On the eve of the 2026 Spring Festival, China’s biggest internet firms launched a cash-fuelled campaign to turn curiosity about artificial intelligence into daily habit. Tencent, Baidu and Alibaba together committed 45 billion yuan in holiday incentives—roughly $600–700m—by tying traditional New Year “red envelopes” to the use of their AI assistants, transforming a seasonal marketing ritual into an experiment in mass behavioural design.
The three-pack of offers differed in shape but not intent. Tencent’s Yuanbao app announced a 10 billion-yuan pool that includes individual envelopes of up to 10,000 yuan and social sharing bonuses; Baidu dedicated 5 billion yuan to weave cash rewards into interactive narrative features inside its Wenxin assistant; and Alibaba’s Qianwen platform unveiled a 30 billion-yuan “Spring Festival host” plan promising free orders and subsidies across its commerce and local-services ecosystem. ByteDance has not matched the large cash pools but secured prime exposure for its Doubao AI by serving as the AI cloud partner for the national gala broadcast.
The tactic is familiar but the target has shifted. Red envelopes have been used since 2015 to teach millions to bind payment apps to social platforms; this year the objective is to embed an AI “super‑entry” into daily life. China’s internet authorities report that by mid‑2025 there were about 515 million generative‑AI users—roughly 36.5% penetration—meaning the market sits at a tipping point where a shallow nudge can scale into habit formation.
Mechanically, the promotions force interaction. Users must open AI apps, hold conversations, generate content or engage with AI effects to unlock rewards. Tencent links extra draws to sharing on WeChat and QQ and to a nascent social feature called “Yuanbao Pai”; Baidu’s “fantastical life” scenario hands out cash as users choose branches of a story or solve riddles; Alibaba’s approach embeds discounts and full refunds into real transactions orchestrated by its assistant across food, travel and retail services.
The logic is straightforward: holiday periods provide dense, socially amplified usage over several contiguous days—an ideal window to establish 7–10 day routines that behavioural economists and some industry analysts say can stick. Citations of the 2015 WeChat red‑envelope moment are apt: that campaign taught billions to pay by phone and rewrote China’s payments landscape. The current playbook substitutes payments education for AI‑first interaction training.
Yet massive subsidy does not guarantee permanence. Early indicators have already surfaced: Yuanbao’s social links flooded group chats and produced user annoyance alongside rapid download numbers; industry observers warn many participants will take the money and uninstall. Experts such as Cui Lili of Shanghai University of Finance and Economics emphasise that migration to a new interface depends on the quality and fit of the AI experience, not just incentives. Pan Helin, a member of the Ministry of Industry and Information Technology advisory committee, notes the promotions are effective at quick market education and social reach, but that long‑run stickiness requires meaningful, repeated value.
The three companies’ tactical differences reflect deeper strategic bets. Tencent is leveraging social graph effects to make AI a layer atop its messaging empire; Baidu is selling immersive, narrative‑driven engagement to anchor users inside search and assistant interactions; Alibaba is converting AI into a commercial assistant tied to transactions across an already vast retail and services ecosystem. ByteDance’s route is exposure and product momentum, using mass‑broadcast visibility instead of direct cash subsidies.
The broader significance is twofold. First, the battle over who becomes the default AI entry point is a contest for future distribution power: whoever wins will route large swathes of requests, commerce and attention through their model and data flows. Second, the spending arms race raises the cost of entry and could entrench incumbents that can sustain subsidies and ecosystem lock‑in. For smaller players with no deep pockets, the danger is rapid marginalisation despite having comparable algorithmic capability.
Ultimately, this Spring Festival experiment will be judged not by downloads or headlines but by retention and habitual use in March and April. If users retain an AI for practical tasks—planning, booking, shopping or social play—the spending will have bought genuine mindshare. If not, the effort will read as another seasonal marketing spree with temporary fireworks but little structural change to the market.