$1.1 billion. That's how much China's four largest tech companies collectively spent during the Lunar New Year holiday to bribe people into using their AI chatbots.
Alibaba led the charge with an estimated $430-720 million in subsidies for its Qwen assistant, offering bubble tea for one cent through 25-yuan vouchers redeemable at 300,000 outlets. Ten million free orders were placed within nine hours. ByteDance threw $215 million at its Doubao chatbot, securing an exclusive partnership with CCTV's Spring Festival Gala (China's most-watched broadcast, with 700 million viewers) and giving away prizes including Audi and Mercedes EVs. Tencent put up $140 million for its Yuanbao assistant. Baidu, the distant fourth, allocated $70 million for its Ernie chatbot.
This is not a tech launch strategy. This is the ride-hailing wars of 2015 all over again, except the product being subsidized is an AI assistant.
The Numbers Look Impressive Until You Check Retention
During the holiday peak, the user spikes were dramatic. Alibaba's Qwen jumped from 17 million daily active users to 73.5 million. Tencent's Yuanbao went from 7 million to 40.5 million. ByteDance's Doubao climbed from 84 million to 144.5 million.
But what happened after the cash stopped flowing tells a different story. Alibaba saw a 51% decline in engagement time, with average sessions dropping to just 3 minutes. Tencent's Yuanbao reverted almost entirely to pre-campaign levels with a 32% drop in usage time. ByteDance's Doubao held up best, losing only 15% of daily usage time and maintaining an average of 8.7 minutes per session.
Morgan Stanley's assessment was blunt: "Short-term traffic explosion doesn't guarantee victory; product practicality, ecosystem integration, and genuine value beyond monetary incentives determine long-term success."
What This Means Outside China
The spending tells us something important about the state of AI competition: these companies believe the race for default AI assistant status is winner-take-most, and they're willing to lose billions to win it.
The parallel to the Western market is imperfect but real. OpenAI, Google, and Anthropic compete on model quality and developer adoption. Chinese companies are competing on raw user acquisition, betting that getting people to form a chatbot habit matters more than having the technically best model. ByteDance is marketing Doubao 2.0 as 90% cheaper than GPT-5.2, competing on price rather than benchmarks.
For users of Western AI tools, the immediate impact is indirect but worth watching. DeepSeek already triggered a global AI price war with dramatically lower API costs after its January 2025 launch. If China's subsidy-fueled growth produces AI assistants with hundreds of millions of sticky users, the pressure on OpenAI and Google to cut prices or increase free-tier offerings will only intensify.
The bigger question is whether any of this spending produces genuine product adoption or just coupon-clipping behavior. Alibaba's 55% of orders coming from rural users suggests they reached new demographics, but a 3-minute average session time suggests those users aren't exactly power users yet. As one analyst put it, "People will try it out, but they won't see the value there, and then they will revert."
China's regulators are already watching. After similar subsidy wars in food delivery, the State Administration for Market Regulation stepped in. If the AI spending continues at this pace, expect similar intervention.