Recently, the token economy has attracted considerable attention within the AI sector. Tokens, as the smallest units of information processed by large AI models, have seen an exponential increase in usage. This surge is propelling breakthroughs in end-device agent technology, paving the way for its large-scale deployment. Consequently, AI is evolving from being cloud-centric to being integrated into end devices, elevating its role from a mere auxiliary tool to an autonomous agent.

By March 2026, China's average daily token consumption had soared to over 140 trillion, marking a staggering thousandfold increase from the levels seen in early 2024. The National Data Administration has recognized tokens as the cornerstone of value in the intelligent era and as the settlement unit that bridges technological supply with commercial demand.

The explosive growth of the token economy and the advancement of end-device agents go hand in hand. Together, they establish value benchmarks for AI services, transforming agents from a theoretical concept into tangible productivity. To minimize token usage costs, end-device manufacturers are enhancing token conversion efficiency through on-device computing capabilities and system-level optimizations. This strategy is fostering the widespread adoption of AI in the consumer market.

With the widespread application of end-device agents across PCs, in-vehicle systems, smart homes, and other domains, a new AI ecosystem is rapidly emerging. This ecosystem is fueled by tokens and powered by on-device computing, heralding a new era of AI innovation and accessibility.