China’s largest internet firms staged an all-out marketing offensive around the Lunar New Year to seed audiences for a new generation of AI-native apps. Tencent’s Yuanbao, ByteDance’s Doubao and Alibaba’s Qianwen deployed massive promotion programs — giveaways, viral social mechanics and one‑click commerce — that pushed their monthly and daily active user figures into the hundreds of millions.
The campaigns produced headline numbers. ByteDance reported 19 billion Doubao interactions on New Year’s Eve and tens of millions of AI‑generated holiday avatars and greetings; Alibaba says users asked Qianwen for help 5 billion times during its event and used “one‑sentence to order” features almost 200 million times to buy eggs, milk tea and other consumer goods. DataEye and company disclosures put MAU for Doubao around 200 million, Qianwen at roughly 130 million and Yuanbao at about 114 million; DeepSeek, a separate “native” AI app, has posted MAU near 150 million.
Those user totals came at a steep price. Industry trackers estimate the three high‑profile products spent more than ¥100 billion on holiday marketing and user acquisition, a reminder that building scale for interactive AI today looks more like a marketing war than an organic, product-led migration. App‑store rankings reflected the flow of promotional traffic: Yuanbao briefly topped Apple’s free chart on February 1 before sliding, then re‑entered the top ten after the holiday as promotions tapered and usage patterns normalized.
The spending spree signals firms’ strategic priorities. Rapid MAU growth buys data, attention and a chance to embed AI into social and productivity workflows across vast ecosystems. Tencent is pushing Yuanbao into office and social scenarios — file parsing, document export and integration with WeChat and Tencent’s public account ecosystem — aiming to turn an AI assistant into a platform connector rather than a standalone novelty.
But the holiday spike exposed familiar growth puzzles: short‑term surges driven by giveaways are easy to buy, persistent engagement and monetisation are harder to earn. Industry analysts note that promotional intensity and ad injection are primary drivers of store rankings; when incentives end, a natural reversion to lower activity follows. Operators will be judged on retention curves, real‑world utility and the ability to convert novelty into daily habit.
Operational risks also surfaced. Yuanbao generated controversial outputs over the holiday and Tencent issued an apology, underscoring continuing content‑safety and moderation challenges as models are deployed at scale. Regulators and platforms will be watching for compliance with consumer protection and misinformation rules as AI assistants mix social, commercial and informational functions.
For the international observer, the episode offers an early view of how scale in consumer AI may be won: not primarily through model quality alone, but through ecosystem leverage, deep pockets and viral social mechanics. The immediate winners are firms that successfully turn festive traffic into sustained engagement; the longer‑term winners will be those that translate traffic into dependable product value, manageable costs and regulatory resilience.