Caixin reports that Chinese internet giants like Alibaba and ByteDance are shifting from pure model scaling to a multi-front competition over users, hardware, talent and capital in AI. The piece details soaring usage of apps like ByteDance’s Doubao, massive MaaS growth, and IPO plans from key startups such as Zhipu AI and MiniMax.
This article aggregates reporting from 1 news source. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
Caixin’s cover story makes clear that China’s race to AGI is no longer just about training bigger models; it’s about owning the “digital gateways” where those models touch consumers. Alibaba and ByteDance are embedding assistants like Qwen and Doubao into phones, wearables and super-apps, turning AI from a destination product into an always‑on layer across commerce, search and social. With Doubao reportedly at 172 million MAUs and leading China’s MaaS market via Volcano Engine, ByteDance, in particular, looks like the country’s closest analogue to OpenAI plus a hyperscale cloud wrapped into one consumer machine. ([caixinglobal.com](https://www.caixinglobal.com/2025-12-29/cover-story-chinas-ai-titans-escalate-battle-for-control-of-digital-gateways-102398121.html))
On the startup side, the “Six AI Tigers” are being forced into hard choices: several are retreating from pure foundation-model races into vertical applications, while others like Zhipu AI and MiniMax are heading toward Hong Kong IPOs despite multi‑billion‑yuan losses. ([caixinglobal.com](https://www.caixinglobal.com/2025-12-29/cover-story-chinas-ai-titans-escalate-battle-for-control-of-digital-gateways-102398121.html)) This environment favors players with integrated data, distribution and compute—precisely the strengths of Alibaba, ByteDance and Tencent. Meanwhile, DeepSeek’s emergence as the top global model by token volume underscores how Chinese open‑source models are driving down costs worldwide and pushing commoditization of base capabilities.
For the AGI race, that means a diverging model: U.S. labs push closed frontier models atop Nvidia-centric infrastructure, while China leans into heavy application integration, open models and talent scale. The competitive pressure between these systems is likely to accelerate both capability and deployment, even as monetization and regulation lag.


