From Models to SuperApplications: a fresh phase of the AI race
Chinese technological giants are moving from building advanced language models to implementing consumer "superplications" based on AI. In November 2025, Alibaba Group officially launched the Qwen app – an intelligent assistant designed as a competitor of WeChat in the era of artificial intelligence.
Although inactive in the beta test phase, Qwen immediately finished 4th in the App store ranking in Hong Kong and 5th in mainland China. The application is not limited to conversations (dialogues) – it offers image generation, research, organization of individual and professional tasks, and above all – full integration with the Alibaby ecosystem (payments Alipay, maps, trade, cloud etc.).
“Using a deep knowing of the user, the app can rapidly turn conversations into business transactions”, said Poe Zhao, the founder of the newsletter Hello China Tech.
AI as an gross engine
Unlike Western companies that are inactive struggling with model coining, Chinese companies show that AI is already generating concrete operational and financial benefits:
- During this year's Singles’ Day, the "AI Business Advisor" tool Alibaby generated over 5 million sales reports.
- JD.com automated 95% of orders that were executed within 24 hours thanks to AI.
- Ant Group (fintech Alibaby) launched LingGuang – a tool that allows users to make simple applications (e.g. a calorie calculator) utilizing voice commands in 30 seconds.
"LingGuang gives each of his own AI-developers in his pocket – coding, creating graphics and transforming ideas into solutions," said He Zhengyu, technology manager of Ant Group.
Increase in gross and regulatory pressure
The financial results of the 3rd 4th of 2025 confirm the trend:
- Xiaomi first achieved profit in electrical car and AI divisions, with a 130% increase in net profit.
- Baidu, despite a fall in advertising revenues, recorded a 50% increase in the AI segment.
At the same time, the anti-trust office of China published a draft of guidelines to prevent algorithms from manipulating prices and creating artificial monopolies, indicating an increased surveillance of AI applications in trade.
A gap in natural computing power
Despite the leadership position in applications, China loses ground in the race for computational capacity. In November, Google DeepMind launched Gemini 3 Pro, which pierced OpenAI GPT-5.1 and pushed the Chinese Kimi K2 reasoning (Moonshot AI) into 4th place in Artificial Analysis’ Intelligence Index.
Limited access to advanced Nvidia chips (A100, H100, Blackwell) makes it hard for Chinese companies to train the largest models. In response, Chinese manufacture leaders met in Shanghai to discuss the strategy – and the key direction was RISC-V, an open processor architecture.
Experts agree that RISC-V processors, valued for low energy consumption and flexibility, will become the main carrier of AI calculations in China within 3-5 years.
Global implications
Although Western media initially ignored Chinese AI initiatives, there is now increasing concern.
If Alibaba and its partners can combine deep ecosystem integration, user scale and applicable applications, Qwen can establish a fresh standard of "agent AI" on a global scale – not only in Asia.
Source:
- South China Morning Post – China Future Tech: The ‘everything’ AI app, November 22, 2025
- Artificial Analysis – Global AI Model Rankings, November 2025, https://artificialanylysis.com
- Alibaba Group Investor Relations – Q3 2025 Earnings Release, https://www.alibabagroup.com
- RISC-V global – RISC-V in AI and Edge Computing: 2025 Outlook, https://riscv.org
Leszek B. Glass
Email: [email protected]
© www.chiny24.com
