Xiaomi has delivered a seismic shock to China’s artificial intelligence sector by announcing a permanent price reduction of up to 99% for its large language model (LLM) APIs. This aggressive move essentially commoditizes the company’s AI capabilities, allowing developers to access five to eight times the volume of data processing, or 'tokens,' for the same price. By slashing costs to near-zero, Xiaomi is signaling that the value of AI in China has shifted from the models themselves to the applications built upon them.
This pricing strategy follows a broader industry trend initiated by DeepSeek and quickly adopted by tech giants like ByteDance and Alibaba. In the Chinese market, LLMs are no longer being sold as premium, high-margin software but are instead being positioned as essential infrastructure. For Xiaomi, this price war is a strategic gambit to lock developers into its vast hardware ecosystem, which now spans smartphones, smart home devices, and the burgeoning electric vehicle market.
The financial timing of this announcement is particularly telling. Xiaomi recently reported a robust first-quarter net profit exceeding 6 billion RMB and announced a significant 20 billion HKD share buyback plan. This formidable war chest suggests that the company is well-prepared to sustain a prolonged period of low-margin or loss-leading AI services to bleed out smaller competitors and capture dominant market share.
While Western AI leaders like OpenAI and Anthropic continue to focus on increasing model parameters and performance, Chinese firms are prioritizing cost-efficiency and mass adoption. Xiaomi’s 99% discount effectively lowers the barrier to entry for small-scale developers and traditional industries to integrate AI. This shift could accelerate the deployment of 'AI-native' features across China’s manufacturing and consumer electronics sectors, further decoupling the Chinese AI trajectory from that of Silicon Valley.