Chinese AI firm SenseTime predicts a major transformation in artificial intelligence commercialisation, moving away from a token-based pricing system toward a 'task economy' where customers pay for complex, completed tasks rather than units of compute or tokens.
- Token prices expected to fall like telecom data costs two decades ago
- Shift towards paying for AI task completion rather than raw computing
- SenseTime focuses on multimodal AI to create practical, production-ready applications
What happened
SenseTime, a leading Chinese artificial intelligence company, outlined its vision of a shift from the current token economy to a task economy in AI commercialisation. CEO Xu Li shared in an interview that as foundational AI models and computing power become commoditized infrastructure, the pricing for tokens—units used to measure AI resource consumption—is expected to drastically decrease over time.
In this emerging model, users will pay primarily for completed, complex AI-driven tasks, such as reviewing architectural blueprints or generating product videos, moving away from paying solely for raw compute power. SenseTime recently introduced its SenseNova U1 Pro multimodal AI model during the World AI Conference in Shanghai, highlighting its investment in technologies that process images, video, and 3D data alongside text.
Why it matters
The expected decline in token prices marks a significant transformation in how AI services are monetized and adopted. This evolution from a token economy to a task economy could lower barriers for AI use, expanding the total addressable market beyond specialized professionals like programmers to include a wide variety of users across industries and demographics.
China's large population and complex use cases give it a unique competitive edge in discovering diverse AI applications. SenseTime believes this dynamic environment can accelerate AI ecosystem development faster than in other markets, presenting an opportunity for rapid innovation in practical AI implementations.
What to watch next
Observers should track how token pricing evolves and how rapidly the task economy model gains traction within China’s AI sector and potentially beyond. SenseTime’s push into multimodal AI technology, which integrates visual and textual data for real-world problem solving, will be critical to achieving this vision and could set new industry standards.
Additionally, monitoring SenseTime’s profitability trajectory amid the ongoing agentic AI surge will provide insights into the commercial viability of the task economy. The company’s ability to maintain strong margins in generative AI applications and expand its customer base outside professional circles will be key indicators of success.