly smarter for the money.
- “selling results”: This is like hiring a driver; you don’t care about the driver’s experience or skills, as long as they get you to your destination safely. Examples include Claude and AI programming tools. Users focus on the outcome—whether the code works and the reports are useful, which is why these products command higher prices. Claude’s average monthly user fee is over $211, eight times that of ChatGPT users.
- “selling transactions”: The AI itself doesn’t charge; it’s just a free tool before you place an order, with the cost embedded in the product you buy. A typical example is Alibaba’s Qianwen, which uses AI to help you select clothes on Taobao or book flights on FeiZhu. The AI’s cost is included in the transaction fees.
- “selling computing power”: The AI is free, and the profit comes from the servers that power it. For example, nearly 60% of Baidu’s AI revenue comes from renting GPU cloud servers, with only 20% from actual AI services.
2. The Dramatic Differences in User Acquisition:
The way users acquire AI services varies significantly between China and the US. In the US, there are three main types of users:
- **“internet-famous” users who actively use AI services like ChatGPT, accounting for 80% of traffic initially but now down to 50%.
- **“pre-installed” users who use AI with platforms like Android phones and Google services, reaching 1 billion monthly active users.
- **“parasitic” users who encounter AI within social platforms like Twitter X, with user numbers growing rapidly.
In China, all AI products are integrated into super apps. For example, DouBao is embedded in Douyin, and Qianwen appears in Taobao and Alipay recommendations. AI doesn’t aim to replace these apps but becomes a built-in feature.
3. The Unexpected Shift in Competition:
The focus of the AI competition has shifted from who has the best model to who can operate more efficiently and with fewer losses. No longer does it matter who has the largest model; the focus is on reducing costs and delivering tangible results.
4. The Biggest Challenge: Finding Paying Customers
Both China and the US have reached massive user bases, but profitability lags behind spending. The challenge is to find enough customers willing to pay for AI services. In the US, the “result-selling” approach is gaining traction, as users are willing to pay for practical solutions. In China, two approaches are being tried: large companies integrating AI into their products, and independent AI companies offering affordable services to developers.
5. The Turning Point:
The industry is no longer about model size or performance but about efficiency and cost-effectiveness. The core metric is task success rate—whether AI can complete tasks correctly without needing extensive adjustments. The competition has shifted to finding ways to minimize costs and maximize profits.
6. The Current Struggle:
Both China and the US are facing the same challenge: they have billions of users but are struggling to generate enough revenue to cover their expenses. The key is to find a profitable model that resonates with the market. In the US, the “result-selling” approach shows promise, while in China, integrating AI into existing apps is more common.
By 2026, it will become clear which approach is more sustainable.